Discussion keeps the world turning.
This is Roundtable.
You're listening to Roundtable with myself, He Young.
I am joined by Steve Hatherly and you should in the studio.
Coming up, when a social media platform clearly states that it will use your input to train its AI model, what ethical considerations arise regarding intellectual property and consent should users rethink their engagement with social media.
And a New Zealand airport has imposed a three minute limit on goodbyes.
Why have they done that?
Wanna be part of our Roundtable Heart to Heart segment?
Share your thoughts, questions, or hot takes on news, tech, life, or anything you think of by sending us a voice memo to Roundtable Podcast at qq .com.
Our podcast listeners can find us at Roundtable China on Apple Podcast.
Now let's shift gears and actually tie off that tech thread we've been pulling through the show.
As technology advances, the risk of privacy violations increases.
Initially, individuals may willingly share their location data with positioning apps, but now they face the imminent threat of their posts being used to train AI models on social networks such as X, formerly Twitter, akin to China's microblogging platform Weibo.
Users are voicing their concerns about this shift in data usage.
In today's AI -driven digital landscape, how can we better protect our privacy?
So, what happened? Let's give everybody a little more details on what's happening here.
Yeah, of course. But actually, before that, I'd like to say, wow, today's topic is so quite tacky.
You're very tech so far.
I don't want our listeners to think about, oh, Huxuan is here then.
Today's show is going to be nerdy and tacky.
Too late. Too late.
I always explain this.
But anyway, so social media platform X, or formerly Twitter, unveiled its newest terms of service, which go into effect on November the 15th, and the updated terms of service said by submitting, posting, or displaying content on or through the services, you grant us a worldwide, non -exclusive royalty
-free license to make your content available to the rest of the world.
And it includes the right to analyze any of that content, including, for example, for use with, and training of our machine learning and artificial intelligence models, whether generative or other, another type.
That was verbatim. If you look at the X website, that's what you will see.
What he just said is what you will see on their new terms of service policy.
So the company updated its privacy policy and to detail the changes.
When the policy takes effect, users are automatically opt -in until they opt -out when, you know, all of your posts that you post on this platform will be used by this platform for the aims that they...
So Steve, when you first saw this and after listening to Huxuan sort of quote for us, then, you know, how worried should I be?
And it seems like a lot of people kind of have their hands up in air for this.
Yeah. I mean, when people saw the new terms of service, that's exactly the point that stood out to them by submitting, posting, or displaying content or through the services.
You grant us worldwide non -exclusive royalty -free license to make your content available to the rest of the world.
Now, for me, that doesn't really matter that much because I'm not an X user who's posting anything worthwhile, if you will, in terms of my own material.
I'm not a musician uploading music or samples that could potentially be used purchased by other companies.
I'm not posting artistic, creative material that I could potentially sell to other people or at least use it for my own personal use because I'm the creator of it.
Now, think about it from the other point of view.
You are one of those people.
And now things that you post onto X, that's going to be available not only for AI training, but it's just it belongs to them now.
I mean, this has just got people really, really, really upset.
And that point about if you continue to use X after the 15th of November, then you are automatically in.
You're in on this. So that's why people are very, very, very upset about this.
Yeah. And also one thing is that the fact that can you opt out or not.
So normally users can opt out of such changes by going to settings and privacy and safety and under the data sharing and personalization heading, toggling the data sharing with business partners option.
But opt outs aren't always cut and dry, noting that an AI model can't necessarily unlearn the data that it's been fat if the user opts out after the training has started.
So that is why I think most of the users are angry about this, especially those who are artists and creators.
This is affecting their bread and butter and or you call it the brainchild or whatnot.
So this is definitely a big deal for those folks.
But also I one thing I would disagree with you, Steve, I think for individuals, this matters too, because maybe you're not having the most fancy new invention with your use of words, but some people post a lot of personal information and pictures.
Exactly. So how do you feel if the AI model is training off the faces of you and your friends and your family?
And what about when you use these names, addresses or referring to different places that you hang out and all this information?
And I'm just using I'm just using a couple of examples out of a swath of personal information that gets somehow published on your social media account.
And that's all gonna be used.
And you know what? You don't get a dime for that.
No, you don't get any money at all.
I found this to be an interesting kind of counterpoint, though, because broad licensing with few limitations is not really an uncommon thing for a social media platform.
Alex Fink is the CEO and founder of something called OtherWeb.
It's an AI based news reading platform that targets misinformation.
And he told CNN that what makes X unique is that its new terms remove any ambiguity, if you will, in contrast to other platforms that don't really clearly spell out their intentions to their users.
You know, from personal experience, I guess you see that terms and conditions and it's feels like it's a 300 page document.
And people usually don't really read it, right?
Well, what he's trying to say is that this isn't necessarily new per se.
It's just X is making it very clear to the user.
Yeah, exactly. They're making it very clear to the user.
This is what we're doing.
This is what's happening.
You have until this day if you don't want it, don't take it.
But if you do, this is what's happening after that time.
Are you kind of insinuating that this is actually well, X is actually the good apple or the quote unquote honest player in this whole dingy place and and it's getting all the flack because actually amateurs can actually understand what's going on.
That was from Alex Fink, CEO and founder of OtherWeb, an AI based news reading platform.
Yeah. And can I just ride that coke tail of irony a little bit further?
So I don't know how to put this is this irony or contradiction.
But also X has made it really clear that it's OK for us to train to use this information to train our AI.
But any other company, if you want to scrape our stuff, no, no, no, you're going to be fine.
Yeah, yeah. Yeah, exactly.
How do you feel about that?
Well, I mean, the the terms and the conditions, they let you know what exactly is going on, but it doesn't mean that you have to feel good about what's going on.
And it almost sounds like a threat, doesn't it, when they use terminology like that.
But will we will we be able to opt out of this or not, you know, before these recent terms of service before this update, X users could opt out.
You know, you just go to your settings and then privacy and safety like you do on so many different devices.
And then under the data sharing and personalization header, there's a tab for Grok.
Do you know what Grok is?
A GROK chatbot. It's the AI chatbot.
So you go to that tab for Grok.
And that's where you can you can opt out.
It's not clear with these new terms of service, whether that option will be taken away or not.
And X can now license all the content on the platform, including using it for machine learning and models, as we mentioned.
Grok. Yeah, if you want to compare it to something else, it's like chat GPT, I guess.
It uses a large language model, that technology that trains it by feeding it a lot of information from all over the Internet.
Well, X has made it really clear with this what feels like 300 page disclaimer or terms of agreement that you guys have just mentioned that it prohibits any external scraping of its data.
So it's talking about, you know, the organization who does that is going to be charged $15 ,000 US dollars per one hundred sorry one million posts and scraping.
Should we talk about what that is just for a sec?
Yes, it's it's where it's a technique, I guess you would call it.
It's collecting content and data from the Internet.
Then it's saved in a local file and then it can be manipulated or analyzed as needed.
For example, here's an easy example.
If you've ever cost copied and pasted anything from a website onto a Microsoft Word document or an Excel spreadsheet, that is essentially what web scraping is on a very small scale.
But imagine bots are doing this, searching websites all over the world and taking what documents or what information they want.
That's what scraping is on a large scale.
Yeah, they're crawling all over the web for that information and then the regurgitate it according to our needs when we write our prompts in our large language models or AI assistance for the passage, the graphs, the drawings or the visual products that we want to enjoy, so to speak.
And X's recent move hearkens back to the dark past of controversies surrounding the scraping of Internet data without consent.
Scrape, and that's not new.
According to a different CNN report in 2023, The New York Times had sued OpenAI and Microsoft for copyright infringement alleging that those companies artificial intelligence technology illegally copied tens of millions of New York Times articles to train chat, GPT and other services to provide people
with instant access to information.
So the complaint is just the latest in a string of lawsuits that seek to limit the use of alleged scraping of a lot of different content from all over the Internet without any compensation.
So it sounds like X, this platform is trying to protect their user's information away from other platforms trying to use X, this platform, to maybe train their AI model.
But they are also saying that we can this we're protecting this information, but this mission information can be used by us ourselves and then to train our own AI model.
So who's surprised that it seems a company prioritizes its own interests over, let's just say, of those of those of their users, because it's pretty much put a price label or it's pretty much on a put a amount on the penalties for those others who are doing that.
We're doing that. Yeah, yeah, exactly.
This situation, a similar situation happened in China here as well, according to one report from TMT post this July, tomato novel.
Have you ever heard of tomato novel before?
It's a platform that's owned by ByteDance and there was a big public outcry here because they were asking novel creators for permission to use their work for AI training.
Very similar situation there, right?
So yeah, we see it worldwide now.
And for creators, it's very easy to understand why they're upset.
Why should you be able to use my content to your benefit with no compensation whatsoever from exes or another platform's point of view, I guess it's this.
Well, you're using our platform.
So in response to this backlash, tomato novel actually revised its protocol, allowing creators to opt out if they had concerns about this.
And that actually reminds me of that many AI software applications, they all are likely have kind of similar terms.
You know, imagine if you're using an image generation AI software, typically the images you upload can be used for training their own software.
So I think most users may not be very sensitive about their rights and privacy when they are using these kind of AI because you are inevitably going to upload something and the thing that you upload will be used by these companies.
I mean, in the beginning of the chat when I said, I personally don't really care because for the stuff that I post, it doesn't really make any difference to me.
This doesn't really impact me on a personal level.
But you are absolutely right, Huyang, when you said it doesn't have to be creators, it doesn't have to be artists only.
It can be anyone who's uploading things.
Because when when you post something, it's yours.
It's your thought. It's your idea.
It's your creativity.
So for other people to be able to use that to their advantage for where you don't get any benefit whatsoever, then yeah, you're right.
It's easy to see why so many people don't like it.
Maybe your behavior, you know, I think there could be even safety issues when you allow these kind of things to happen because many actually criminals already simulate someone's account, let's say the way that you talk.
And then sometimes just by altering their idea to appear as a friend of yours to commit fraud, because they're just talking to you, I say, oh, I need money right now.
But you think it is actually your friend and using the way that you're talking.
So with AI training, they could create even more realistic scenarios that might deceive people into believing them.
And that is maybe when we are not noticing it and they are actually using the way or your behavior.
And Grock, by the way, has already been in a little bit of controversy from spreading false information about the 2024 election.
That's one example, generating violent content, generating graphic fake images of famous politicians as well.
There's already been a lot of controversy surrounding the AI for X or Twitter.
Oh, yes. And also what you said earlier, Steve, about, you know, when you're a user on a platform and you're not paying anything for it and then, you know, something unexpected like this comes up.
It just reminds me of the saying in the tech space, if you're not paying for it, you are the product.
And here, I think, is a salient example of that, that our user data is seen as commodity and it's being used to train AI, which is another huge tool of these companies.
And let's just face it.
Almost every major tech company has one.
And this is an arms race among these tech companies and they're hungry for data.
And you might wonder how originally were these AI large language models trained?
And actually companies competing in this chatbot war have consistently turned to something known in the industry as the pile to train their large language models.
And this is very much in the English speaking world.
The pile is a massive trove of open source data.
And that's made up of several smaller data sets.
It includes text scraped from all around the Internet, Wikipedia, the European Parliament's website, even a collection of emails from employees at the now defunct energy company Enron.
Yeah. So I read this report by Annie Gilbertson, an investigative journalist of proof news who looked into how YouTube subtitles have been scraped as well.
But Google owns YouTube.
So I wouldn't be surprised if Google's doing something under the table without making a big fuss out of it, but it will definitely be unhappy if other companies do it.
Yeah. I mean, some companies are taking measures to block AI as well.
CNN, for example, they added code to their websites earlier in 2023 and it blocks open AI's web crawler, which is called GPT bot, from scanning their platform for content.
So there are there are measures that can be taken.
And one last question for this topic before we move on.
We well, users to this platform can opt out.
But what do you think, more importantly, needs to be done to sort of, I guess, shift this situation to make it a little bit better?
You know, I was thinking that when our content is publicly posted online, it is often already used in many AI training or AI databases.
But if platforms require users to just agree on these kind of terms unconditionally, users have the right to say no.
So at least before you actually put this policy in effect, you need to let users choose.
And also legal experts advise that AI generated content should be clearly labeled to avoid public confusion or misidentification.
They also pointed out that the copyright status of AI generated content depends on its originality, uniqueness and intellectual contribution.
Protecting the author's rights.
Platforms need to be doing that as well.
Yes, I can't agree more on that one.
Coming up next, there's this wild new thing happening in a New Zealand airport, hugs with a time limit.
Let's chat about these speedy farewells and the bittersweet vibes they bring.
Stay tuned. Looking for passion?
How about fiery debate?
Want to hear about current events in China from different perspectives?
Then tune in to Roundtable where East meets West and understanding is the goal.
You're listening to Roundtable with myself Hye Young.
I'm joined by Yushun and Steve Hatherly.
All right, picture this.
You're at the airport, heart in your throat, getting ready to say goodbye.
But wait, there's a new rule in the New Zealand city of Denedine that puts a timer on these hugs.
How do you make every second count when it feels like your heart's about to break?
Well, you've got three minutes to cram in all that love and feels before you part ways.
And what's the reasoning behind this hug time limit?
Yeah, right. New Zealand's South Island is where Dunedin's Airport is.
And it's a really quite small airport, a very modest terminal.
It serves about one hundred and thirty five thousand people.
And they have recently put into place the cuddle cap, as they are calling it.
It was imposed this September to quote, keep things moving smoothly in the drop off area outside the airport.
That's according to CEO Dan De Bono.
It's the airport's way of reminding people that the drop off zone outside was for quick farewells only.
So they have put a time limit on the hug that you can give to your loved one.
You have three minutes to stay in the area to get your hugs and kisses out of the way and then get out.
You can't you don't have to go home, but you can't stay here.
De Bono, his explanation is kind of funny.
He said a 20 cent he said, look, a 20 second hug is long enough to release those well being boosting hormones, oxytocin and serotonin.
Anything longer is, quote, really awkward.
I don't know if that's science or the CEO's personal opinion.
But before that, let's, you know, clear about what is the drop off area you you when you're getting out of your car.
That is the area of the drop.
It's not the parking lot.
It's the it's in front of a place when you pull up to the departure domestic or international departure area in an airport.
You'll see it here at the airports, too.
I guess cars temporarily stopped and people get out.
They take out their luggage.
They can't park there.
And actually in other airports around the world, they do things like wheel clamping or they'll find the drivers for parking there for staying there for too long.
So people will generally pull over.
They'll put on their four way flashers.
Right. They'll get out and then they'll stay there too long while they say goodbye.
This is happening outside the airport, not inside the airport.
It's all about pulling your car over, parking illegally and staying there for too long.
This is just a cute way that they're trying to get people to please don't park there anymore.
I think in this way, people mostly can understand that because that is an area that you're normally should not stay there for too long.
And I think by the sound of it, oh, you're not allowing hugs in the airport.
That sounds very harmful, I would say.
But if we're saying it is just in this kind of designated area that's not allowed, I think people understand.
People don't understand.
Actually, people are really angry about this.
They've called this policy inhumane.
They're accusing the airport of infringing on basic human rights by limiting emotional farewells.
Comments like hug police.
Is this just weird?
There's a lot of people who are very upset about this.
It seems like they do have their supporters as well.
They view it as kind of a tongue in cheek approach to getting people to move their cars.
And I think we do have to take this with a grain of salt.
There's no hug police at the airport.
It's just I think the CEO was trying to come up with a fun, cute little way to remind people you can't park here and stay here for a long time.
But some people are very upset about this.
Yeah. Well, it should be something I would assume people know that if you want to well sit down for a coffee or if you don't have that time, then you know, really get your hugs and kisses and then just park your car in the parking lot and write up the escalator and say goodbye to your loved ones properly
at the departure exit.
Well, do you think that this hug time limit would even matter in Chinese airports since hugging isn't really a traditional thing here?
Maybe that is why I said that people can understand that.
Then I think the impact might not be that significant.
There seems to be a cultural difference as Chinese people appear to be less open about expressing their emotions or especially intimate ones.
So I believe attitudes are slowly changing with time.
You know, for example, I now hug my parents.
You're a hugger? Fair wear gesture.
So yeah, times change.
Yeah. I'm a big, huge, ginormous hugger.
And I picked up this disease in the U .S.
And I love it because I think it's such a wonderful way to express yourself that you're opening up your chest to another human being, letting them in and close your arms.
And there are different kinds of hugs.
But if you're really pressing yourself against the other person, you're you're letting them in.
And I just and for those in your life that you think deserves that treatment, enjoy it, embrace it.
I think it really gets the oxytocin and serotonin running.
And that's a fantastic feeling.
And stay away from Dunedin Airport on the South Island in New Zealand.
All right. Park by God.
Yeah. Thank you so much, Steve and Yushun for joining the show.
I'm He Yang. We'll see you next time.
