Siri Founders give AI 2034 predictions

Should be running. Yeah. All right. So I learned in my life that you can speak to the world and have the first augmented reality app on the iPhone happen. I spoke on my blog. I told the Internet what I wanted to see the future be. And an Internet yell. Read that blog and made the first augmented reality app on iPhone. And tonight, we're gifted with three legendary people who pushed the world into artificial intelligence. And I wanted to hear them talk about what's AI in a decade gonna be. I think that was the question we missed this morning. And I'd like to hear the three founders of Siri talk about what they think the world after augmented reality cars starts really happening and after. I'm sorry, autonomous cars and augmented reality glasses happen, after those things happen and we are in this new world, what does that look like? So let me turn the camera on and Adam Shire, you want to stand up and tell us, I'll move my camera on you.

I think as we move forward in the future, some things will shock us. I mentioned today chatgpt melted my mind. Despite having literally having predicted it, watched it day by day evolve, it still shocked me and. And I thought it achieved things I never thought I would see in my lifetime. So that can happen. On the other hand, the more we get new things, the more it seems the same to me. Maybe that's just a question of getting older, being old. All of the phenomenons, all of the things, it feels so much the emergence of ChatGPT, just like what it felt when Siri came out. The same viral TV commercials, the same usage, going crazy and they can't handle it. All of these things happened. You know, the reporters saying, this is the greatest thing ever. This is the worst thing ever. You know, this is evil. This is. It's like all of these things happen with ChatGPT and with Siri. So my prediction of 10 years from now, in some ways I think will be just like Siri today is, you know, on the curve that was talked about, the hype curve today, just be part of our lives. We won't even really call it AI anymore, right? It'll be part of, like, when's the last time people said, oh, Siri is artificial intelligence? Like, nah, it's just Siri, you know, of course it understands my voice. Of course I can retrieve information and do tasks, but it's not. It's not AI. AI is magic. So first prediction chat GPT in 10 years will become that style, will become a fabric of our lives. We'll accept it. We'll use it, it'll be amazing. But it's not AI, right? The definition of AI just changed five months ago. Six months ago, AI was machine learning. Now AI means conversational systems, which this field is perfectly positioned in my head to capitalize on it. So next 10 years we're going to apply it, we're going to use it, we're going to maximize impact, we're going to get forward with it, we're going to accept it, and then something will come along in 10 years that's going to melt our mind and will be the new definition of AI. What is it? I don't know. I'll pick human brain interface, brain implants. I'll just pick that. And why do I say that? I was talking with my other co founder because I would never put a chip in my head. And I go, well, you realize there are millions of people who already do, who wirelessly electronics to neurons in open brain surgery, and it's called cochlear implants. There are deaf people who lose their hearing. We literally open up their head, wire in electrodes, two neurons, put a little microphone with software that controls the pulses going to those neurons. And because of the brilliance of our brain neuroplasticity, we adapt and we literally our brain rewires itself to these new signals and we can now hear again, right? I go, if you do that, meaning we're now hearing through software, right? There's no audio wave necessary to stimulate those that sound. I go, if that happens, you know, no reason I can't be sending and receiving thoughts that, you know, inter inter person. So I actually think it's possible that that will happen in my lifetime. And there are things that happen, that have happened that I never thought I would see in my lifetime. So that, that'll be my prediction. Do I want it?

Not sure.

Kind of creepy, you know, the idea of, of controlling. You can skip the glasses, you can skip the batteries, if you can just change your senses directly, you know, use your retinas, you don't need new glasses or even contact lenses. So I'll pick that as my next frontier. And that will go. That's truly artificial intelligence. Because you're taking your real intelligence, melding it with something artificial and hopefully this something is better.

Thank you. Dad. Dad.

Nothing like getting put on the spot.

Yeah, there is.

Pizza.

So I just did the calculations after lunch and it's right around 9:30am 2034 is when it's lights out for us. So just so you know.

I ran.

A few algorithms, I actually asked chatgpt like do the countdown and added a few parameters and that's what it came up with. So Enjoy the next 10 years and 3 months of your life because it's.

Going to be a ride.

I agree a lot with what Adam said. I'm not sticking any chips in my brain. I'll just talk to somebody. I don't need to do it through my brain. But yeah, I think, I think ChatGPT will become a tool that is sort of like Adam said, the fabric that's going to be. I think there's going to be many, many unicorns that develop these industry specific domains that are very, very good at this. They're going to add incredible voice, text to speech so it sounds human when you're speaking with it. Customer service is going to have a lot less employees in it. That's a big aspect of it. I'm in the healthcare business right now. I think that's going to be greatly impacted by through personalized medicine. We're already working on algorithms that detect, for example, when you use your mobile device to measure your blood pressure, there's something called the blood pulse waveform, which is what my new company does, is track changes in that you take the blood pressure cuffs and the stethoscopes and you toss them out. You're using your phone for everything and it's watching whether you're taking your meds. We're interacting through both chatbots and real people to answer questions about people's condition, what they should be doing, should I up my dose? Just day to day medical interaction with the medical system through artificial intelligence. And I think that's going to be a big trend. So personalized medicine I think is going to be a part of that. Companies like Alpha, like Google Alpha, DeepMind, if you guys are familiar with protein folding, have you ever heard of that before? I mean, when you understand what they're doing now with that, they're basically getting close to understanding the very basics of how the human body works. That's the cause of all disease. It's not just one disease. So I think medicine using AI on that, you've got also you may have heard about in imaging, if you have enough data, you can look at a million different pictures. AI is already better than doctors at predicting cancer because they're seeing the things that happen prior to the tumor forming. So I think just examples of how, I think in medicine in the areas we talked about earlier today, I also think that the vision of the personal assistant that we have all been dreaming about for many years becomes a lot closer to reality. I think there's a lot of work to do on the personalization, some of the other things we talked about, but hopefully we inspire people to actually work on this stuff. And yeah, I think we're. It's going to be a big part of our life. It's going to be the biggest AI, as Adam said. I mean, we've kind of known it for a while, but of course, you know, timing is everything. Computing power gets there, the algorithms get there, the new techniques like LLMs get there. But it becomes the fabric of everything that we do from medicine to personal assistants to everything else. So that's a little glimpse and especially on the medical side of what I see. But eat well, drink well, spend time with your family.

You only have 10 years left.

We're almost there.

We're almost there.

But from what I understand, it suddenly just. It'll be lights out, so it'll be painless.

I'm trying to figure out what the prediction is. I'm thinking sea level rise, floods, the data centers or something. Something happens in 30 years. I love talking about the future because it's something that you can't be wrong and you can't be right about it. Guaranteed you're wrong about it. But I can tell you about the near future and then a bit out there, what's really what. I'm an old fart, okay? I've been in AI literally since 1981. That was gradual level already. So I've seen a lot of it and has done a lot of this. But I've never seen it move as fast as it has in the last 18 months. It is exponential and I hate that word ridicule, people using that word inappropriately. It's actually a appropriately used now to talk about the advance that's happened in the last couple years.

And what.

What's interesting is almost. It's almost. There was a bunch of things leading to it. We've all heard about it, the GPUs and all that kind of thing. But they stumbled over a technique, a software technique, Transformers with Attention. That's the basis for all these large language models and when. And that thing is now being used from everything. Not just English language, but FMRI data. I mean, things that have nothing to do with language. Anything that has a sequential pattern to it or can be turned into a sequence, including images, are all our game for this architecture. It's just one invention, right? And there's a ton of hungry people working on inventions like that. It's the next Nobel Prize kind of sort of invention. So. And that invention, by the way, came out of translation. It came out of attempts to do their job at translation. And that is blowing everyone away. We're finding out that they may have stumbled on a basic principle of why we have evolved intelligence. We have a completely different architecture for intelligence than large language models. Don't let them tell you it's nerves and neurons.

It's not.

We are completely different. But we have something weird. We have a short term memory limitation. We have brilliant minds, a trillion connections, 80 billion neurons, and. And yet we have this seven plus or minus two things in our head at one time. What's that all about? It turns out that that thing, that restriction might even be the basis, that might be the attention mechanism in the transformer. That may be correlated with that. They may have come into the digital parallel computer version of that idea, which is why I started to get this general purpose stuff. So as a technology nerd, I'm saying there's something really there. Now the good news is obviously we can start aiming this thing at all kinds of interesting problems like we're starting to. And it's not just, you know, not just again, it's not just language and chat, boss. It's all these other things working on any problem that can be thought. The thought will improve the progress in most of the problems. We don't need thought like we have to global climate change. This isn't a thought problem, it's a will problem and a communication problem. But then let's like talk about what could also be happening with what we already have today. Not only did they discover a possibly amazing general purpose way of getting high intelligence, we also discovered of all the modalities that we've been playing with in AI, we've been playing with, you know, video and music and voice and language and so on, and just pure mathematical abstractions. Language was the magic thing. And it turns out that language, as Yuval Har says, is the operating system of human culture. It is in fact the universal UI for humans. And that means that this is not only an AI that's smart, it's a smart at dealing with humans. And that's what makes it this time different. Also makes the future really hard to predict. So in this sort of chaotic future, there's lots of branching. One, a good, a positive direction is that we actually learn how to figure out how to get this little buddy to be our little buddy, right? To actually augment us, serve us and the other dystopian futures, other, other People get it to play an adversarial role with respect to us, manipulate us, take advantage of us, convince us, do vote against our best interests. Things like this, they're both possible with this technology, right? There's one thing that hasn't been solved yet, which is what if we could use this technology to help us collectively make better decisions so that when we do vote and we do make meetups like Alan worked on, or, or these other kind of collective actions like when we form new sorts of entities. Corporations might be the last year's version of economic formulation. There might be new, new structures created that actually have AI in, in the middle of it, but are owned by humans and governed by humans. There are like all kinds of ways you can imagine creating things. Now when we do these things, is the AI making us work better as a species together in large numbers, or is it driving us towards our brainstem? So that's where the future bifurcates. Now what I wish would happen is like I said, I alluded to the day, education is a absolutely killer app. Education all levels, but there's just. You can rise everybody, every human being can benefit from this and learn better. Healthcare in many ways is bottlenecked by thinking, particularly in the grinding experimentation phase and other kinds of phases where you just have to do a million experiments and they can just have a worked out okay. But the real frontier is understanding human psychology. That's the real. Up until now, we have had pathetic data samples for human psychology. You get a PhD in psychology, you might have a thousand subjects, wait five years with them. Nothing. You do 100,000 subjects in five minutes in social media. If we have chat bots talking to 100 million people every day, we can learn so much about human nature. We can literally help figure out why we keep killing ourselves and causing poverty. All these things that we don't like about the way we live are caused because we don't collectively think well. Right? So that's my hope for the big 10 years from now. We have harnessed AI to help us make us collectively more intelligent.

Just so you guys know if it's true or he says that education is to filler out, I would recommend getting the associate two year degree since that gives you nine more years instead of second.

Thank you. Thank you everybody. Yeah, thanks.