Transcript

Proving Identity in The Digital World | Max Penk | TEDxMitte

URL: https://www.youtube.com/watch?v=CukfGzqItGQ
Video ID: CukfGzqItGQ
============================================================

Who in here is a robot? Anyone lift your robotic arms? No one. Exactly. They never admit it. This is how they win. No, seriously. I mean, in the real world, it's relatively easy to see who's a robot and who's not. You can look around. You can touch your neighbor, maybe. Yeah. If they give you consent, you can see them. They behave. They breathe. They laugh. They smell. But if I would have asked the same question in a digital space, it wouldn't be so easy. In fact, statistically speaking, half of you just lied because 2025 is the year robots outnumber humans online. We have 51% robots and only 49% humans in the internet. Half of the internet is no longer human. Now, not all robots are bad. Most of them are incredibly helpful. They help you to do your research. They help you to recommend flights. They filter your spam. They recommend films for you. And they're also becoming more intelligent. They're becoming more embedded, more like us. They're even commenting now on some Instagram posts, LinkedIn posts, etc. And they're making memes. They have their own accounts. I mean, they're not only like single users in some Facebook groups. They really have their own accounts. They're part of our online community. if you notice or not. And they're getting so good at acting like us that the main question is how do we know who's real in the digital world? And will it even become more difficult in the future? Now, let me show you something you all know. It's part of a little comic and it's a lifelong question. And they ask Buddha, what makes us human? And Buddha says, "Secting all images with traffic lights." I mean, if you selected traffic lights this week and clicked on a little checkbox saying, "I'm not a robot." Congratulations. You were at the last human fronture of the internet. Digital identity. This is called a capture. So, I don't know if you heard this term before, but it stands for completely automated public touring test to tell humans and computers apart. So there's the word touring inside and touring is Alan Turing. He was a famous mathematician and computer scientist and he said this in 1950. A computer would deserve to be called intelligent if it could deceive a human into believing it was a human. 1950. Now we are here 75 years later. The touring test has long been solved. >> [snorts] >> In fact, a recent study showed that chat GPT convinced 73% of the people chatting to it that it was human. And also, it convinced even more humans that it [snorts] was human than a human chatting with a human. That's totally crazy, right? The whole thing, the whole discussion gets now reversed because you know what really gets me? Why do I have to prove to a robot that I'm not a robot? I mean, who's testing who here? Yeah, that's confusing. So, and sometimes I'm sitting there and I'm wondering, wow, there's this super intelligent machine on the other side of the screen and it goes like, hm, I don't know if you're a robot. Please select all the traffic lights. And then you get this and you see it and you think, wow, is this little piece in this square, is it still a traffic light? What happens if I click on it? It's wrong. Does the machine think I'm a robot? Yeah, the overthinking begins. Yeah, you panic. No, it's it's really funny. I mean, we are building rockets which can fly to Mars. We're mapping the genome. We're building artificial intelligence more intelligent than us. But yet, we still have to select traffic lights to say we are not a robot. It's I think it's not inconvenient, it's ridiculous. It's the year 2025. But how did we get there? So, first of all, robots started to influence our language. So originally we came up with a concept of a hashtag. I don't know this is an awkward sign people who are doing it. I don't know if you remember every one of us used the hashtag once in a while. We posted something on Instagram # sunset # yolo #txmitter. Yeah. Um so we marked our words for the robots so they could see okay there is a word there's a message the user wants to get to me. So we started to work for them. But now it's reversed actually because we are now talking in terms the robot would like to hear and the algorithm would like to hear and this is called algo speak. So let me give you an example because this is a language we use to either avoid moderation or to boost engagement. So if you write online someone killed himself this will get flagged and will be hurtful content. So not a lot of lot of feats will play this out rightfully so. But if you write someone unal alived himself, this will work. Yeah. So this is our way to get around moderation. And we also use it to boost engagement. Maybe you heard the word rz. It's a short form of charisma or dulu for delusional. Yeah. And this is not only internet slang. This is something which gets into our offline language. people start to use it in their daily lives. Yes. So this is algo speak. So the straight line the straight line between sender and receiver is not a straight line anymore because now we have algorithm as a third point in this picture. So essentially we're not just tagging bots anymore. We're performing for them. So really for the bots. So the second reason is um bots participate in online conversations and they get really good at it. They are really good. They're winning arguments against us. So there was earlier this year an experiment from the University of Zurich. They released AI bots into Reddit discussions and they were not telling anyone about it and they were strategically responding into one of the most reflective subreddit called change my view. It's an open forum where you can post an opinion and you are ready to be convinced by another opinion. Really like a really cool public civil debate you can do. And the bots acted as real people and replied to over 1,700 people. And their goal was to see how convincing those AI arguments could be. And they had to end this experiment earlier than expected because the results were mind-blowing. The bots were up to six times more convincing than real humans. Yes, bots can be 600% more convincing than real users. And to top this, nobody noticed that they were bots. So this was not only a touring test to see if the bots could go unnoticed, it was a test of trust and we failed. We failed hard. So in the third reason, the platforms actually like having bots around because think about what most social platforms are built on. Personal confirmation, recognition. You want to be seen. You want feedback. That tiny dopamine hit when someone reacts. Now imagine this. You post something and you get zero likes, zero com comments, no response. You probably wouldn't post again, right? But if you post something and within minutes you get 15 likes, three comments, someone saying, "Wow, I love the colors on this sunset or this makes me happy just looking at it." You feel encouraged and you will probably post again, right? You will probably post more than ever before. And this is the thing. So maybe those likes and those comments are [snorts] just bots designed to boost your engagement because more engagement on the platform means more time spent on the platform. More time spent on the platform means more opportunities to play out advertising for you. And more advertising means more money. So in the end those bots are not only users, they're somehow the rocket fuel for some platform's economy. And to add to this, it's really hard for advertisers and brands to see technically a difference between a real user seeing an advertising and the bot seeing the advertisement. But these platforms count for every few. So technically half of the money spent by advertising is in front of bots. Half of the money is gone. And a lot of people call this ad fraud because the money is lost. But here's a different view on it. What if we are advertising for humans and for bots? Because bots are also making buy decisions. Bots are also recommending stuff. But the difficulty is if you want to advertise for humans, you have to be emotional. You have to trigger a feeling. You need to be touching somehow. But if you want to advertise for bots, you have to be very structured, very datadriven, very formal. If you want your advertising, your brand message to be played out by Chat GBT, Gemini or Perplexity, you really have to talk to bots. You cannot avoid it anymore. Yeah. But good luck in combining one advertising with which is touching, emotional, but also informal, wellruct structured, datadriven. It's not possible. And this is becoming a huge problem as well. But now let's zoom out. What happens when these systems not influence only attention but power? I think the experiment from the University of Zurich showed that it's very easy to change a public sentiment, a public opinion about something and we should be very aware of this happening because it all comes down to one thing which is human digital identity. Because if bots can fake engagement, if they can fake outrage, they can fake citizens, they can also fake democracy. And this isn't sci-fi. This is now. This is happening. So where do we head from here? Do we try to divide the web? One place for bots and one place for humans, or do we try to merge everything together like we're already doing it? Do we need new rituals, new signals, some kind of systeming, a system where we can show here, this is me. I'm real. Yeah, these questions don't have simple answers to it. But they're the ones we have to ask because if we don't choose a direction, the platforms will. And they don't care about the human space. They care about outcomes. They care about numbers. But what happens if we don't act? Well, then your whole feed might just be full of bots with artificial content. I mean, open any social platform. What do you see? Probably not your friends anymore, right? So, social media, this big promise of human connection has already quietly turned into an algorithmic wasteland. It's sometimes like a ghost town full of perfectly optimized content. But the engagement rates are collapsing. Real users are starting to leave the platform. The numbers show that. And the ones who stay, they're just fighting for attention all the time. They're fighting against bots. So influencers are trying to automate the authenticity. They are brands are optimizing more for algorithms than actual for people. And your feed gets an endless stream of infinite content, but with zero soul. And here's the bad part. Most people won't even notice. But they won't notice that this is bot content. They just keep scrolling. And one day you will refresh your feed and it will be just bots talking to bots. This is called the dead internet theory. And this is not a theory anymore. So the scariest part isn't bots taking over. It's more us not noticing when they do. So what can we do? So this franch this last human franch can't be solved with better captures or more traffic lights to click on. We need a new framework for digital personhood. A way to signal not just that we're human but how we are human. So I give you another example. My son loves to play Roblox. So Roblox is a famous online gaming platform and also there bots started to invade the space. Yeah. And he loves to play role-play games. And in roleplay games, you want to play against humans. So what was their way of proving human identity? They made voice chat with a microphone in most of the games mandatory. So you really have to be reachable. You have to talk. People can come to you. You have to response. This is their way of proving at least a little bit that you're human. And for them, this worked. Yeah. A lot of people like were convinced here are no bots and I think it still works but it's just a matter of time till bots get intelligent enough and the human voice systems get copied so exactly that you cannot see a difference anymore. But you see they're fighting in every area of the internet against this problem. So some ideas of how to prove human identity are emerging and some of them have been around for a while. I show you some. So we have decentralized ids. Decentralized ids are basically a unique identifier which follows you everywhere through the internet every website you visit. So it helps everyone to see it was once verified as a human. Also biometric authentication there's a project called world from openai or tools for humanity where you scan your iris and you become a unique human in the internet also has has its problems. There are reputation graphs. There are trust networks and also connected smart devices like smart watches, smart earpods, smart wristbands, devices so close to the human that no bot could be between it. But all these devices have a problem because you need sometimes money to buy something. You have to get a membership and that should not be the case if we want to integrate everyone. Right? So the core idea is this. In this age of intelligent bots, authenticity becomes our most valuable currency. Not productivity, not virality, authenticity. Because what if the future of the internet isn't about machines pretending to be human, but about humans proving they're still human? Maybe the next touring test is not for machines, but is for humans, and it will not be a technical one. Because you know what worries me the most is not that bots will take over the internet. It's that we slowly stop acting like humans. Slowly stop talking like humans. We adapting bots language. We are arguing like bots using chat GPT in discussions to convince someone. This is when we are really lost. And I think the franch isn't out there. It's in here. It's in the way we show up online as unpredictable, as emotional, as sometimes beautifully irrational beings. Authenticity becomes our most valuable currency. And the next time you post something, you want to uh comment something, just ask yourself this tiny little question. Is this me or is this what the machine wants me to be? Because that's the new human test, right? That's the line we have to draw. not with captures, traffic lights, but with our behavior. If we get this right, we won't just prove we're human. We'll keep the internet human. And that, I believe, is a funure worth protecting. Keep the internet alive for and with humans because humans want to connect with humans. Thank you.