Did You Hear About This AI News Yet?

Last week, Google owned the news with their Google I O event. And this week ChatGPT and OpenAI took the news back. Google held the spotlight for most of the week last week, but by the end of the week on Friday, ChatGPT just had to go and make the announcement that everybody was waiting for. They announced that they're rolling out web browsing and plugins to all ChatGPT+ users over the next week. And if you're a ChatGPT+ user and you've logged in anytime this week, you've probably noticed that these little buttons at the top look a little bit different. If I hover over the GPT4 button, I now have a dropdown with browsing and plugins. If you don't see this dropdown with browsing and plugins come down to the bottom left, click on your name, click on Settings, click on Beta Features, and make sure web browsing and plugins are turned on. If you turn on the browsing function, ChatGPT can now search the web, but honestly it's super slow and we often get a lot of failures as we're trying to search. Like this reading content failed that we're seeing now. Now, it did eventually give me an output, and it did cite all of its sources for where it found these outputs, but it did take quite some time and this little progress bar showed multiple failures to and multiple failures to read while it was trying to search this out. In my opinion, the bing chat with GPT4 built in is actually still a little bit better for web browsing than what we're getting out of ChatGPT. But again, this is in beta, so it's still very early when it comes to plugins. You could click this dropdown, click on plugins, click this other dropdown where it says no plugins enabled, click on the plugins store, and you'll see a ton of different plugins that you can play with for ChatGPT. Out of these, the Wolfram one's probably the coolest because it can do some really complex calculations and pull up real interesting real time data. So for example, with this Wolfram plugin set up, I could type things like how many of Shaq's shoes would it take to span between the Earth and the moon? It used Wolfram to figure out the average distance between the Earth and the moon at 385,000 kilometers. It then figured out that Shaq's shoe size is a 22. It actually did some of this math wrong here where it said a 22 shoe size is equal to about 7.33 inches. Pretty sure Shaq has bigger than 7.33 inch feet. So I'm gonna go ahead and tell it. A 22 shoe size is equal to about 22 inches. And according to Wolfram, it would take about 689 million of Shaq shoes to span the distance of the Earth to the moon. That's some very relevant and valuable information that everybody should know. Now, the real highlight of the week was the Senate hearing about AI regulation that featured Sam Altman, Gary Marcus and Christina Montgomery from IBM. Now, this Senate hearing was over three hours long. I watched the entire thing and I condensed it down into about 27 minutes of the most interesting part. So if you want to see the TLD watch, you can head over to my Twitter and see the 27 minute condensed version. If you're watching on a desktop, you can even speed it up to 1 1/2 to 2x speed and speedrun the entire Senate hearing in about 15 minutes or so. In this hearing, basically the creators of all this AI stuff were essentially begging the government to regulate them.

Regulatory intervention by governments will be critical to mitigate the risks of increasingly powerful models. IBM urges Congress to adopt a precision regulation approach to AI.

We obviously need government involved.

A lot of the discussion focused on how AI is going to impact jobs, technological revolutions.

I expect there to be significant impact on jobs, but exactly what that impact looks like is very difficult to predict. GPT4 and other systems like it are good at doing tasks, not jobs. And so you see already people that are using GPT4 to do their job much more efficiently by helping them with tasks. Now, GPT4 will entirely automate away some jobs and it will create new ones that we believe will be much better as our quality of life raises and as machines and tools that we create can help us live better lives. The bar raises for what we do and our human ability and what we spend our time going after goes after more ambitious, more satisfying projects.

A lot of time was also spent during this hearing around how AI has the potential to manipulate elections and what can be done to prevent that from happening.

We're going to face an election next year and these models are getting better. I'm nervous about it. I think people are able to adapt quite quickly. When Photoshop came onto the scene a long time ago, you know, for a while people were really fooled by Photoshopped images and then pretty quickly developed an understanding that images might be photoshopped. This will be like that, but on steroids.

Another big topic of conversation was what happens when all of this AI power is concentrated into a few small companies like Google, OpenAI, Microsoft, et cetera. I've heard a lot of chatter, especially over on Twitter about how OpenAI is trying to force the open source out of the market and they want to control it all. But after watching this whole hearing I get the very strong impression that Sam Altman in OpenAI want to encourage open source and small builders to continue to.

Build the open source community is amazing, but there will be a relatively small number of providers that can make models at the at the true. Is there danger in that? I think there is benefits and danger to that. Like as we were talking about all of the dangers with AI, the fewer of us that you really have to keep a careful eye on, on the absolute like bleeding edge of capabilities, there's benefits there. The regulatory pressure should be on us, it should be on Google, it should be on the other set of people in the lead the most. We don't want to slow down smaller startups, we don't want to slow down open source efforts. We still need them to comply with things. They can still, you can still cause great harm with a smaller model, but leaving the room and the space for new ideas and new companies and independent researchers to do their work and not putting a regulatory burden that say a company like us could handle, but a smaller one couldn't.

And another really interesting moment during this testimony. Gary Marcus, somebody who's typically kind of an AI doomer, gave a lot of credit to Sam Altman himself.

Let me just add for the record that I'm sitting next to Sam closer than I've ever sat to him except once before in my life. And that his sincerity in talking about those fears is very apparent physically in a way that just doesn't communicate on the television screen.

Thank you.

From here we actually learned that Sam Altman is only taking enough money to cover his own health care and that he actually doesn't even own equity in OpenAI.

Make a lot of money, do you? I paid enough for health insurance. I have no equity in OpenAI. I'm doing this because I love it.

So I've been keeping a little list here of potential downsides or harms, risks, generative AI, even in its current form. Let's just run through it. Loss of jobs. And this isn't speculative. I think your company, Ms. Montgomery has announced that it's, it's potentially laying off 7,800 people third of your non consumer facing workforce because of AI. So loss of jobs, invasion of privacy Personal privacy on a scale we've never before seen, manipulation of personal behavior, manipulation of personal opinions, and potentially the degradation of free elections in America. Did I miss anything?

Yeah, there was a lot of fascinating questions and a lot of really interesting answers. I was very, very impressed with how Sam Altman specifically handled himself. He was very articulate, he shared his thoughts very well and I have a so much stronger respect for him and for OpenAI. After watching this entire hearing now again up on Twitter, I narrowed it down into 27 minutes. I didn't want this entire video to just be a recap of this hearing. So I highly recommend that you go and watch this, see all of the different topics that they covered. I've watched a lot of these past hearings with Facebook and Mark Zuckerberg and the CEO of TikTok. And the senators always seem to make really off the wall comments that just show how clueless they really are. In this one, I was fairly impressed with the Senators questions as well as how much they were completely upfront about how they don't understand the technology and that they need to lean on the people that are inside building this technology to understand how to best regulate it. Now the day of that Senate hearing, there was an interview with Elon Musk on CNBC where he said this.

I AM the reason OpenAI exists. It wouldn't exist without you.

It wouldn't exist without you.

I came up with a name. The name OpenAI refers to open source.

Elon also talked about how OpenAI is essentially just kind of a puppet for Microsoft right now and how Microsoft is really pulling some of the strings exactly.

Is the relationship between OpenAI and Microsoft. And I do worry that Microsoft actually may be more in control than say the leadership team at OpenAI realizes. They have control rights to all of the software, all of the model weights and everything necessary to run the inference system. At any point Microsoft could cut off OpenAI.

Satya Nadella from Microsoft responded in another interview with CNBC.

Actually not correct. I mean, as I said, OpenAI is very grounded in their mission of being controlled by a nonprofit board. We have a non controlling interest in it. We have a great commercial partnership in it. And quite honestly I am very comfortable in partnering with a cap profit company that has a mission of fundamentally pursuing this very powerful technology that ultimately is going to be controlled by nonprofit. In fact, the last time I checked, we are the only for profit company that is comfortable with a nonprofit company and a board controlling technology and I would welcome others to do that as well.

Now this actually happened later in the week. But since we're on the topic of OpenAI and ChatGPT, OpenAI just released their official ChatGPT app on the iPhone. As of right now, it appears that it only works in the US and it's only on iPhone. But the PS of their announcement here says, Android users, you're next. Now, there have been a lot of fake ChatGPT apps that went on the App Store and the Android Store claiming to be the ChatGPT app and taking people's information and getting credit card details and charging people and not actually giving them chatgpt. So be very, very careful when you're looking for the ChatGPT app in the App Store. Make sure that it is the one produced by OpenAI. And your best bet is to actually go to the OpenAI blog and click on the button to download on the App Store just to make sure that you're getting the authentic OpenAI ChatGPT app. Because there are a lot of fake apps out there. Now, noticeably, when you do use the ChatGPT app, it doesn't yet have the new browsing features and the new plugins built into it yet. However, my friend Matvet Pro found a little bit of a workaround. Oh, okay.

This is really interesting. So you actually can use the plug plugins and web browsing. You just can't start a new chat unless you start it on the computer first. But once you have access to it, you can see Wolfram is still in here.

So essentially, if you start using one of the plugins from the ChatGPT inside of your browser, and then you continue that same conversation from your phone, that plugin or the browser feature still continues to work. It just doesn't seem that you can start with that plugin enabled from the Chat GPT on your phone. I have a feeling that'll be something that's updated fairly soon, though. I'll link to Matt's full video below in the description so you can watch his demonstration of the new app itself. All right, now, Moving off of OpenAI news, Google did make another announcement this week. They announced that they're adding AI powered coding for free inside of Google Collab. Now, if you've watched some of my past videos, I actually show you how to use Dream Booth inside of Google Collab. It's basically a platform where people can write code, allow others to access that code, and then run that code inside of Google's clouds. Well, they are now adding generative coding directly into the platform. You can see an example here where somebody asks a Question. And it actually gives coding help right inside of Google Collab. And they say access to these features will roll out gradually in the coming months, starting with the paid subscribers. But they do point out anyone with an Internet connection can access Colab and use it free of charge. So it sounds like it's going to be available for free users eventually. They're just rolling it out to the paid users first. Now, if you've ever generated images with Stable diffusion and you've used Stability AI's Dream Studio, and you really like this platform for generating images, maybe you prefer it over something like automatic 1111. They actually just announced that this Dream Studio platform is being open sourced. So you can actually download the code for this and install it locally on your computer. You can build off of it, iterate off of it. You have full access to the code for this Dream Booth platform now. You can find it over at stability AI's official GitHub. You can see all of the code is available to download. They have instructions on how to install it. Now, one thing to keep in mind about using Dream Booth, while the Dream Booth platform itself is open source, they do say right here you'll need to have your API key handy to use the default Stability API plugin. In order to use the API, there are actually costs involved. If you check out Stability AI's documentation, they have all of the details about how much this actually costs, but they appear to be moving to a model where you can open source the platform that actually creates the images. But to access the API behind the scenes, you still need to pay Stability AI to use their API keys. So the tool is open source, but it's not necessarily free to use. So just keep that in mind. Now, in the world of robotics and AI. This week, a company called Sanctuary AI demoed their new humanoid walking AI robot. This demo here went somewhat viral on Twitter, showing off the robot that looks like it could be straight out of iRobot. And it's called Phoenix. They call it a general purpose robot. So essentially it's a robot that they see being able to do anything that humans wouldn't possibly be able to do. I mean, maybe it'll be able to do your laundry or wash your car for who knows? One thing that's good to know is that it's got a speed of 3 mph. So if this robot ever does try to attack you, all you have to do is walk briskly away from it. The first time I saw this, I thought it was somebody having fun with AI video tools. Not to Be outdone. Several hours later, Tesla dropped their own video of their own humanoid robot called Tesla Optimus. And while this first video went semi viral, Tesla's video went really viral. And this video actually shows the robot walking around a scene that looks straight out of iRobot as well. Some crazy jumping abilities of the robot and also the fact that, look, it actually won't crush the egg. And this guy has to prove that it's a real egg. It also memorizes its environment and it showed off some of the computer vision technology that's built into it. They also showed off that it's actually trained by humans doing things. So you can see that humans do things and then the robot can actually repeat the motions and actions that the humans did. So right now, we're seeing a bit of a battle between all the large language models and the company's for large language models. Are we also going to start to see a battle between AI robotics companies and who's going to release the first multipurpose humanoid robot? I don't know. We'll see, but it's still fun to watch. Now, in the world of bizarre AI news, a Texas A and M professor fails an entire class of seniors, blocking them from graduating, claiming that every single one of them used ChatGPT. And the way that he determined that every single one of them used ChatGPT was that he took their essays, copied and pasted them into ChatGPT and said, hey, ChatGPT, did you write this? And for every single one of them, chatgpt said, yes, that looks like something I would have written. But here's the problem with ChatGPT, it says that about everything. You can literally copy and paste almost anything into ChatGPT. And ChatGPT will tell you that that looks like something I would have written. So it's not a very good test of whether or not something was written by ChatGPT. And that's pretty much the breakdown of all of the cool stuff that happened in AI this week. And next week is going to be exciting as well. I've even got some sneak peeks of what's coming next week. One of my favorite AI video companies called Kyber, just dropped this teaser on Twitter of a new Text to video AI that they're about to release next week. May 23rd is the date that they're going to drop access to this, and I'm really looking forward to getting my hands on it and playing around with it. Also, one of my favorite AI art companies, Leonardo AI, is releasing a brand new pipeline for their image generation. It's called Leonardo Alchemy and they actually sent me some teaser images to show off the type of generations that we can get out of this new pipeline. So Leonardo Alchemy is also dropping next week and these images are just awesome. So that's something else we have to look forward to next week. And finally, in Seattle next week, Microsoft is putting on their Microsoft Build event. It runs from May 23rd to May 25th and this is essentially an event that they put on for developers to show what they're working on that developers can build off of. Some of the speakers include Satya Natalia Nadella himself from Microsoft, Greg Brockman, the co founder of OpenAI, as well as other various executives over at Microsoft. We can all expect some huge AI announcements around the future of Microsoft and OpenAI coming during this Microsoft Build event. The last few weeks in AI have really ramped up in pace again and I really don't see next week slowing down just based on what I've seen already. And if you really want to stay in the loop with all of this crazy AI stuff that's happening, make sure you check out FutureTools IO. This is where I curate all the coolest tools that I come I keep the AI news section up to date on a daily basis and if all of this is just way too overwhelming, you just want a once a week TLDR of the entire AI space. Make sure you join the free newsletter because every Friday I send out an email that just has my five coolest tools that I came across, a handful of news articles, a handful of cool YouTube videos, and one cool way to make money with AI. I said it every Friday. It's the grand overview of the most important things that happen in the world of AI that week. And again it goes out every Friday. You can sign up by going to FutureTools IE and clicking on Join the free newsletter. And if you want to see more videos like this in your YouTube feed, make sure you give this video a thumbs up and subscribe to this channel. And I'll make sure videos like this keep popping up in your newsfeed. Thanks so much for tuning into this video. I really appreciate you. I'm going to keep on making these news videos every single Friday for you to keep you in the loop with AI as well. So if you're not subscribed, why not do it now? Really, really appreciate you. See you guys in the next video. Bye.

By.

Sam.