Mo Gawdat warns that AI-generated deepfakes and recommendation algorithms are eroding truth itself, making it impossible to trust what you see or hear. He argues governments must mandate disclosure of AI content and that we need to embed ethics into AI systems now, before they learn the wrong values from their training data.
Governments need to ensure many things. I mean, you know, when I talk about some of the immediate threats, for example, one of the most important immediate threats is what I call the end of truth. Right. If you look at hashtag AI model and see how unbelievably undetectable, you know, it is to see if something is AI made or not, you know, and you can create all of those. Now, what would prevent us from in, you know, the U.S. election in a year's time or in a, you know, 18 months time or whatever, you know, what would prevent someone from going to one of the.
Endlessly developing graphics tools and say, give me a 25 minute documentary that stars Donald Trump and, you know, that basically talks about all of the atrocities that this other person did when they were teenagers in school with, you know, footage from what appears to be security cameras and, you know, testimonies from humans that don't exist. Right. And you, you could literally, if, if maybe by next year, that would take hours of processing, but in two to three, four years time, this will be just a tool available to everyone, like a face filter on Instagram. Okay. I'm actually working on an app myself that I call Pocket Mo, which basically all of my work about happiness and stress and so on, I gave, I fed that to the, to the language model. Okay. Created an avatar that looks like me. I'm trying to make it a little cartoonish so that it's distinguishable. Okay, well, you can go and ask Pocket Mo anything and, and simply say, you know, I. My husband is feeling a little depressed since the loss of his child. What advice do you have for me? And I've been asked that question before. So pocketmo will find it somewhere and say it to the user as if it's me. Right. I mean, I'm not promising this yet, but it's almost ready and if I don't release it next year, I'll release it the year after. It's. It is inevitable that we will get to that place. Right. So what would prevent someone of, from taking that app and making Pocket Mo say something that appears to be me, undistinguishable from me, indistinguishable.
So you're. When you say that, it makes me think like this. In three, four years time, I could be interviewing AI Mo instead of real.
Yeah, I was talking to a friend in the morning that there will be. I mean, the minute the app is ready, I'm gonna host Pocket Mo and Slow Mo on my podcast. I'm gonna have a chat. Yeah, I mean, One of my dear, dear, dear, dear friends, Peter Diamandis. Yeah. Has I. I think he calls it.
Yeah. Peterbot. Peterbot. So he has. He has a podcast episode of Peter the Mendez, you know, interviewing Peter Bot.
Amazing.
Yeah. And. And the. The opposite will be true. Maybe. Maybe Pocketmo will interview me.
I love that.
So we, we took. You know, you have. No. Harari basically summarized it in a very interesting way. Of course. Everyone looks at those things from their own perspective. To him, it's all about knowledge and about human history and about human ability to tell stories and debate and so on. And his description is that, is that AI has hacked the human operating system, right? If I can. It doesn't matter, by the way, if I am an AI or not. If what I told you right now entered your brain, I've influenced you for the rest of your life. Whether, by the way, what I'm saying is true or not, whether what I'm saying is by a human or by an AI, whether you agree with it or not. If I told you, by the way, I. I realized that New Zealand has the lowest number of redheads in the world, right? Doesn't matter if what I'm telling you is true or not. Doesn't matter. I've occupied the part of your brain that is either trying to believe or not believe, okay? That is trying to prove or this or. Or. Or disprove, right? That is constantly going to that. You know, if you listened to it, it will pop up in your head next time you see a redhead. I've influenced you forever, right? And. And that's the point. The point is that in the absence of a government intervention that literally criminalizes.
Disseminating content that's created by AI without saying that it's created by AI, we will completely lose touch of the truth, right? By. By not having anyone invest in tech that can actually find out, you know, what it is that is true or not. We're done. Okay? We're going. It is happening today when you swipe on social media that the Instagram recommendation engine or the TikTok recommendation engine is not only showing you videos to entertain you, they're shaping your life completely. Completely. You know what that means? What. What that means is that there are people out there in the world today that believe that one of the most important skills in life is to shake your hips, right? And, and, and how else. Why, how could they believe anything else if every video that they see, which has a million likes on it, is someone shaking their hips, right? And you start to reform your Perception of the world informed by a machine. And of course, as a result, more influencers stand in front of cameras and shake their hips and the cycle becomes even more vicious.
What kind of control and ownership do we have as individuals over the power of the machine?
The most beautiful question of all. Okay, so when I wrote Scary Smart, I have to admit to you, I wrote it out of what I believe was a message from my son, my departed son. I, I, you know, I was, somehow I felt that he was telling me, what are you doing wasting your time? We agreed we were going to do this right? And believe me or don't, it doesn't matter. But I was so driven to write that book. I wrote it in three months because I felt an obligation to write it. So when I started to write Scary Smart, the book is to, is to fold. The first half is the scary part, as I call it. Okay. And the second half is the smart part, as I call it. It's not very smart, honestly, but no, it is actually reasonably smart.
No, that's fantastic. I can say that.
Yeah, it is, it is, it is the answer to Marvin Minsky's question of how can we have, how can we make them have our best interest in mind? And so the first part of the book was scary, all that I spoke to you about so far and more. The, the second part was, let's admit for a second, what it is that we have control over, okay? We can't stop them, we can't dictate how they develop their intelligence. And they're autonomous, they can make their own decisions, basically, a little bit like your Lovely children, your 7 year old when he becomes a teenager, right? You have no control over what he will do, right? But if he loves you enough as a teenager, he will say, hey dad, let's go together and do this, right? If he loves you enough when he's 21 and, and you fall sick, he was going to say, hey dad, let me drive to you and take you to the hospital. Right? And, and this is not because he's intelligent. This is because of the ethics you instill within him. Right? And so if you, if you assume that artificial intelligence is, is autonomous, it is, it evolves, it learns on its own, it develops its own in, you know, intelligence. Accordingly, it will also develop its own ethics. And, and the definition of ethics is very interesting. The definition of ethics is this wishy washy, never really discussed concept that society at large believes is the right thing. Okay? And, and so with enough intelligence, we, we don't make decisions based on intelligence understand that we make decisions based on our ethics, as informed by our intelligence. So, you know, if. If I. If, you know, if you take a young lady and raise her in the Middle east, she will say the, you know, the. The trend, the. The way I should show my beauty is to be a little conservative. Right? If you take her and raise her in on the Copacabana beach in Rio de Janeiro, the answer will be a G string. Yeah, Right. Is one of them smarter than the other? No. It's the same woman. Right. Is one of them right and wrong? No, of course not. Okay. It's just that in, you know, the Brazilian culture, for a woman to show her beauty is worthy of praise. Right? In the. In the Middle Eastern culture, for a woman to appear conservative and share her beauty with her loved one only is. Is what deserves praise.
Neither is right or wrong. But this is how ethics are formed. Now, if we can show artificial intelligence, what ethics they should follow because they're learning from us. Chad, GPT did not develop its knowledge by contacting an alien civilization somewhere and saying, you know, how do I solve physics? No, it read all of the physics work, all of the literature, all of the chats and conversations that humans generated, right? So it's learning from us. If we can show it the right ethical code, it will learn the ethical code. It will grow up, as Marvin Minsky said, to have our best interest in mind.
