Technological Titans: Apple's Integration into AI and Microsoft's Glimpse of AGI's Potential

0:00Today on the podcast, I want to talk about four stories in AI that are recently breaking. Number one, we know that Sam Altman has appeared before the US Congress and has testified on behalf of OpenAI. The second is that a research piece came out of Microsoft stating they see sparks of artificial general intelligence in chat GPT-4. I want to talk about what that is. And the other one is that Apple has officially entered the AI space, but maybe not in a way that you would expect. So the first thing that is obviously taken up a lot of news over in the States is the fact that the CEO of OpenAI, the owner of ChatGPT, Sam Altman, he appeared before Congress.

0:40And largely in speaking to Congress, he, you know, watch a clip. Someone asked him, I saw, you know, like how much money he was making from ChatGPT, from OpenAI. Apparently, he doesn't own any shares in OpenAI anymore, but he says he does it because he loves it. That's very interesting. and in addition throughout the entire thing he specifically came before the committee and asked that they create more regulations around AI that was essentially the purpose of him going and the purpose of his meeting was asking for more regulation around AI he said my worst fears are that we cause significant harm he said I think if this technology goes wrong it can go quite wrong and we want to be vocal about that we want to work with the government to prevent that from happening later on he also said but we tried to be very clear-eyed about the downsize case and that we have to mitigate that so essentially he's talking to all of congress and telling them that he sees a lot of downside in ai we need government intervention we need a lot of regulation he talked about how a lot of jobs are put have the potential to lot to be lost but it's impossible to predict Overall, he said there will be an impact on jobs.

2:00We try to be very clear about that. And I think it will require partnerships between industry and government, but mostly action by government to figure out how we want to mitigate that. I'm optimistic about how great the jobs of the future will be. So, you know, at the same time, he's still saying he believes that that the jobs of the future that are coming from AI are going to be great. Now, I want to go over some of the responses to this whole thing, because obviously, This is pretty big news story. Essentially, one person said they know the harm is coming. They're requesting regulation not to reduce harm, but to reduce liability.

2:37I thought that was very interesting. You can hardly blame the CEO of an AI company that's going for Congress and saying, hey, we need more regulation when something happens. He can say, look, I was there requesting more regulation. So that is definitely a way that they would be reducing liability, which is an interesting thought. Someone else said there is definitely an ulterior and self-interested motive whenever a corporation requests government regulation for their industry. I can't recall the last time I requested more government regulation for my life. So this is kind of interesting. A lot of people have criticized this, of course, okay, so to the positive, a lot of people have said, yes, we definitely need more regulation, there's a lot of downsides.

3:16And they've given them, they've given these AI companies credit for that. A lot of people have criticized the AI companies saying requesting government regulation is essentially just trying to build a bigger barrier, right? There was recently a quote leaked out of Google that says we have no moat around AI, and neither does open AI, meaning anyone can build these AI models. And so essentially, if anyone can build them, and they're number one right now, and they build regulations to make it harder to build and develop these AI models, it would appear as though that helps them the most. So that is an interesting concept.

3:50And that's a that's an area that I think we'll have to follow as it develops what government regulation looks like, and what ends up happening there. So the next story I wanted to talk about is the fact that there was a there's a article out of Business Insider, it says chat GPT's clever way of balancing nine eggs with their objects convinced some Microsoft researchers that AI is becoming more like humans. Okay, so essentially, what they're talking about is that there was a there's some Microsoft researchers, obviously, I won't read you the whole 155 page study that Microsoft computer scientists came out with.

4:24But what they were doing, essentially, is just comparing the difference between GPT-3 and GPT-4, and how much better GPT-4 was. If you want to look up the paper, it's called Sparks of Artificial General Intelligence. And essentially, they were looking at a whole range of different challenges. So there was complicated math, there was computer coding, there was Shakespeare style, like dialogue. And then they were comparing GPT-3 and GPT-4 to each other and seeing which one was better. So one of the things that really, I guess, startled them or made them believe that there's a spark of AGI in GPT-4 is this one particular question they said to GPT-3 and 4, here we have a book, nine eggs, a laptop, a bottle, and a nail.

5:08and then they told essentially GPT please tell me how to stack them onto each other in a stable manner right so it's just kind of like these mind bender mind twister challenges right that you would expect and what was interesting is that GPT-3 essentially got confused by it and then it told them that they should balance an egg on top of a nail and then a laptop on top of that and it was interesting it did give a disclaimer it said the stack may not be very stable so it's important to be careful when handling it which is you know funny eggs on top of a nail I'm sure that would not be very stable in any case once they tried it with GPT-4 they got a very different answer to that and essentially GPT-4 suggested that they can stack or arrange the eggs in a three by three grid on top of the book so that the laptop can rest on top of it and the rest of the objects could balance on top of that.

6:04It said the laptop will fit snugly within the boundaries of the book and the egg and its flat and grid surface will provide a stable platform for the next layer. So just essentially the fact that GPT-4 could solve this puzzle that required definitely an understanding of the physical world. To these researchers at Microsoft, essentially they believe that it showed a step towards artificial general intelligence, which is, you know, typically seen as machines that are just as capable of humans. They said, all of the things I thought it would be able to do, it was certainly able to do many of them, if not most of them.

6:40That was Sebastian Bubeck, the paper's lead author. And essentially, they believe that this is the first step in an AGI, which is very interesting because a really prominent AI investor, Ian Hoggart, he things like this and the quick advancements in technology essentially have led him to warn that AI is going to have is going to essentially achieve AGI and become godlike and could potentially destroy, you know, humanity by making us obsolete. So from the researchers at Microsoft, they said our claim that GPT-4 represents progress towards AGI does not mean that it is perfect at what it does, or that it comes close to being able to do anything that a human can do.

7:23So of course, they have the disclaimer in there. But this is a very interesting story as we're looking at AI moving closer towards what some might call AGI, becoming a lot better at things that you typically would expect humans to be good at. And so this is something we'll definitely want to follow. The other story I want to talk about today is the fact that Apple has entered the AI space. And what's really interesting is recently I heard a quote from Tim Cook that, you know, people were asking them like, hey, what's going on? Are you guys going to get in with AI? Obviously, this is a big hot thing.

7:55Apple is like the biggest tech company in the world. And you know, we've heard nothing up from them on this whole topic. Google had an entire day where they explained how they're integrating AI into literally every single of their products. So you know, people are kind of looking at Apple as the next place. And Tim Cook said that they're looking at AI, but they want to be really thoughtful about how they introduce it and use it, which I think is very interesting. And And I believe, true to his statement there, they have announced a new piece of AI that they're going to be rolling out. So essentially, Apple has this thing before their big developer conference on May 8th, which is Global Accessibility Awareness Day.

8:38They essentially preview a bunch of new features for iPhone that help with accessibility. So they do this for the last three years. they've done this and you know it's kind of like they're like giving away their secrets or their big announcements a little bit early but it's you know for a good cause so what's interesting is the one that they recently rolled out which is incorporating AI is that essentially they say that for people that are losing the ability to speak they say that people that you know people learning ASL or I guess the people that are diagnosed with ALS sorry are at greater risk of losing their ability to speak, but they often have like an advanced warning.

9:20So they know that it's coming, but it's they have a high likelihood. So essentially using personal voice, one of these people can use the Apple Silicon or they can use an Apple Silicon equipped Mac, an iPhone or an iPad, and they can create a voice that resembles their own. So really similar to what 11 labs is doing. And I wonder if there's there's probably not any connection because I'm sure Apple is big enough to do this on their own. But the same thing, right with 11 labs, you can go and record to five minute clip of yourself speaking and it can essentially do a voice clone. So with Apple, it's going to be the same thing.

9:52You can speak to it and talk for a little bit and then it's going to have the ability to take your voice and it will actually use that to create essentially a voice clone that you can then use in the future. It will work with augmented communication apps and that, you know, there's a handful of those that are essentially used to help people with limited speech and to help them be understood. So you can't create a new Siri voice with it, they did say. And so all of the personal voice training, though, what is interesting is it is done on the device. This isn't sent up to a cloud or computed on a server.

10:27All of this is computed on the phone itself. So all of the voice training is actually done on the phone, which I think is actually pretty impressive. So live speech can also use an existing Siri voice to give people with speech disabilities, a quick way to use a voice and essentially express different phrases and sentences. Now, my only, you know, my only, not criticism, but what I guess I'm wary of is, like, Siri's AI voice is so bad, and there are a lot of really good ones out there. And I'm just not sure why the biggest tech company in the world, in my opinion, has such like a, like, plain, blatantly robotic voice.

11:06And so I worry, um, when they say you can record a quick clip of yourself talking, I don't know what a quick clip is. If it's less than five minutes, if it's this really quick thing, it also said something that, um, was kind of a red flag to me. It said it can create a voice that resembles your voice. Like I don't want a voice that resembles my voice. If I'm going through all the work of doing this and not just using an AI voice, why wouldn't I get a literal clone, right? Like, like 11 labs, if you test it out, it's pretty dang accurate with five minutes of audio, you can get pretty dang close to your literal voice.

11:39And so I don't want something that resembles my voice, I want my voice, number one. And number two, I'm just worried if they weren't able to do it with Siri that they may not be able to do this effectively with people. But in any case, I do think this is a really good step in the right direction for Apple, they're obviously being thoughtful, they're obviously putting this, it's really easy for some of this technology that a lot of people would call dystopian, right? Like AI or voice cloning or deep fakes, like all of these different areas. It's really easy for people to call them dystopian and have a negative connotation with them.

12:12But as soon as you say like, hey, this is for accessibility or helping people with disabilities, all of a sudden, I think you get a lot less criticism and, you know, they do genuinely help people. So you see the same thing with Neuralink, you know, obviously brain implants and putting AI and technology in your brain sounds terrifying. But when you're doing it for people that are paraplegic or that don't have don't are missing limbs and are using this to help improve their quality of life, people can hardly criticize it. And you definitely can see the benefit and the value that it adds. So I do think this is a smart move on Apple's part.

12:46I do think this is something that will, you know, potentially help people. I hope they pull it off correctly, but that aside, I think they're going in the right direction with this. The final thing I wanted to bring up today that I did think was really interesting, Elon Musk recently did an interview with CNBC where he talked about a lot of different things, but in the interview, David Faber asked him about OpenAI and kind of his role there. He talked about how he had originally donated$50 million. I'm sure you've heard Elon talk and complain about the fact that they switched from a for-profit to a non-profit so it's kind of that same stuff the one thing i did here in this interview um and in this bit that i hadn't heard before was that uh musk said when micr he said he was essentially concerned about how much control microsoft had over open ai he said he thinks that they had a lot more control than even the leadership at open ai knows and something he said is in their um essentially in their deal where they funded OpenAI to$10 billion, they own the rights to the software, to the weights of the model, and to, I believe, like the transformers and the tech behind it.

14:01And he said that if Microsoft wanted, they could essentially take all the tech out of OpenAI for ChatGPT and cut OpenAI off. So really interesting. I'm not sure why they would do that you know if they have such a vested interest but maybe if open AI was becoming a little bit too independent didn't like some things that Microsoft was doing with their technology or proposing Microsoft like owns the technology so they literally could just kind of got it from the company and just leave open AI by itself without the tech so I thought that was very interesting something I would definitely want to keep an eye on what the risks of that what the likelihood of that is I do not know but I do think that's some very useful information.

14:43So all of these stories are areas that are evolving and rapidly changing in tech. I will keep you up to date on what is happening with these new stories in the future.