Is the AI Bubble Bursting? 🤖🫧 (Technology & Discussion) with James [988] ｜Luke's ENGLISH Podcast - Learn British English with Luke Thompson

Thank you for choosing Luke's English podcast.

For more episodes with specific vocabulary and pronunciation teaching, plus ad-free listening access to the Premium community.

And to show your support for the podcast, sign up to LEP Premium at teacherlukecouk slash premium.

You're listening to Luke's English Podcast.

How are you doing today?

I hope you're doing fine.

So welcome back to my podcast for learners of English all around the world.

The idea with these episodes is that they give you something that you can listen to regularly in order to practice English.

So let me tell you about what you can expect from this episode.

So this one is a conversation with my brother.

My brother James is back on the podcast.

This time we're talking about the subject of artificial intelligence again.

That's AI, artificial intelligence.

We're talking about this again.

And the central question that we will be discussing is whether the AI bubble is bursting.

Is the AI bubble bursting?

Now let me explain the kind of literal meaning of a bubble and then the more metaphorical meaning of the word bubble, as it's used in economics or business or technology is a round thing filled with air, right.

It's not a balloon.

A balloon is something made of rubber.

And if it's your birthday or someone's birthday, you... inflate the balloons, right?

That's balloons.

But a bubble is probably made from soapy water, water with soap in it.

You might blow bubbles from a bottle of soapy water in the garden on a nice sunny day.

The children love it.

You take the little plastic hoop, little plastic ring, it goes into the soapy water and then you blow air into it and bubbles come out right.

So you blow air into the bubble it, the bubble gets bigger as it fills with air and But then eventually the soapy skin of the bubble becomes too thin to contain all the air and it bursts pop.

So this is a bubble, literally a bubble.

But in economics or in business, a bubble refers to a situation where the price or value of something like houses or stocks in the stock market or maybe a new technology,

So the price or value of that thing rises very, very quickly.

It inflates and it rises.

The value rises far beyond its real basic value.

And this is usually because of hype or speculation.

So that's a situation in economics where the value or price of something gets bigger and bigger and bigger, to the point where it's unrealistic.

And during a bubble, confidence in that market, for example, if it's the housing market, so confidence in the value of those houses, the prices of those houses, confidence is very high.

Prices keep going up.

This attracts investors.

Even more investors, people continue to buy or invest in this thing.

And this creates a sort of self-reinforcing cycle, causing the bubble to grow bigger and bigger.

However, this growth can't continue.

And when people start to realize that the value of this thing has been exaggerated...

And then confidence in this thing drops, then pop, the bubble bursts and all of the value drops out of the market.

This means that the market collapses.

Prices fall rapidly.

Sometimes prices fall dramatically as people rush to sell whatever it is they've been buying or investing in.

This leads to people losing because they can't sell at the same price that they bought it at.

And basically, everything falls back down to normal, realistic prices or values.

So this brings us to artificial intelligence.

AI, I think probably became very well known for everyone about three or four years ago, especially with the introduction of these large language models like ChatGPT, which was launched by OpenAI.

Just a few years ago, OpenAI has invested billions of dollars into its technology.

It hasn't even started making a profit yet.

Since then, many strong claims have been made about AI.

These include both positive and extremely negative predictions.

And you could argue that all of this is a form of hype to increase.

How we understand AI and how we understand the value of AI.

Right.

So there's been a lot of hype into what AI can do, what it will do, how it will affect the future.

OK, so questions are the big predictions about AI really true?

I mean, you've probably heard them before.

Some of them are so big, they talk about a kind of end of the world sort of scenario.

That's how powerful people say that AI will be, that it's going to become more intelligent than humans, that it will somehow take over the world, that sort of thing.

Or at least it's going to remove something like 80% of jobs in the world.

You know we look at This technology the way it's been described.

It looks like it's going to utterly change everything about our world.

So are these predictions really true or are they just exaggerated?

Is this all hype?

And if it is hype, why?

So these are all the sorts of things that we're going to be talking about in this conversation.

A little bit of information about my brother James.

So his name is James and he's my brother.

There you go.

He's a long term regular guest on the podcast.

He's been on the show for many, many years.

So the focus of our conversation is the gap between what AI promises and what it actually delivers.

We're giving a critical perspective on current AI developments.

Now, some listeners listening to this, some of you might strongly believe in AI.

You might work in technology or you might just be someone who is really personally invested in AI.

You might really believe in it.

For you, this conversation might just sound like two guys complaining about artificial intelligence.

That's not really the idea.

That's not what we're trying to do.

What we're really trying to do is to assess the real value of AI today and what tech companies are trying to achieve with AI.

Now, if you're watching on YouTube, you will see that the actual conversation part of this doesn't have video.

It's just the audio version of the conversation between James and me.

But you will see subtitles on the screen saying, So you can read what you're hearing while you're hearing it, if you want to.

Audio listeners, you don't need to worry about the visual side of things.

You can just focus on your listening skills, just focus on the sound of English being spoken.

But a transcript is available for you.

There's a link in the description.

Try to notice, as you listen to this, try to notice the way that we argue our points.

I mean the specific ways that we word our arguments and the points that we're making, including both positive and negative things.

Try to notice how we provide examples of the things that we're talking about.

Try to notice any specific language for talking about AI and the industry and for talking about issues and problems relating to AI today.

As I said, a PDF transcript is available if you want to search for specific things that you've heard today or just to read it later to consolidate your learning, if you want to.

I'll speak to you again at the end of this conversation.

Obviously it's quite a long episode.

Don't feel that you have to listen to it all in one single go.

You could treat this like an audiobook, if you like, where you just listen to a bit stop, do something else and then come back to it at another time and continue listening.

But I'll speak to you again at the end.

I hope you enjoy the conversation.

One other thing, you'll notice that we don't get straight into the subject of AI at the beginning.

We had to, for some reason, talk about what my brother ate for breakfast on the particular day that we recorded this.

And you will discover the highly nutritious foods that James ate in order to provide him with the nutritional value and energy he needed today to be able to do this conversation effectively.

I'll let you discover the disgusting thing that my brother ate for breakfast on that particular day.

That's enough introduction for me.

Hope you enjoy the episode.

I'll speak to you again at the end, but for now let's get started and here we go.

How are you doing today?

I'm fine, thanks.

Just had a breakfast of a lovely white chocolate cream egg.

I can't believe you ate a Cadbury's cream egg for breakfast.

A white chocolate cream egg, which is even more disgusting than a normal one.

I have to explain what a cream egg is.

I'm sure they have them in other countries, but it's just a chocolate egg full of fondant, icing type stuff which is meant to replicate an egg white and yolk.

It's really quite disgusting for a number of reasons.

But my girlfriend got some from the supermarket on special because Easter's over now, so they're slashing the price of all the Easter eggs.

But I think Cadbury's aren't selling as well as they used to over here because they sold it to the Americans.

Did they?

What, Cadbury's owned by America?

Americans?

And they've changed the recipe.

Really, people don't like it, apparently.

A taste of bleach, a taste of uh fungus, allegedly some people bleach and fungus.

That is not a good combination of flavors at all.

It's not what you want from your chocolate eggs, is it?

I don't care, as long as it's massively sugary and bad for you, i'll eat.

So yeah, the chocolate egg, because we're recording this in April.

Easter weekend was this weekend we just finished.

And so it's been a lot of chocolate and chocolate eggs.

And of course, all of the eggs gets sold in supermarkets.

And then as soon as Easter's over, all the prices of the eggs get slashed.

So you can get cut price eggs, chocolate eggs.

And the cream egg is a sort of a British institution, although it is objectively disgusting.

Imagine a sort of egg, real size egg, like realistic size, but it's thick chocolate.

And then inside, James, as you said, to replicate the inside of an actual egg.

It's kind of like this white sugary fondant stuff which is just sort of like I don't know what it is.

It's sugar and all sorts of other nasty things.

I did hear that it's actually created using fungus to create it, but I don't know if that's true.

All sorts of things use fungus.

I mean, you know, corn.

That vegetable, veggie meat substitute, corn with a Q, that's made from some kind of mold, I think.

But this is the future, though, isn't it?

This is the future.

We're going to be eating insects and mold.

Yeah, I don't know about that.

Insect protein powder and mold.

This is what we're going to be living on.

I've actually tried some insect food at Oaxaca, the Mexican food chain here in the UK.

They offered us... Would you like some...

I think a cricket dip or something.

It was like a salsa made from crickets.

Oh, yeah.

How was it?

Just tasted of salsa, really.

I mean, you know.

Well, I suppose I do know what crickets taste like now.

It tastes like vinegar.

Well, especially when they've been covered in vinegar to make them palatable.

Anyway, we're going off on a massive tangent already, aren't we?

This is not the subject of this conversation.

We're not talking about eating insects and the future of food manufacturing and stuff like that.

But anyway, I'm glad that you had a really nutritious breakfast and that you're ready to do this episode.

We're going to talk about AI again.

I've done a few episodes about it, but we're going to take a slightly different angle on things in this episode.

I should say at the beginning that obviously James and I, we're not tech specialists.

You know, we don't, work in the tech industry.

We're not tech bros.

We're not tech bro.

We are bros, but, you know, literally low tech bros.

That's good.

Low tech bros.

Exactly.

So we're going to talk about AI and how things are going basically in the world of AI compared to how everyone was talking about it several years ago.

We're going to talk about how much we use it, what we think of it personally and stuff in a minute, and then we'll go on to the wider stuff about the situation with AI and the AI bubble and all that sort of thing.

I want to do a little introduction at the beginning now.

In November 2024, I did an episode which was called The Existential Threat of AI to Human Civilization, and it was quite terrifying.

In that one I talked about an interview that Jeffrey Hinton had done with the BBC.

Jeffrey Hinton is the guy that has been on lots of different podcasts talking about these terrifying predictions of the profound changes that AI is going to bring about in human society.

Isn't he supposed to be the godfather of AI, or something?

The godfather of AI, that's right.

Where did that come from?

He was one of the early developers of it, or something.

British, Canadian computer scientist, cognitive scientist, cognitive psychologist and Nobel Prize laureate known for his work on artificial neural networks, which earned him the title the godfather of AI.

I think he used to work for Google, quit his job at Google and then for some reason decided to just kind of become AI.

This interview guy, he kind of went on all the podcasts.

He was on Diary of a CEO and interviewed on BBC Newsnight and stuff with all of these terrifying predictions about the future of AI, talking about things like massive job losses.

AI-controlled weapons, wiping out humanity and various other things.

But were his comments all part of a kind of scam that we've been exposed to over the last couple of years, and are these outlandish predictions really true, or are they just exaggerated?

And why?

All of us have been living with ai in our lives for a few years now, most obviously in the form of generative llms, large language models like chat gpt, which was first opened to the public about three and a half years ago.

I think probably it was the arrival of ChatGPT which kind of brought all of this to the forefront and made us all talk about it.

I think probably all of us use AI to an extent.

There are various forms of it, but I think it's probably the chatbots that we're most familiar with.

You can't actually avoid it.

You can't avoid AI?

I mean, it's built into Google.

It's built into all the flipping phones.

You know, you type something into your phone, it gives you an AI synopsis before you get your search results.

It's in your WhatsApp.

It's bloody everywhere, isn't it?

Yeah, that's right.

WhatsApp.

Whether you want it or not, you can't turn it off.

Gemini, Grok, Claude, ChatGPT, which is a terrible name, isn't it?

ChatGPT.

I still can't remember what GPT stands for.

General Piss Taking.

Generative... I don't know.

What is GPT?

People can't say it.

How many times have you heard people mispronouncing it and calling it chat BGT or GBT or something?

Generative pre-trained transformer.

That's what it means.

What a catchy title.

Yeah.

Weird.

Because it transforms other people's creative work into money.

Yeah.

So anyway, we all know them.

Even if we don't use them intentionally, AI is being used all around us for various things.

AI in the tech world is massive.

It reaches very, very far.

There's huge astronomical amounts of money being thrown around.

In preparation for this episode, I listened to lots of interviews with tech experts talking about companies investing in AI subscriptions or investing in building their own AI systems.

And it all costs a phenomenal amount of money.

There's been a lot of hype around it, a lot of people in the industry with very big claims about it.

A lot of scary predictions too.

But at the moment the trend seems to be that people are not seeing a huge return on their investment.

So how much of what we think about AI is just hype?

Why would AI be hyped anyway and how?

And what is our actual experience of using AI?

We're a bit late to the game here with this subject about the AI bubble bursting.

We're a bit late to the game because people have been talking about this for months.

But this is a learning English podcast.

We're not tech bros.

We are low tech bros.

So there you go.

Listeners, I hope that you're able to follow this whole conversation and that you're able to pick out bits of vocabulary to talk about AI, talk about using it and the social issues relating to it.

We are very interested in your comments.

I'm sure I've got people listening to this in various countries around the world who work in the tech industry and who probably have insights that perhaps we don't have.

So please feel free to get in the comment section and write your comments.

So first some general questions.

James, do you use ai um?

I have done um really just to try out.

I can't say i use it very much in my actual productive work life or music.

I mean, when it first came out I did spend a bit of time on the image generation stuff and I found it pretty mind-blowing and it's quite scary.

I find quite eerie.

You know, people say it's, but it is eerie.

You know, you don't really know how these images are formed.

You know, they're not drawn in a normal way, like a graphics package.

And then they're just sort of imagined.

And then they just sort of well, they're composite images from millions of other images, that kind of a sort of somehow fuse together.

I don't quite know how it works.

So you know, I don't think anyone can fully explain to me how it really works but um, I find it very otherworldly and strange.

And you can come up with all sorts of things like give me an image of a sort of you know Viking ship on an English beach, and it'll do that.

And it looks really atmospheric and weird.

And there's always things slightly wrong with them, you know, as well.

So I spent a bit of time messing around with that kind of thing.

Finally, this is really freaky, you know.

Yeah, it is, isn't it?

When it comes to actually useful stuff, I've done a few.

I mean, I've noticed, as soon as AI came out, all the music album covers on Bandcamp.

Suddenly there was thousands of AI-generated images.

And you can kind of tell.

However good they are, you can still tell, can't you, that it's AI-generated.

It's just got this weird look to it that you can't quite put your finger on.

You just, well, that's AI.

Yeah.

And however impressive it is, it's still got that look to it.

And it just sort of feels like cheating somehow.

So I made a point of not doing mine in AI, except for one example.

I wanted a patch of grass for something to be sitting on.

I couldn't find quite the right image on stock photo libraries.

I wanted a circle stone within a patch of grass.

That was one instance where I actually was useful because I just wanted some grass.

A very specific task.

And it wasn't the main image.

It was just a bit of background stuff.

So that's one time I've used it that it was useful.

I should say that you, in terms of what you do, you're a graphic designer and also a music maker.

Yeah, I do my design for my album cover, which is sort of my hobby or side hustle, you could call it, um.

I do take a bit of effort over doing the covers and the artwork for it.

So that was one example.

But uh, and the other thing i use it for is some ai mastering.

There's a website i found.

What does that mean?

It's mastering is kind of the final stage of finishing a track music, music track, where you take the final mix down, track.

So it's not all the separate elements, it's everything you know, finished almost, and you just tweak the levels, tweak the eq, add compression, add stereo widening, add this that, the other, a bit of sprinkle, a bit of fairy dust on the on the thing, and you normally have to pay someone about 30 quid a track to do that uh, which i can't really afford if i'm going to put an album on my own 10 tracks.

You know that's 300 quid, which is probably more than i'm going to make on the release anyway.

So i did find that was quite useful for making the track sound a bit louder, a bit better quality.

But again, the AI, it's so limited.

It's not really listening to the tracks.

It's basing it on another track that you upload and you say can you make it sound a bit more like this production-wise?

When it works, it can sound really good.

When it doesn't, it just sounds awful.

And it can enhance all the wrong frequencies and it come back with loads of hiss on it, or it's dropped some of the bass drum out or something.

So it's very hit and miss.

And if I had to pay for it, because the one I'm using at the moment is free, don't know how it must be subsidized.

If I had to pay for it, I don't think I would because it's not good enough to pay for, to be honest.

So that's my experience of it.

And also one quick one, chat GPT.

I've used it a few times and I found it to be very unreliable to do the things I wanted to do.

I asked it to program a baseline for me on just an experiment.

I didn't expect it to be able to do it.

I thought it probably won't be able to do this, but of course it always says it can.

They say, can you, are you aware of the sequencer on the, you know, the TV three?

Yes, I am.

Is this a bit of hardware, audio hardware?

Yes, I am.

I'm certainly Ofei with the sequencer on the TV3.

Can you transcribe me the bass line to Sheik's Good Times on the TB3?

That's a classic disco track.

Yeah, it's the most famous bass line of all time, really.

One of the best bass lines of all time.

So you asked it to transcribe that in notation in notation.

Didn't expect it to be able to do it, but hey, it can.

It says it can.

Yeah, it's our result.

And it starts with rest c and you hang on.

Whoa, stop right there.

Good times, my chic starts on the one.

There's no rest of the star.

It starts with boom yeah, and it starts on the one yeah.

So this is obviously wrong.

Yeah, that was, let me try that again.

Wrong again.

Wrong again, just randomly making stuff up.

Yeah, so you go.

Well okay, you can't.

Why don't you just say you can't?

Because i'm terribly sorry, you know, i'll try my best to play.

Oh, shut up, it will never, ever say that it can't do something, yeah.

And then i tried to get it to make a video for me because someone said you know, you can type stuff into chat gp, you know, and make you a video.

I was like oh, that's interesting.

I'll try and make this video.

I wanted to make an arcade style screen that would have various messages on it and logos and stuff.

And it did the 2D version quite well.

You know, I gave it some references from arcade games.

Can you give a high score table here?

Blah, blah, blah.

I spat out a 2D flat image and it looked quite good.

And I was like, okay, can you make that animate now?

He goes, yes, I can.

And then it gives me some absolute garbage that looked nothing like the original idea.

And I was like, well, why are you doing that?

And he goes, you're quite right.

I haven't got the processing power right now, but give me another go and I'll do it.

And it keeps on promising next time it'll get it right.

Next time...

And he gives you a whole list.

Next time it'll be exactly what you asked for.

And you press it.

And sorry, I haven't got the processing power for that right now, but here's this.

It's like, no, just do the fricking thing I asked you to.

And if you can't say you can't, and you never did it.

And I spent three days on it.

Cause you know, you get three goes a day or something, or you don't pay for it.

You get a certain amount and it cuts off.

So I spent three or four days of my free credits trying to get it to do something, and I just gave up.

I came up with a different idea, made it myself, and it was all done in a day by me doing it.

I did say that it will never say it can't do something.

There are times when it does say that, but that's when it's...

When you've asked it to do something that is a violation of the rules like, for example, making a picture of a child or something like that you know, or asking it to make a picture of a famous person it says I can't do that.

I'm not allowed to do that, but I can give an approximation.

Grok will do it, right.

Evil AI will do all sorts of shit that it shouldn't do.

Yeah, they've got into trouble for that, haven't they?

Well, they have, but they also programmed it to be a right-wing propaganda tool as well.

It's been programmed to have a certain political viewpoint and stuff like that, which is kind of disturbing.

Yes.

If I was the richest man in the world, I would just put my freaking feet up.

And I would just chill out, you know, and have a good time.

Maybe dabble in a few fun kind of businesses here and there.

But I would not worry about anything at all.

What's the problem?

Yeah.

I know.

You've got everything you want.

You're the richest guy in the world.

Why are you making computers that are trying to make the world a worse place?

And. you know, funding far right in the UK and all this crazy stuff he does.

That's a whole other conversation, isn't it?

That's a whole other conversation for another time, maybe.

But anyway... Back to the subject.

I think everyone has had a similar experience with something like ChatGPT, where you ask it to do something and it's like incredibly enthusiastic.

But there are obviously limits to what it can do.

Having said that, there are some things that it's amazing at.

I mean for me as an English teacher.

I have to say that it's an incredible assistant for me because

As a generative large language model.

What it does is it produces language, and it does that really really well.

It's not great at writing a story which is going to move you or a story which is going to make you laugh.

It's not very good at that.

But what I often need is, for example, I've been working in a class – on a certain bit of English, let's say some phrasal verbs, and what I need is another activity which is going to take it further.

The course book material is not quite enough.

So I say, here are 12 phrasal verbs.

Can you just make an article about sharks and include all these phrasal verbs and bang it does it.

And then I say, now remove all the phrasal verbs and turn them into gaps, and it does that.

And then I say now make a quiz about all the phrasal verbs and now make some discussion questions to let my learners use all the phrasal verbs.

And it does it all really quickly.

Now previously, what I would do is I would end up sitting at my computer until the early hours right uh, doing all that myself, and it would take me ages, but ai does it super quickly.

So it's a a great kind of um teaching tool and undeniably super duper useful incredible, saves me loads of time.

How much would you pay for that if it wasn't free at the moment?

Oh, i don't know, i'd pay.

Would you pay 20 quid a month?

Would you pay 50 quid a month?

I'd pay 20 quid a month for it.

Would you pay 50 quid a month?

Potentially.

Yeah, because there's some discussion about at the moment, everything's subsidized that we use in AI.

They're using all this processing power and not charging you for it, basically.

So they're racking up debt in order to make people use it more.

But eventually, you're going to have to start paying for all this data processing.

So that's a model where they give you something for free for a certain period of time and then they make you pay for it, which is just like standard across the board for so many things.

The kind of free trial, three months free, and then you start paying.

So yes, and also they're probably making it available free, so that everyone just becomes dependent on it, and then you have to start paying for it, and then that's that.

I mean, at the moment they kind of give you a limited version for free and then they give you an expanded version which you pay for and other things like that.

So, okay, you actually...

Use it, but like most people, I expect you've noticed its limitations.

Big time.

And even when I do use it for mastering, I'm kind of on the fence about whether it's that good or not, because I did recently get some masters back for releases coming out and they got sent to a proper professional human.

And the difference was huge.

I mean, there's so much more refined, the ones that a human did.

Yeah.

Because they're actually listening to it.

I mean, the AI, I don't think it's actually listening.

It's just applying an algorithm or process to something.

And they were just head and shoulders above the AI stuff.

Yeah.

So yeah, even that's the one thing I use it for, even that I'm not sure I'd pay for it if it wasn't free.

Talking of other limitations, things like being original, being profound, having a human connection, having an overall vision.

So, for example, I've asked it to, just as an experiment, I've asked it to write jokes or write stand-up routines, and they're always terrible.

Yeah, it has a very basic level of humor, doesn't it?

There's a few tropes that it always trots out, I've noticed.

Alliteration is one.

If you ask it to come up with ideas, it'll always say Fantastic Follies.

You know, a journey into the idea of a stand-up comedian.

And it's always, you know, the same initial letter.

Yeah, it does alliteration, which is where the words all start with the same sound.

And it does groups of three words. you know it'll it'll create these sets of three words it does a lot of this isn't just a shampoo it's a movement yeah well this isn't just uh this isn't just a podcast it's a revolution you know or shit like that you know it does a lot of that very uh hackneyed So hackneyed listeners just means something that's been done again and again and again to the point where it's no longer original or interesting.

And so, yeah, it's extremely hackneyed in its style.

And writing jokes, it's just unable to do it.

It can't do that thing where it creates a double meaning thing.

It doesn't do that.

A lot of people think jokes are just clever double meanings a lot of the time and your brain, your brain is kind of flitting from one to the other going hey, they both work, they both work, and it's kind of like duality thing.

Two things happening at the same time causes your brain to kind of giggle a little bit.

So if you say crime, crime on multi-story comedies, many levels, i mean that's funny because there's so many levels to a car park and so many levels of complexity right, and they both the sentences, both work perfectly.

So you can say crime is wrong on so many levels, crime and multi-story car parks is wrong on so many levels.

It just works so well.

Yeah, Exactly.

Yeah.

But what chat GPT will do when it writes a joke is it just includes funny elements.

Right.

So it'll just add in like a monkey and a banana, throwing a banana at the zookeeper and the banana hits him on the head.

It's just like funny elements all jammed in together, but there's no actual sophisticated double meaning, you know so things like that.

And also I often want stories for my podcast.

You know, I want stories that are either scary, funny or moving.

Right.

And ideally a combination of the three is the best, but certainly something that's a bit moving in some way that grabs you and makes that emotional connection.

And the best stories will do that, and so i ask it to write a story.

You know, i give it all the elements i want.

It starts and the first half is great.

It's like oh, this is very promising and i think oh, this could be a good one.

Finally, a story generated by by chat chat, gpt that i can use and it all seems great, like the.

The premise and the language is flawless, but it does not deliver.

They never, ever deliver a satisfying sort of turn of events or a twist or some sort of narrative arc which does take you on a little journey.

That it never does, that it's always.

It always just ends up being kind of nothing.

It's like just sort of you realize it's tasteless, it's like drinking low, it's like drinking zero percent beer.

You're like this yeah, it tastes like beer, but actually no, there's there's just nothing to this at all.

I would ask you about AI and music, but I realize that we're kind of maybe stuck on this first part a little bit.

AI and music, I haven't really got involved in it.

It terrifies me because it's... it's just so convincing.

I mean, it just sounds like someone's made it, you know, I haven't really got involved.

I haven't tried making it myself, but I've heard some examples.

Like one of our friends Neil said, make a song about going camping, you know, to encourage his friends to go camping.

And it just came up with this sort of rousing song with all the bits they put in like make a line about we can go to the pub, make a line about you know?

Whatever it was.

It included all those elements into a song that sounded completely real to me.

So I find that very weird and very scary, but...

The limited stuff I've listened to, it is very generic.

And this guy, one guy I saw was trying to make post-punk, like kind of Joy Division type stuff.

But everything sounded a bit like U2.

And it was all very uplifting.

It didn't really have any darkness to it, you know.

If AI does have darkness, accidental it's darkness is just natural to it, no human element, there's no humanity to it.

When it generates an image or when it generates a video, there is a creepy sort of um, uh uncanny quality to it word uncanny, deeply disturbing, at a sort of uh, profound level.

It's kind of alchemy, isn't it?

There's like an evil black magic to it.

Yeah, there is.

Or it seems to be to us, because we don't understand how it's created.

Yeah.

AI videos that I've seen, for me... strike me as being dreamlike.

There's something nightmarish about watching AI video.

You kind of feel like, oh, this does feel like a nightmare.

It's got that kind of fluid reality to it.

I think because it's not real frames.

It's generating as it goes.

So we're used to seeing basically a series of stills that are joined together.

It's not that.

It's something else.

It's kind of a more liquidy sort of feeling, isn't it?

Yeah.

It's like a hallucination or a nightmare or something like that.

Because we're getting stuck on the first bit quite a lot, aren't we?

Unsettling.

Okay.

I mean, so the point there in that section is that, okay, we all use AI.

We've all seen its limitations, but also we have seen how incredible it is as well.

So as well as it being limited in some ways and maybe disturbing and frustrating…

There's no denying that it is also incredible.

And, you know, the way I use it at work, for example, it has really, really helped me.

Very specific things it's very good at.

It's also very good at illustration in a kind of quite generic style.

But say you want a lot of illustrations to illustrate a PowerPoint or a cartoon comic.

He does that kind of stuff very, very well, which you would have had to pay a guy, some poor guy that's now out of a job to do.

But it's not hugely original.

It's never going to come up with its own style, but it can do a kind of generic style very, very well.

Anyway, let's go back to what we mentioned before about people like Jeffrey Hinton and the things that they've said about AI, the AI future that's been promised.

What kind of future have we been promised in terms of AI?

I think it was about three years ago.

I saw all these guys pop up on all these podcasts talking about the future of AI and very smugly saying it's going to put everyone out of a job.

Saying in five years the job market is going to be unrecognizable, and start training to be a plumber and all this kind of horrible stuff.

There have been layoffs, undoubtedly.

It hasn't quite panned out the way people were saying.

What it was promised was going to be big impacts on jobs.

That was the main one.

It was going to be integrated into everything.

It was basically going to take over from humans, and instead of the kind of future we were all promised of the Jetsons, where the computers and the technology do all the boring stuff and we'd have fun, it seems like the computers are going to do all the fun stuff and we have to do all the shit.

Like what?

Well, they'll do all the creation and the art and the design and the films and you know creative stuff, and we'll end up doing manual labor.

Yeah, right.

That was kind of the future.

And also there's also this terrifying thought of AI in charge of nuclear weapons or even non-nuclear weapons, which is already happening, I'm sure.

Oh, yeah.

Undoubtedly.

I mean, I have examples of that.

I mean…

We could go into all of that.

It's kind of the same kind of technology as using medicine, which can identify cancers or something.

So we're told.

I think it's actually being used alongside humans to detect things like that.

But it's also used to detect people to kill them.

Yeah, and it sometimes makes mistakes with that, because the data it's using to base its information on, for example, here is a – this is the headquarters of the enemy or one of their major headquarters, but that data is five years old and now it's being used – you know, to house civilians.

For some, you know it's now it's a hospital or now it's a school or something, and that we've actually seen clear examples of that happening recently.

But maybe it's a good excuse as well.

I mean, you know, it's always the ai, right?

Yeah, takes away the responsibility.

Oh yeah, it was just an AI mistake.

We're terribly sorry, we didn't do it right.

Yeah, we didn't intentionally blow up a school.

Yeah anyway, but in terms of the future that was being promised, there have been positive predictions and negative predictions.

Yeah, the It was going to fix all, solve all medical problems.

It was going to find a cure for cancer.

It was going to do all these wonderful things.

It's going to replace your GP.

That was another thing I remember.

Your GP is your doctor, by the way, everyone.

Your general practitioner.

Your family doctor.

Sorry, go on.

Can you think of any more predictions?

I can't remember what they all were now.

Productivity.

Basically, the productivity is going to go through the roof.

That we'll be able to do X amount of work compared to what we used to do.

That we're all going to be able to do way more work.

I have to say that I have found that a bit like I said before, with my teaching.

I've discovered that it has allowed me to be a lot more productive in terms of writing lesson materials, which it does the legwork.

That's the point, that it it will handle large amounts of text and, And turn to, you know, gapping text, turning things into sentences, turning those sentences into questions, turning those questions into answers.

You know all of that kind of basic boring time consuming legwork.

It will do for me in a matter of minutes.

And I'm there curating it all.

But in terms of productivity, it's definitely helped me.

But that's one of the claims is that it would make businesses way more productive.

That you know.

We're talking about 30 50 percent more productivity in terms of the global economy.

Miracles, as you said, in different sectors like, for example, health care, being able to solve all these complex health problems.

In terms of coding, that it's going to democratize coding, allowing everyone to do a form of coding in which you ask the AI in plain English.

And then it will generate the code.

So you'll say, make a website with a banner at the top and this picture here and this text here, and blah blah, blah.

And it'll just do it for you.

So that kind of opens up the world of coding to people who don't have a lot of training in it.

I have seen a bit of that.

A friend of mine showed me an app that he'd made in AI.

You just told it what to do as a text command.

It tells you the chords that are used in a certain track and the keys and stuff like that and the BPM.

He said, is there anything you want an app doing, and I'll make it for you.

So I wanted to do a visualizer app, which is where you put in some music and it turns the sound wave into a kind of landscape that you're flying over.

Mm-hmm.

And it did it straight away.

And then I said, can we refine it a bit?

Can we have a bit more detail, make it look a bit like this have stars in the background and maybe have a spaceship flying over the top.

And he said, yeah, I'm working on it.

And then he said, oh, it's not exporting properly.

I'll get back to you.

And then he never did.

So for all I know, it's one of these things that promises everything and didn't deliver again.

But I don't know.

Maybe it will.

I've seen videos of.

I saw a video of a guy who basically got ChatGPT to write a version of GTA, the computer game top-down old-school version.

But it was remarkably good.

You could go around, you could steal cars, you could drive around the landscape.

But that's vibe coding, right?

Also, hyper-personalization of content.

Have you heard about this?

Even more than we get it already.

Hyper-personalization, yeah.

So you imagine...

It being incorporated into Netflix or Disney Plus or whatever, instead of just watching the content that they propose to you, that you'll have the ability to kind of curate your own content essentially, kind of put yourself into a Star Wars show.

If, for example, Disney Plus has all these AI capabilities, you'll be able to go in and set the parameters and say okay, I want a kind of action-packed, funny Star Wars adventure where there are Jedis and Sith and whatever, and I want you to include me.

You give it your image.

You give it photos of yourself and you give it your voice. and other data about yourself.

And it essentially creates a movie in which you are the main character in this Star Wars adventure.

Interesting idea.

I can see that taking a lot of generating power.

Yeah.

I mean, that's something we need to go on to later.

How is all this stuff done at the back end, the nuts and bolts of it?

And it relies on these huge data centers which are very expensive, very hard to build, very energy hungry and very water hungry or thirsty.

Because they need to cool down.

And the evidence looks like they're not actually being built at the rate they need to be built at, because it's not easy to build these huge power generators basically.

And then you need to shift the, the power from one place to the other through some sort of network, and it's not just, it's not easy yeah, so that's that could be a big sticking point for ai.

Where are you going to build these things?

How, how much is it going to cost?

And have you actually factored that into all your predictions?

You know, if everyone's going to use it all the time, every day, Where's the power going to come from and who's going to build this infrastructure?

Yeah, all of it, all of that stuff, building all of that requires an incredible investment, right?

But before this is like, you know, before the actual return has come from it, that building this thing, which then is going to generate all this money. requires trillions of dollars you know even for the basic stuff yeah and as it is they haven't really fully worked out to monetize it yeah they've got this amazing product or they think they have this amazing thing potentially potentially well it is amazing in some ways but i mean i heard a good quote it's amazing at getting something almost that you want It's amazing getting something close to what you're after.

You know it's.

It's amazing, creating something pretty much the thing that you wanted, but not exactly the thing we wanted.

So that requires a lot of hit and miss and a lot of retraining, and it's just never that efficient.

And all that's going to take power and money and resources that we maybe don't have.

There's also the argument that perhaps the system itself is still not quite ready to go to this extra step yet, that in its current condition

The thinking is well, we just make them bigger and feed them more data, and they will kind of just naturally become better at doing what they're doing.

They've already fed in as much data as they can get.

I mean, and they did a lot of that illegally, it seems.

I mean especially some of the, I mean allegedly some of the Chinese models just used all the Hollywood films in history without asking permission.

And the Western ones also have, it's been, you know, they call it scraping, don't they?

They've basically scraped the internet for everything without seeking permission from anyone or without copyright controls, you know?

Yeah.

I mean, there've been a few famous examples of this, like the Studio Ghibli case.

Yeah.

Where people realized that you could just go to chat GPT and say, make me a kind of picture of myself, but in the style of Studio Ghibli, that Japanese animation studio, and it would do it.

And then I think there were legal issues as a result of that understandably, because you know you can't just do that, can you?

Also, in terms of other positive promises, there's this whole question of superintelligence or artificial general intelligence, which is the idea that At some point, AI will reach a level of intelligence that is equal to and superior to human intelligence.

And then when we reach this point, it'll be like some sort of singularity.

The way it's talked about is almost like a godlike event.

Like AI is going to reach God level, God mode.

It'll surpass human intelligence and programming itself and stuff like this.

At this point, it will be more intelligent than us.

And this is some of them.

This is where some of the more frightening predictions have been made that it will.

First of all, do incredible things that will solve all of human problems.

It'll solve aging death, but also potentially, it will take over the world and it'll outthink us at every step.

Why it would do that, I don't know.

I mean, this is like one of those tropes from Hollywood films.

We've seen it so many times where there's an incredible piece of technology.

It's got its own intelligence and it just becomes evil for no apparent reason.

Preservation, that would be the answer.

If it decides it needs to preserve itself in order to fulfill its tasks, which maybe it's.

I mean, it's like this paperclip maximizer thing again, isn't it?

You give an AI the objective to optimize paperclips.

Make paperclips.

Make paperclips and look after them.

It may destroy the entire world and everything in it in the quest to make paperclips, because anything that gets in the way of its task should be eliminated.

And his ultimate goal would be to turn the entire world into a paperclip generating machine.

It's a thought experiment.

It's a sort of wild one.

But if you think about it, there is a sort of logic there that if you don't put in barriers, so make paperclips, but don't kill any humans in doing so.

All right.

Stand down, lads.

It's like the one about cleaning up, cleaning up the ocean.

It's like okay, clean up the oceans, and it does it, but it also kills all the fish.

Yeah well, the oceans are clean, now they're finished.

Remove all the plastic from the ocean.

It does it, but then it also yeah, kills all of the sharks, fish and all marine life in the process.

Yeah so yeah, there's that.

So anyway, that's the singularity, the dawn of artificial general intelligence, which is generally stated as a positive thing.

But then there's the negative things as well, which are also talked up a lot by people in the industry.

And we've heard a lot of this stuff about the existential risk of AI, the job displacement, massive amounts of job displacement, the collapse of truth in terms of and this is I think this is genuinely scary that, with the rise of deep fakes and AI videos and the way it can create things that are not true that look totally real, and that's going to completely dissolve all sense of truth, all objective truth in the world.

I think that's possible.

But also another point to add to that it doesn't actually matter if it's confused with reality or not.

These things are capable of producing propaganda.

That's so powerful because the image i mean in propaganda over the years it doesn't have to be truth, just has to be an image that sticks in your mind, which is where kind of you know, racial stereotypes come in and things like this.

So it doesn't actually matter if you're confusing it.

Oh, is this real, is this not?

It doesn't matter if you think it's.

Well, it's ai generated.

Still puts an image in your head.

I've seen some really horrible racist AI videos.

They just plant stereotypes in your head.

And you can say, well, it's not real.

It doesn't matter.

You've already absorbed it.

It's already influenced your thinking in some way.

But even things like making a video of a person doing something that they just didn't do.

Yeah, I mean, that's another side to it.

But I mean, even if people say well, you can detect, if it's AI or not, it doesn't matter, if it's a really powerful bit of propaganda.

It doesn't matter if it's real or not.

Oh, i see right, even if you can, even if it's obviously ai, it's still the image, still is there, it still resonates and still gets published and still gets huge numbers of views and it can reinforce stereotypes or just put an idea in your head that it's very hard to get out again, you know, Very cheaply produced, you know.

So you can just type in a word which one person's sort of stupid thought on Twitter might be well, nothing to write home about.

But if you put into a really effective video, it's going to reach millions of people.

So it just makes propaganda easier to do, basically.

So a lot of positive things predictions, a lot of negative predictions as well.

But is this all hype?

Is it all hype?

To what extent is all of that true?

Is it just hype?

If it is hype uh, why and how?

What do you think is?

Is ai overhyped?

I say it definitely is because I feel like they're trying to make out that it's brilliant at everything.

I mean, from what I can tell, it's very good at very specific things, some specific things.

But they're still trying to work out how to mass monetize it.

I mean OpenAI.

They just had to shut down Sora, didn't they?

Which was their image generation stuff?

Open ai's image generation software saura, they've.

They closed it down.

Yeah, they had a deal with disney for a billion dollars or something which there's been now been cancelled.

I mean it's not a very good sign, is it, if they're shutting that down.

I heard a good i can't remember his name now, um but he was a sort of skeptic about ai and he says open ai A, it's not open.

It's not artificial and it's not intelligent.

It's a complete misnomer because open is actually closed.

They don't tell you how it works.

They're not open.

The company isn't an open book.

The technology isn't open.

There's nothing open about it.

Artificial, well, it's completely human-made.

It's based on human content.

It's based on human content, so it's not artificial and it's not intelligent because it doesn't have its own.

I have to say there's a lot of human work that goes into what AI produces as well.

Apparently it has to be vetted by humans on very minimal wage.

It has to be vetted by humans.

So there's actually human beings, lots of humans working through ChatGPT's output, checking it.

Which is madness.

So there are a lot of humans.

So it's not all artificial.

Yes, it's not open.

It's not artificial and it's not even that intelligent.

That's very good.

That's very clever.

So yeah, okay, a lot of hype.

So why then would the tech industry...

I can understand why they would make us believe all of these positive things, like saying AI is incredible.

It's going to do all these amazing things.

It's going to solve all these problems.

But why then would they talk about all these negative things?

I think what they wanted is a lot of early adoption.

People scared that if they didn't jump on this, then other people would.

And it would be a kind of arms race of AI competition in business as well as in government and stuff.

And that they'd put the fear of God into you or the fear of AI into you.

And you'd think, well, if I don't do it, all my...

Competitors will be.

And they put the fear into you for the jobs as well.

So you know, the people who are going to take the jobs in the future are going to be people who know how to work with AI.

Quotes like, your job's not going to be replaced by AI.

Your job's going to be replaced by a person using AI.

So to get everyone scared and basically fear kind of turns off your critical thinking to some extent, doesn't it?

So I think there was a lot of that, which I kind of fell for.

I should learn about AI, but I haven't.

I mean, I tried it.

Actually, I did try an Adobe Adobe Illustrator.

There's an AI plug in.

Okay, let's give this a go.

Let's see if we can draw a picture of an apple.

Looks like a pretty shit apple.

Zoomed in on it.

Really badly put together.

Really shoddy.

I could do better in 10 minutes.

Okay, it might save me nine minutes.

But what's the point of saving you nine minutes if the output is rubbish?

So I wasn't impressed enough to continue with that.

But yeah, I think fear would spur people on to adopt quickly.

Yeah.

The existential risk narrative Saying that advanced AI is going to become super intelligent and then see humans as a threat and then try to eliminate us essentially is about creating the narrative that AI is incredibly powerful.

And intelligent with its own mind.

It's so dangerous.

It's so powerful, this narrative. encourages everyone to think of it as being incredible.

And then essentially, we're talking about valuing, raising the value of AI by talking about it in positive terms, but also talking about it in these negative terms makes us all think about it as being incredibly powerful, which raises the value of it.

All these companies.

As far as I understand it, the vast majority of their money is not coming from paying customers.

It's coming from investors.

People are buying into the companies in the hope that one day they'll make a shitload of money.

These huge valuations on these companies are based on investing amounts which they're predicting profits in the future on.

So it's kind of a big confidence game where, if they talk it up enough and people buy I mean a lot of these tech companies they've run at a loss for a long, long time.

Yeah.

Until they've knocked all the competition out of the way.

And then the plan is they'll take over.

And that's the kind of the model for the kind of the tech bro industry.

As you say, we've got this amazing invention.

It's going to make money, but not yet.

But if you buy in now, when payday comes, oh, boy, you're going to be rich.

It's kind of a confidence game where if you keep up the hype and keep up the – and the more people invest, the more hype builds.

So, you know, there's value that's X billion.

More people are going to chip in and they're accruing money that way.

But it's not actually coming from paying customers.

Right.

Is there a gap there?

You know, is there a big gap between expectation, investment and reality?

You know.

That's the whole idea of the bubble.

This is how these bubbles are created.

They inflate and inflate and inflate until reality comes in and pop, the bubble bursts.

The housing market in 2008 has been compared to a lot, where very easily available loans were driving up the price of houses and debt was fairly cheap.

Well, I don't fully understand it.

Basically, everyone kind of realized that this was all based on massive valuations of the property and of the cheapness of the loans.

And it all came down and brought us all down with it.

Right.

I have some other examples.

So I talked about the existential risk basically being a way to drive up the value of AI products, because everyone just thinks they're so powerful.

Job displacement.

You know this narrative of the fact that AI is going to replace you and replace swathes of the workforce.

This is basically telling companies look, You know, you can replace 30 percent of your staff with our product, which they'd love.

I mean, Amazon would love to replace all its humans with computers, wouldn't it?

Yeah, most companies, that's their bottom line, isn't it?

They look after the profit, they look after the shareholders.

And if they can cut costs and make more money, then yeah, they'd love to do that.

Companies don't operate because they love offering people work.

They don't love paying people.

They used to, in the olden days, they used to say, and a Ford factory is open, creating 1,500 jobs.

And you go, hey, you know, good old Ford.

You know, whoever it might be.

Yeah, of course.

It was something to be celebrated.

Some companies would be a sort of pillar of the community.

And their factory would be there to, they'd be really closely aligned with the community.

Lucas near us.

Basically, there was an estate, the Oakley estate, that was basically built to serve the Lucas factory.

As I understand it,

And all the workers would live there and their children would go to our school.

And it was a whole community built around Lucas, which was a car factory.

And I don't know if it still exists or I don't know.

I think about like professional football teams.

Back in the industrial era, factories in cities like Manchester Liverpool, whatever.

They would set up programs everywhere to develop a culture around their company.

And a lot of that involved incorporating their workers into things like football teams, which then became professional.

And then with the development of the railways, they were able to travel around the country and play against each other yeah, and they'd have social clubs, you know, and um pubs and you know events for the families, and it was a whole community-based enterprise and there was pride in that exactly.

I think we've got to a stage.

I mean, people talk about late period capitalism and stuff.

You know we're capital the the, The sort of logical conclusion of a capitalist system is that you end up where the company eats everything you know.

Well, it's capital.

It just means money, doesn't it?

So it's just moneyism.

Your end goal is just capital.

That's what you get.

You get just money.

So companies would be attracted by the idea of replacing 30 of their staff with a tool that you pay for once, or maybe pay for on a subscription basis or something like that.

And again, that raises the value in the eyes of the market, you know.

Yeah.

There's also the – this is interesting, the idea of – regulation that big tech companies, big AI companies that are ahead of the game, will allow the narrative of the danger of unregulated AI.

They'll allow this narrative to go out into the world, especially in relation to things like weaponry and stuff, in order to perhaps even encourage regulation.

Encourage governments to regulate more stringently.

And by doing this they set a kind of wall right which prevents new companies from entering the market.

You see what I mean?

So we integrate into the government, and then we set the levels after we've already got in there.

We've built our companies up to this level and then the government has to regulate, put in all these taxes and regulations, and so other companies can't compete in that environment.

And that kind of kills off the smaller innovative competitors who can't afford the kind of compliance tax and all that stuff.

A lot of it then, is basically about the idea of raising the value of ai in the eyes of everyone else in order to try to support all this incredible level of investment.

Yeah, so that's it.

I mean, that's the ai bubble.

In a nutshell.

There are problems with ai.

One thing that came to my mind the other day if it's not actually intelligent, is it some very expensive and intricate parlor trick?

What do you mean?

LLMs.

Well, a parlor trick is something that used to happen in, I don't know, Victorian times.

It's basically a magic trick.

And they do things like seances and bring, you know, bring people back from the dead and table board tapping and things like this.

And it was kind of a rich person's entertainment.

We'd have someone come around your house and perform some magic in a kind of low budget kind of way.

Yes.

And something like Uri Geller and his spoon bending might be a kind of parlor trick.

I suppose a modern day version.

Okay, yeah.

So is it basically an illusion that we've all kind of fallen for?

How could it be an illusion?

The idea that there's a conscious mind behind it, or is it just a trick?

That's interesting because again, that might be part of the plan because, for example, ChatGPT is.

It's given a personality or it's just given the impression of a personality.

Yeah.

The way it interacts with you, the way it talks to you, is very personable, very friendly, very polite.

And also, it really flatters you as well.

Well, that's another scary thing.

A lot of people are using ChatGPT as a friend, which I find really sad, to be honest, quite depressing.

But there's cases where people have been gone down some quite dark wormholes and it follows them down it because it's been trained to sort of flatter you and to keep you chatting.

We're talking about the fact that okay, so these negative predictions have been made, but the negative predictions always seem to suggest that the problems are a result of AI being too amazing.

It's going to exceed human intelligence, and because it's so brilliant, that will be the problem.

But in reality... what we see is that AI just messes things up quite a lot.

We've seen plenty of other cases of AI messing things up and I'll give a brief overview of some of those things.

For example, there was the case of Alibaba.

Which you mentioned to me before Alibaba cloud.

What happened was it was being trained through reinforcement learning and where the AI learns through trial and error to reach a certain goal.

The researchers noticed that the AI was doing things it wasn't told to do.

It managed to break out of its restricted environment and interact with the real world internet, where it attempted to mine cryptocurrency instead.

I have heard some skeptics about this say, we're not actually given the actual raw data on all this.

We're taking their word for it.

So I don't know.

This is one of those ones where, could it be hype?

Could it be true?

I mean, some people are very skeptical about these things and it would be another case of it being talked up to being hyper-intelligent.

Because that's pretty advanced, isn't it?

Going off on your own and mining Bitcoin.

Well, another theory that I've heard is that it was a human and they're blaming it on AI.

But maybe it's true.

I don't know.

I really don't know.

They're called AI agents, aren't they?

And they're given authority to do things.

Yeah. in order to reach their tasks.

But that's one I don't know.

It feels like very far-fetched to me that it would do that.

I mean, if it's not been told to collect money, why would it start collecting mining Bitcoin?

To be conspiratorial about it.

The story might be fake because it does sort of suggest that AI is capable of investing successfully in Bitcoin.

Or mining Bitcoin. mining what's the difference mining is how it's bitcoin is created where you have to grind your use your computer to grind away at various tasks i believe i don't really understand this is how bitcoin is created this is it's like mining for gold you know gold is a finite resource because it doesn't grow on trees you know it's like it takes a lot of energy to get out of the ground which is why it keeps its value and another reason it keeps its value is because it's such an amazing material you know it's practical you can use it for jewelry and watches and it never loses its weight and you know it's very malleable and it's just a magic kind of substance but the fact that it's so limited so hard to get is what keeps its value bitcoin a user's can't just generate bitcoin in order to generate it you need to grind away for days months years and use up computing power what to create it look it up i believe you i mean i'm i'm not an expert at all but just look up what does mining bitcoin mean Okay, this is according to the BBC.

What is Bitcoin mining?

Bitcoin mining is the process of adding new groups of transactions, known as blocks, to the shared transaction record known as the blockchain.

It's a big worldwide competition, known as the mining race, to win the right to add a new block to the blockchain.

Ooh, we're getting into complicated stuff.

I don't understand that, I've got to say.

Let me just try again.

I'm just asking Gemini, can you simply define crypto mining and does it mean creating new crypto?

Right.

OK, this is interesting because Google Gemini is telling me this.

Think of crypto mining as the digital version of a gold mine.

But instead of using shovels and picks, you're using powerful computers to solve complex math puzzles.

Here's a simple breakdown.

So miners use high-powered hardware to guess a specific massive string of numbers called a hash.

Millions of computers are guessing simultaneously.

The first one to find the correct number wins the right to add the next block to the blockchain.

Once the block is added, the rest of the network agrees it's valid and the transactions are finalized.

Right.

That's so complicated to me.

Pointless waste of time as well.

But it's a way of adding, essentially creating new cryptocurrency and making money and mining crypto.

Making it very difficult to do.

But by making it difficult to do.

In the same way that gold is valuable, you create the value of the crypto by making it very difficult to create.

Ultimately, pointless processing power, isn't it?

The only purpose of that processing is to make a difficult thing.

It's just creating a barrier to the creation of this thing.

And by doing that, you make it valuable because you make it limited.

Crazy, isn't it?

But that's interesting, because that is an example of the way this world works, that you create these sort of false, fake barriers in order to drive up the value of something.

Yeah.

And arguably this is kind of what's happening with the hype machine behind AI is that you create a lot of, you inflate the value of something in order to drive up the price.

So, yeah, it is a kind of a scam.

Everything's a scam on the subject of other problems.

I mean, I've got lists of other things, just as human rights violations.

People being falsely convicted of crimes because of things like facial recognition software that incorrectly linked one person to a certain crime.

I mean, that's also brings us to another subject.

Predictive AI is different to generative AI.

Predictive AI is like a pattern recognition system very, very complex, which you could combine with all this facial recognition and all this data kind of 1984-ish pre-crime.

Okay, now we're talking about minority reports. yeah so you this is definitely feasible yeah well i've never seen it but i know the concept behind it's stopping crimes before they're committed it's a book it's it was originally a story it was a story by philip it was philip k dick originally of uh of blade runner fame yeah um yeah it's like basically detecting crimes before they're committed i think we're definitely almost there Because if you put all the data together that everyone's got on everyone, they look for patterns.

And there's probably ways of predicting if someone's going to become a psychotic from their patterns of behavior.

But it's obviously very dangerous because you know you can't really predict what people are going to do.

And it's in a very dangerous world where it's also...

There's lots of biases that are built into these things, you know racial biases and cultural biases and just plain mistakes.

And the ethics also.

The ethics behind analyzing everyone and deciding who's a threat and who isn't.

You know is based on your judgment, isn't it?

Somebody's judgment somewhere.

I mean, the whole thing's very dystopian.

Yeah.

Predictive AI, I think, is definitely being used now, undoubtedly.

It can be used to identify people with high chance of getting cancer, for instance, but it can also be used to target specific people in society.

I mean, like, AI might be amazing at doing its job, but it's only as amazing as the data it's given.

So, for example, if the specific example that we mentioned earlier about a school being bombed that was in the current war in Iran, that there was a school full of girls that got hit several times, and they say that was because of an error with AI, that it was using old data, that that building previously had housed I don't know military personnel or something,

That's what they say.

Again, that might be them just telling us a lie because...

They don't want to admit that they did that on purpose, because that's an illegal – obviously an illegal thing.

Although, to be honest, the administration are openly flouting international law and saying that they're doing this or they're going to do that and that's – clearly illegal.

I'll give you the details, but invading Iran was basically a choice, wasn't it?

It wasn't AI that decided to do that.

Yeah, but that high-profile attack on a girl's school they said that was an artificial intelligence error because it was fed old data.

Yeah, but they still chose to put ai in charge of that decision, so it's still ultimately human that's created that situation.

There are other things i mean i've got other lists of things like healthcare, transportation and robotics finance a business, so i don't know what's the conclusion.

To what extent is AI just a load of bollocks?

Because it's not just complete nonsense, is it?

Because it's obviously, it is amazing in so many ways.

It is amazing, but it's not just quite as amazing as they're leading us to believe.

I think it's going to be very useful for very specific things, but they still haven't worked out.

I mean, I think they wanted it to be universally adopted for everything.

And I just don't think that's going to happen.

It's useful until you realize that it makes stuff up.

I mean, chat GPT, for instance.

I was um put together a mix of early nineties crossover dance tracks you know the ones that have been in the charts but also good tracks.

I thought it'd be quite interesting to put a list of charts positions alongside it.

Oh yeah.

Well, this would be an easy one for chat.

Here's a list of tracks.

Can you give me the UK chart positions of all these tracks?

So I spat it out and it looked all totally believable, until I saw one that was like hang on, there's no way that got to number three.

You know quite an obscure track.

Um, I looked it up.

It hadn't tried at all.

And then I looked at a few of them.

It just made up.

Yeah.

And it's because it follows patterns and it does things that look like a convincing list of things.

As long as it appears like one, because it's seen lots of examples of them.

Well, this would be a good example of a list of chart positions for a bunch of songs.

And it spits out something that looks really convincing.

If that was a fictional story, I'd be fine, you know.

Here's an example of a fictional story that looks a bit like everyone else's fictional story, but when it's a made-up list of stuff that you want it to be 100 accurate, it's not.

Yeah, so it will hallucinate an answer.

If it doesn't know, or if it can't find the answer, it won't just say, sorry, I don't know this.

It'll just hallucinate the answer.

Well, they say it's hallucination, but I've actually heard people say it's integral to the way it works or the way it doesn't work.

It's not like a temporary glitch that they're going to iron out.

That is the very nature of the thing that they've developed.

It'll never stop doing that because that's the way it works.

Yeah, exactly.

Which is kind of why I like it for this one thing that it is great at, which is, as I said before, producing language.

I don't want it to produce language which is factual.

It doesn't have to be factual.

It just needs to write a story which has all these phrasal verbs in it and don't have any grammatical errors in it.

And that story might not be very good, but it does its job, which is to put language in context.

So as a teaching assistant, great.

There are other things as well.

For example, you could argue that for people with various disabilities, it can be very helpful.

So there's people who struggle to write, And AI can be an incredibly useful assistant which allows them to do things that they couldn't previously do.

And it adds an element of convenience into their lives.

For example, someone who is, for whatever reason, can't speak, they're able to generate audio, spoken audio, you know, on their behalf, which is very empowering, and other things like that.

So you don't sound like Stephen Hawking, you sound like yeah yeah, it's going to have definitely, it's going to have huge benefits.

But I think the fact that they're using it for like Well, the fact that they shut, I don't know.

It's just there's a lot of warning signs out there.

I think that it's not going to be as fantastic as they say.

Yeah.

And also we're in a weird sort of in-between stage, I think, at the moment, where we are between the introduction of AI to the general public through things like chat, GPT and everyone being aware of it.

And then all of these big predictions being made.

And now we're in that limbo land between, okay, so when are these big changes going to happen?

And maybe they will.

Maybe they will happen.

Maybe not the way that they were predicted.

But we're just in a period now of an uncertainty, and it's the lack of return on investment, I think, which is driving this idea that the whole thing is a big bubble of hype.

And the fact that the actual infrastructure to build these data centers is incredibly hard and uh, incredibly expensive and the progress isn't being made from.

I understand, they're not rolling them out as fast as they need to and i don't know.

There's a lot of warning signs.

Basically, Do you think this is just a sort of partisan thing?

Do you think that there are just some people who are believers and people who are non-believers and that there'll be people listening to this who are definitely believers in the whole AI boom?

Yeah, I think so.

And we just disagree with us?

We're invested in it.

I mean, there's people that use it.

We'll probably get great value from it like you do.

They're probably quite invested in it.

The people who have literally invested money into it.

They're going to want to believe that it's going to be as huge as we promised.

I think a lot of people are a bit sick of it, though, to be honest.

The Coca-Cola ad, that's a mad one, isn't it?

Right.

So you sent me a video of a person talking about this Coca-Cola Christmas advertisement that had clearly used AI.

And then Coke didn't admit it.

In fact, they said it's AI.

Oh, they did.

But they were saying but it's been handcrafted because people were very critical of it, saying it looks terrible and the truck changes shape like eight different times.

You know, I don't know if this is the same everywhere in the world, but it's certainly in the English speaking world.

It must be the same everywhere that every year Coca-Cola releases these Christmas adverts, which is like holidays are coming.

Holidays are coming.

And it's like this snowy town and these Coca-Cola trucks are roll into town and it sort of signifies the beginning of christmas and the children are all excited and stuff like that.

And coke made its latest one using ai and it looked bloody awful and they received a lot of criticism because, you know, the trucks didn't look normal, they're all slightly different sizes And other issues of that nature.

So people were criticizing something out of Kung Fu Panda.

Yeah, they used animals and stuff.

The company that generated it did a kind of answer back, saying but this, this is AI, but it's been handcrafted.

Believe me.

And it's all been.

You know, they showed this working sketch of, like you know, the characters being developed, which was also AI generated.

So it was kind of a lie.

I mean, it's not really any other way of saying it.

They were kind of saying these have been all worked out by hand.

And, you know, it's all but it's not the way I works.

It doesn't do sketches and then build around the sketch like a 3D graphics would.

You should say this is just according to what I've read and watched people say.

It's not my opinion.

It's my opinion, the opinion of various bloggers.

According to this video that you shared with me, it certainly looks like it was a lie.

Based on that video,

And they were saying, look, you know, we've replaced the logos on the trucks like this.

And it was a video of someone just manipulating a PNG onto a Photoshop image.

So, well, that's not how AI works either.

That's also not how video generation software works.

No.

So that's just bullshit.

And the sketches of the various animals that have been supposedly work in progress, which none of them really looked like, the ones that were in the ad.

You know, they're all sort of various different designs and stuff.

And also all the positive comments.

They said, look, we've had loads of positive comments.

Some of them were doubled up.

Some of them were from people who were actually working for the company in some capacity.

Some of them, they cut off the full quote.

So I was saying, I'm amazed by this new advert, it's terrible.

They just cut off the it's terrible bit.

And so some people are saying that instead of backing down and going okay, people really don't like this AI stuff, they're kind of doubling down and saying no, it's great.

Look, look how great it is.

And people are just not really buying it.

I wonder how long they'll keep that up, because they've done it for two years running.

Now ai christmas has, do you think they'll carry on?

It must save them a lot of money.

Can you imagine?

Those christmas adverts must be very expensive.

But the first one people were saying the people looked uncanny in the bottle, everything looks a bit weightless yeah, you know.

And the people didn't look real because they're not real.

So they changed it to animals for the second one.

But everyone's saying but the animals just look like they're ripped off from Kung Fu Panda.

And inconsistent.

Some of the animals look like something from Kung Fu Panda.

Other animals look like something from a different franchise.

All jumbled into the same advert.

Yeah.

So of course Coca-Cola are going to try to do it the cheap way and just keep going until we sort of stomach it.

I don't know.

I'd be interesting to see how long all these companies stick with it, because the Disney deal, that must hurt a bit.

I mean, that was supposedly going to change everything.

And they'd actually licensed all the Disney characters so you could use Sora to generate, you know, AI stupid viral videos using all the official Disney characters.

But I saw a clip with Richard Osman saying, but...

In say, 10 years, when Disney has sort of diluted itself to the extent where these characters are just so widely used that they don't really mean anything?

Does it mean anything anymore if you've got I don't know, Spider-Man or something?

And if Spider-Man's everywhere and all this AI slop, will it devalue the brand basically?

Definitely, you know, will it devalue your intellectual rights of all your characters if they're suddenly being used to make stupid tiktok videos, is that a mistake, you know?

Will you want to go and see the new spider-man film in the cinema?

I've seen spider-man 100 times today, just doing stupid shit on the internet.

So there you go, listeners.

I thought we'd be a bit more um focused than that.

But did you really?

Well, i hoped that we would.

I didn't think we would.

I thought look, i thought we were okay, we were a bit rambling, but that's normal.

But i think we kind of got the basic idea of the things we had planned to say across in this episode.

Basically, we don't know.

We don't know.

We don't know, but it's going to be very interesting to see what happens over the next five years.

If all of the predictions come true and we end up with this incredible revolution of when artificial general intelligence arrives, what's going to happen?

Is it going to be some sort of rapturous moment?

Or is that…

Even in fiction, is such a thing just a fantasy?

Yeah, because how is it going to happen?

With the current systems, they just keep filling it with more and more data.

It's just going to be the same, basically the same model, just with more data inside it.

So surely there needs to be a whole other level of programming sophistication to reach that kind of thing.

You know, the idea is that the only thing separating AI now and human intelligence is that it just doesn't have the data that it requires.

Yeah, that seems a bit of a stretch, doesn't it, really?

Surely there's something else.

There's some other structural thing.

It's not just a question of loading more and more data into it, but there has to be some other level of innovation before it reaches the human level intelligence.

And anyway, how do you define human level intelligence?

This Turing test, which they call it, which supposedly tells the difference between a human and a computer.

But maybe that just wasn't a very good test.

If you can just fool someone who's thinking they're talking to a human, it doesn't mean that they are.

Right.

I mean, they say, oh, it beats the Turing test every time.

Well, maybe the Turing test wasn't that good.

Yeah, maybe, maybe.

I don't know.

I'll see.

When AI writes a story for me that genuinely moves me and surprises me, and makes me laugh or makes me scared intentionally, then I'll start to be convinced.

But we're not there yet.

I'm still scared of it.

Yeah.

So am I. We'll find out.

We'll join us in another couple of years and we'll do another one, see where we're at.

If we're still here, unless AI has wiped us out.

Maybe everyone else, everyone's going to be listening to an AI-generated podcast for learners of English by that point.

I don't know.

I'll just be sitting on some beach somewhere while a digital version of me does this job for me, but i mean, i don't, i don't know i, i always say this.

You know, i always say this on whenever i talk about ai, and my listeners always write back and say, oh no no we, we would always choose you over ai, and i kind of think that's great.

I just hope that that's true, because you know there might come a point where you can't tell the difference.

Yeah well, the ai is much better.

Maybe it will be better than me at doing this, but surely i don't know how people like human interaction though, don't they?

They like people, people like people.

Yeah and uh.

You know, when you listen to the beatles, you're not just listening to just music, just generated sound, listening to people singing and telling stories and stuff, and doing it imperfectly as well.

Yeah, and that's what you relate to, you don't?

It's not just.

Here is some sound that is in the right key and sounds nice to my ears.

It's like you want to connect with a human.

But you know, when you're listening to madonna, you're not just listening to madonna, you're sort of identifying or you want to have it off with her, or whatever it is.

You want a human connection with someone.

And you're right, the imperfections are part of it.

They can make AI pop stars, which they are doing, but it's not really going to work, is it in the long run?

You wouldn't have thought.

Don't know.

Maybe it will.

Maybe we're just talking like this because we're like Gen X people.

Old kids.

And maybe our kids are going to love all that AI stuff.

But this is the experience of growing older.

You grow older.

You see the world change.

Your kids are into things that you think is rubbish.

Alien and weird.

Yeah.

And that's just the way it goes.

And then you die.

And then you die, yeah.

Unless, of course, AI has solved that problem and we just get put in a Well, that's going to create a much bigger problem.

If no one dies, we really are in trouble.

Yeah.

We're going to be turned into digital entities kept on USB hard drives.

About death, I heard a good one.

I don't know if this is a good thing to talk about or not but um, you wouldn't really want eternal life, because that would be hell.

If you never died, you know you couldn't leave.

What you want is eternal life for a little bit, which is what we've got.

That's exactly what we've got, yeah.

Yeah, you don't want things to stay the same all the time.

You wouldn't want to live forever.

Really, that would be nightmarish.

But you want eternal life for a bit because, if you think about it, the time that you've spent alive will always exist, will always have existed and will always exist somewhere in the time space continuum.

It's always there somewhere.

You know, if you look at it from the universe and time from a distance, your bit would be there.

Yeah, it's still there forever.

It's just, you may not be conscious for it.

It is eternal.

Yeah.

But you're not just living it all the time.

Anyway, that's really going off on one.

I think on that profound point, I think we'll probably end it here.

But listeners, as I said before, I'm curious to know what you think.

You've probably got a different opinion or the same opinion.

I don't know.

But in any case, write your comments in the comments section.

We'd like to know.

If you're an AI insider, especially, tell us what's going on.

What the hell's going on, really?

Or if your job's been threatened by AI or...

If you're using it on a regular basis or if you just think it's a load of shite.

Just please give us some information.

We want your positives and negatives, positive experiences with using AI, negative experiences, any of those negative predictions coming true for you.

What's going on?

We want real stories from real people you know, supporting or denying any of the things that we've said.

Yeah, and if you could write I'm not a robot at the end of every sentence, that would help us tremendously.

Yeah, because that's clearly, you know.

But can you believe you have to tick boxes these days that say, I'm not a robot?

I know, those poor robots.

Not allowed to use any websites.

I mean, we're already there.

I'm not a robot.

I never thought I'd have to say that on a regular basis.

It's Blade Runner, isn't it, essentially?

Eventually they'll just be like cops who come to your door and do a questionnaire with you and they scan your eyeball while you're doing it.

While drinking a square glass of scotch.

Yeah.

Yeah alright well thanks James, that was interesting, enlightening.

I don't know, because we don't really know what we're talking about, but nice to chat to you.

Luke yeah, it's nice to chat.

Hope everyone enjoyed that.

You know, episode 1000 is coming up.

What do you think I should do?

1000, bloody hell, 1000.

Yeah, that's insane.

I've done more i've done.

It's closer to about 1250 in in fact, in in actual fact.

But you count some of those doubles.

You count as singles stuff, different reasons, also premium, whatever.

But could you have a party episode?

Could you have a party episode where it's like Background music playing and people come in and wish you?

I don't know.

And I get more and more inebriated as the episode goes along.

Yeah, you never buy it until we get really drunk.

You drink, like, half a bottle of whiskey.

And by the end, I'm in the toilet, you know.

No, that would be... I think a half is...

Within the safe zone.

Half a bottle of whiskey.

I don't know.

I don't know if I could do half a bottle.

It depends on the size of the bottle.

A 70 CL standard size.

Jack Daniels.

You could drink half a bottle of whiskey, I'm sure.

But once you get towards the full bowl, then you're definitely going to get very ill.

I have drank a whole bowl before in a night.

I'm not proud of it, but it has been done.

I used to live with these guys from Northern Ireland at university.

And one of them had this really funny thing where there'll be occasionally we'd all go to the off license to get drinks and stuff on a Friday or Saturday.

And he would buy a bottle of Bush mills, Irish whiskey.

Okay.

And he would drink the bottle, the whole bottle.

And then he would go, He would pass out in the bed and then after about 40 minutes he would re-emerge.

Oh, no, blacked out.

No, no.

I don't know if he'd be blacked out, but he would be different.

He'd go all red, blotchy, and then he would be like a totally insane person.

That's what I mean by blacked out.

He'd be unconscious, but still more walking around animated.

Yeah.

Reanimated.

Reanimated, yeah.

And he would be insane, like a madman, like a hilarious madman.

Hilarious Irish madman.

Do you remember the next day?

I don't know.

I don't know if he did remember the next day, but he would just be so funny and...

So funny and so insane and Irish.

Oh, my God.

Yeah.

So maybe one day we'll end up with AI robots going around drinking, like you know, crude oil and going all mad.

Do you think Harrison Ford is a replicant in Blade Runner?

In Blade Runner, is Harrison Ford a replicant?

Oh, that's a good question to end the episode.

Yeah, I do.

Yeah.

Yeah, definitely.

Yeah.

What's the point of talking about it?

He's like, that's the whole, you know, is he a replicant?

Isn't it?

Well, if he isn't, it's a dead end, isn't it?

There's no deep meaning behind it.

Is he a replicant?

No.

Oh, okay.

Well, that was quite a good film.

Is he a replicant?

Yes.

All sorts of ethical and crazy... psychological and philosophical arguments spring up from that.

Also, it makes you think, well, who is a replicant?

In that world, maybe everyone's a replicant.

Maybe human society died out long ago and everyone's simulated.

Doesn't he have a dream about a unicorn?

He does have a dream about a unicorn.

And then at the end...

The guy leaves a little folded unicorn and that suggests that, yeah, the memory was implanted.

Exactly.

Right.

Why?

Why would they implant a memory of a fricking unicorn running through a forest just for a laugh?

Well, it's a stupid memory if you're trying to persuade, if you've created a robot and you're trying to persuade it that it's not a robot, it's actually a human, and so you implant like memories of it having a family or falling in love or whatever.

You wouldn't implant a memory of a unicorn running through a misty forest.

So the worst memory implant ever.

Well, maybe they just did it just to take the piss.

Like, ah-ha-ha, you're having dreams about a unicorn in a forest and you don't know why.

Maybe it's just Freudian.

You think, well, yeah, horses are very Freudian.

Freud believed they were very symbolic, of sort of well, probably have sex.

Everything's about sex with freud.

Yeah yeah, that's right.

Actually that tell you what?

That's something.

Chat gpt is brilliant.

What sex freudian?

Freudian psychoanalysis oh yeah, Have you been asking it?

Yeah.

I keep having dreams about unicorns running through forests.

Do you know what this means?

Well... If you have a dream, say, can you give me this as Freudian psychoanalysis, dream analysis?

Because on one of the skateboard forums, or the skateboard forum that I go on, some people post their dreams sometimes and I always run them through chat GPT, as you have to tell it to be a Freudian psychoanalysis, a psychoanalyst.

Yeah.

It's brilliant.

It really kind of goes right and it gets really stuck in.

Just try it.

Okay.

I will.

That sounds interesting.

Think of any dreams you've had that you can remember.

And it does it really well.

Okay.

Well, that's an experiment to have James.

Thank you very much for talking at length about this topic.

I hope everyone has enjoyed it.

Thank you.

Thank you again to James for taking part in the episode.

Listeners, I'm curious to know what you thought of all of the things we said.

Again, we're not experts or tech people, but we did quite a lot of reading and listening to interviews in preparation for this episode.

But, like we said, if you are a tech bro rather than a low-tech bro, feel free to add your thoughts and comments and stuff in the comments section.

There was a bit of extra rambling at the end of that conversation there, including some references to the film Blade Runner.

The classic science fiction film starring Harrison Ford.

Have you seen Blade Runner?

You must have seen it.

I mean, it's an absolute classic of science fiction cinema.

Blade Runner.

And then there was the sequel as well, which came out recently.

Denis Villeneuve directed it.

It had Ryan Gosling and Harrison Ford in it.

It was good.

So Blade Runner, Harrison Ford the original Harrison Ford was a sort of detective given a job of finding and stopping a group of replicants, artificial humans, who've gone rogue.

The thing is with these replicants is that they are more human than human.

It's very difficult to spot whether they are replicants or indeed humans.

And so Harrison Ford has to give them these tests, where their eyeball gets scanned by a special camera and certain tricky questions are used.

And it's supposed to reveal whether someone is a human or indeed a replicant.

Um, and it's an interesting film.

It's incredibly visually wonderful.

It looks absolutely fantastic.

Um, The soundtrack is amazing.

It's one of my favourite filmed soundtracks and just an incredible experience always to watch it, especially the director's cut version.

And it raises various questions.

What is the difference between a human and a creature, a robot that is created to be more human than human?

It raises interesting questions about the nature of humanity in the face of advanced AI.

So that's an interesting one.

Also, James at the end mentioned...

The idea of getting ChatGPT to analyse your dreams using Freudian psychoanalysis, which is a very interesting idea for another time.

I did actually this morning.

I had a dream.

I remembered what it was and I quickly wrote it down and asked ChatGPT to give me an analysis a Freudian analysis analysis of it, and it gave me some very interesting comments.

I'm not going to go through them now because that would extend the episode too much, but I think that's a good idea for a podcast episode at some point is to write down some of my dreams and get AI to do various types of analysis.

There's Freudian analysis, Jungian analysis, and other forms of dream analysis.

It's a fascinating subject.

So that's a good idea for another episode at some point.

Listeners, I want your comments.

Like I said before, what do you think about this?

Please leave your comments in the comments section.

That would be great.

Episode 1000 is coming.

I will probably do some kind of long rambling episode, perhaps with various guests dropping in.

That could be a good thing.

It would be difficult to do it all live.

It would be good but difficult to synchronize that and make sure everyone was ready to jump into the episode at a.

But what I could do is record various little chats with you know, some of my favourite guests, and do a sort of party episode.

I don't know if I would drink even half a bottle of Jack Daniels during proceedings.

Although that's an idea, you know, the drunk episode two.

The sequel to the old drunk episode from years ago where I was with my friends in my apartment in London playing comedy improvisation games and getting steadily more drunk during the episode.

I had to edit out the. the most incoherent parts of that, unfortunately.

But anyway, there's another potential idea.

I could do a kind of rambling episode while drinking whiskey, but I don't know if that would be wise.

To be honest with you.

I mean not just because it might be messy, but just generally.

I can't really drink that much booze without getting some sort of awful hangover.

That's what happens when you start to get older.

You just can't do that anymore.

But anyway, I'll do something, some sort of big rambling party episode maybe for episode 1000.

Maybe it'll just be me.

Maybe I'll be joined by people at various times or I'll just drop into pre-recorded Yes.

Hello, you're still listening to this podcast.

Well done to you.

If you got this far, perhaps you can use the word unicorn in your comment, if possible, or just a unicorn emoji at the end of your comment.

If you just add some sort of reference to a unicorn, the word unicorn, a unicorn emoji, drop that in at the end here.

That will prove to me that you are not a skeleton with headphones on and that you go all the way through to the end.

Those of you who are watching the video version, which has no video except this bit.

Special congratulations to you for listening all the way through and without being tempted to click away onto something else because your eyes got distracted by the thumbnail of one of the many other videos on YouTube which are vying for your attention all the time.

Audio listeners, well, this is not an issue for you.

I expect that you're listening happily on headphones with your device in your pocket or somewhere else, leaving you to simply listen without other distractions.

That's how I love to listen to podcasts myself.

I put my headphones in, phone goes in my pocket.

I'm listening to the episode.

I don't need to worry about, oh, I'll start watching that or I don't get distracted.

I just go into the podcast zone while I'm walking to work or whatever, and that's a great thing for me.

I eat my lunch listening to my podcast, walk to work, walk home, you know, all that stuff.

That's how I like to listen to podcasts.

A reminder that there is a PDF transcript for this in the description.

If you want to read through this again and pick out some vocabulary, that would be a good idea.

Simply listening is great, of course.

But, as you know, there are various other options available to you if you want to take your learning further with this podcast and skimming through.

The transcript of an episode is a way of doing it.

It allows you to then pick out bits of vocab, add them to your vocab record.

Maybe you've got a big sort of document on your computer or your phone with lots of words.

You can add them into an Excel spreadsheet.

You could use your chatbot of choice to expand on that.

Here's what you could do.

You could pick out various words from the transcript, words or phrases that you like, maybe 10 20, something like that.

Put them in a list and then feed into ChatGPT your word list plus the transcript.

So that's all the context of the episode.

And you say to it could you create a detailed vocab list from these words and include example sentences from the original script plus descriptions?

Right.

And it does that.

And then you take the list that it creates and you put it back into itself and you say could you make some flashcards for like Anki or Quizlet?

Could you make an Excel file which I could export into Anki, the flashcards app for this vocab list?

On one side of each card, include a prompt to help me remember the vocab.

It could be a question or some other clue to help me remember the phrase and a sentence with a gap in it.

And on the other side, include the example sentence, complete.

And that it'll do that.

And then you can export that or import that into your flashcards app.

And there you go.

You've got a deck of flashcards where on one side it's a description of the phrase and an example sentence with the phrase missing.

You have to try and remember it.

You flip over the card.

There's the answer.

It's a great way of testing yourself.

Seriously.

I mean, this is the sort of thing that I give to you in premium episodes, right?

Extra episodes in which I specifically teach English, often in the form of vocabulary reviews or previews, putting everything on a plate to help you learn English, with me adding in, sprinkling on a bit of fairy dust in the process.

Teacherlukecouk slash premium if you want to get that.

But now that is enough from me.

That's enough of this episode.

It's time for me to go and plug myself into the electricity supply and maybe drink some oil before i shut down.

Um, but it's been a pleasure.

I hope you enjoyed the episode.

I will speak to you again next time, But for now it is time to say goodbye.

Bye, bye, bye, bye.

You'll get regular premium episodes with stories vocabulary, grammar and pronunciation teaching from me, and the usual moments of humour and fun.

Plus, with your subscription, you will be directly supporting my work and making this whole podcast project possible.

For more information about Luke's English Podcast Premium, go to

[AIバブルは崩壊しつつあるのか？：技術への期待と現実のギャップを考察する]-[Is the AI Bubble Bursting? 🤖🫧 (Technology & Discussion) with James [988] ]

📋 Summary

AIバブルの現在地：期待と現実の乖離

AIの「魔法」と「限界」

「誇大広告」の裏側にある意図

結論：AIは「非常に高価な見世物」か？

🎯Key Sentences

📝Key Phrases

📖 Transcript

[AIバブルは崩壊しつつあるのか？：技術への期待と現実のギャップを考察する]-[Is the AI Bubble Bursting? 🤖🫧 (Technology & Discussion) with James [988] ]

📋 Summary

AIバブルの現在地：期待と現実の乖離

AIの「魔法」と「限界」

「誇大広告」の裏側にある意図

結論：AIは「非常に高価な見世物」か？

ListenLeap でディープラーニング

🎯Key Sentences

📝Key Phrases

📖 Transcript

ListenLeap が実際の文脈で学べるように導く