Google Just Went ALL-IN on AI (Everything You Need to Know)

So today was the Google I O event and this is actually an event that I decided to come and be a part of in person. This is the first Google event I've ever been to. It was a really cool experience. I'll probably make a video in the future where I talk a little bit more about the whole experience of the event. But for this video I want to break down some of the big announcements that Google made during the I O event. And make no mistake about it, this event was all about AI and and the various things that Google is now putting AI into. I'm actually hearing a lot of people over on X claim that they thought that OpenAI was a bigger announcement yesterday than what Google made today. I feel like OpenAI's sort of major announcement yesterday was bigger than any one announcement today, but there was a whole bunch of announcements out of Google today. So let's break them down real quick. Starting with the fact that all Gemini Advanced subscribers, the people that actually paid to use Gemini, now have access to the newest model, Gemini 1.5. And it's got a 1 million token context window. So the amount of words, input and output that you can get back from this model is about 750,000 words. It's a huge context window and they also announced that this context window is going to expand to 2 million tokens, which is about 1.5 million words input and output when you're working with one of these large language models. One really cool demo that I liked from the keynote was when they showed off the ask your photos feature where you can ask questions like what's my license plate number? And it will look through all of your photos and based on what it sees in all of your photos, find your license plate number. Or you can ask it when did Lucy learn how to swim? It will search all of your photos and find photos where it sees Lucy swimming for the first time and then can reflect back to you. When Lucy learned to swim for the first time, they showed off Gemini being in Gmail and at any time if, if you're wondering what Gemini is, it's basically their large language model that is powering pretty much all of their AI tools. It's the little chat window that will pop up that you can have conversations with. And they showed off an example of it being used in Gmail where you can ask it a question like summarize all of the announcements that came from my kids school and it will look through all of your emails inside of your Gmail, find everything related to your kid's School and surface it for you inside of the AI chatbot that so you don't have to click through and look into individual emails yourself. They showed off the new features being added to their notebook lm. I thought this was really cool because they showed off an example where you could put a whole bunch of documents in there, you can put some audio notes in there that maybe you just recorded on your phone, and then it'll essentially create like a podcast of this content for you, which almost sounds like you're listening to a radio show or something describing the information inside of NotebookLM. But then you can interject as the conversation's happening, you can sort of stop it and say, wait, I have a question. Ask the question. And then it goes back into this like podcast mode. But while answering your question, I thought that was a really cool little feature. They also made it very clear that they're working towards AI agents, things that will do multiple steps for you. Instead of just going and saying, hey, answer this question for me and you give it a prompt and you get a response back. You could tell it to complete a task for you and, and then it will go and try to complete all of those steps to complete the task. One of the examples was return these shoes for me. And it went and figured out where the shoes came from, how much it costs, the customer support details, and then could actually contact the shoe seller on your behalf and get a refund on the shoes. So I think this AI agents concept is going to be a huge concept that we hear about more and more and more. And a lot of these companies that are developing AI are probably, probably going to start showing off their agent like features. And today I think Google made a big step forward, being one of the first companies that I know of to show off a really easy to use AI agent that can use all of the tools that you kind of already work with anyway. I mean, we're talking about Gmail and Google Drive and Google Sheets and Google Docs and Google Meet and all of these tools that you use under the Google umbrella, these AI agents are going to have access to that information. Now my one worry with the AI agents is that we saw a demo today. We got really excited about what it's going to be able to do and work across all of our data. But Google sometimes has a tendency to announce stuff, get people excited and then take forever to ship or just never actually release it to the public. So hopefully that doesn't happen. But these AI agents are the next sort of a wave that I think we're going to see a lot of AI companies showing off. Even when we saw OpenAI's demo yesterday, the whole chat thing that they were showing off is sort of a step closer to AI agents. I think this is what a lot of these companies are pushing for. This is the future of what AI really should be and what all these companies want AI to actually be. But from the presentation, it looked like it was going to be really easy to use and have easy access to all of the data we wanted to have access to. They also had Demis Hassabis there, who is the leader over at DeepMind, and he shared some really cool stuff that they've been working on as well. He showed off their new model, their lightweight model called Gemini 1.5 Flash, which is a much smaller, much lighter model designed to run really fast on, like, mobile phones or when you need a really quick response. That's what that model is designed for. And the real showstopper, in my opinion, was when they showed off Project Astra. Project Astra is their attempt to create a real time AI agent that's really, really useful and, and can use your camera on your phone. Now. I was actually able to see this and demo it and experience it in real time. And it did work in real time. The demo they showed me, they had this sort of downward facing camera and they would put stuff below the camera and then ask questions about what it saw on the camera or ask it to tell a story about what it saw on the camera. And it worked. It worked pretty quickly. They kept on making a point during this whole keynote that what they were showing was real time with, which is an obvious sort of overcorrection from the last Gemini announcement, where it wasn't shown in real time, but everybody was sort of led to believe that it was real time. So this event, they definitely overcorrected for that and constantly told us, this is live, this is real time, this is actually how fast it works. But they showed off a really cool demo where they pointed their camera at like a speaker and then they drew on their phone and said, what is this part of the speaker called? And it looked at that image plus the drawing and said, oh, that's called the tweeter. And then they looked around the room some more with the phone and were able to ask questions about their environment. What's different about this from what we've really seen before is that this was just looking at the video feed instead of snapping a photo every single time, it was just watching the video of what was going on in the camera and you can ask questions and get responses in real time of what it was seeing on the camera. To me, that was one of the most impressive demos that they showed off. And normally I would say I want to wait till I get my hands on it so I can play with it, so I can really tell you how I feel about it. But I was able to get my hands on it and play with it and it was cool and it did really work. They showed off Imagine 3, which is Google's version of like Dall? E. It's their image generation platform. To me, it didn't look, you know, head and shoulders over everything we've seen before. It looked pretty good. The biggest advancement that this one made is that this one now does text pretty well. So it kind of has caught up with Dall? E3 in that way and Ideogram in that way, where it can actually inject text into your images. They showed off their generative music tool, their music effects, which we've been able to play with for a little while now. So I've had some experience playing with that. That wasn't anything super new to me. But what was new was when they showed off VO or VEO V E O. This is their new video generation model, which looks like it was designed to compete a little bit with Sora. Doesn't quite look the same quality level as Sora, but we've really only seen sort of cherry picked examples from Sora anyway. But it does shoot video in 1080p. It can generate for longer than 60 seconds and they open the wait list, meaning they're going to let people actually use it. Unlike Sora, which we have no idea when we'll be able to use it now, a lot of the stuff that they showed off at the event today, you can actually try it out. I don't know if it's available in every single country yet. Some of the tools are only available in the us. Some of them are opened up worldwide. But if you go to Labs Google, a lot of this stuff is available for anybody to sort of play around with and experiment with right now. You can also sign up for the wait list to get access to the Veo text to video model that they showed off today that is going to kind of compete with Sora. Now nobody really has access to it yet, but you can get on that wait list again. You can find that over at Labs Google. One of the funniest moments I thought was when they talked about Google getting more AI built into it and that Google will be able to Google things for you. I thought that was interesting. Google also showed off their new AI overview feature that's going to be rolling out into the Google search engine. This new search feature has what they call multi step reasoning. You can ask the search engine multi step question questions and the search engine will actually respond with a rundown responding to all of the steps that you asked it. The example they gave in the keynote was find the best yoga or Pilates studios in Boston and show details on their intro offers and walking time from Beacon Hill. And it managed to look up all of this information and give a summary back inside of the Google search results that answered all of those questions and find the best response for for the person who asked the question. Something like this could totally change the way people actually use the Google search engine. You're not gonna just go and type Pilates studio San Diego. You're gonna say, I live in this part of San Diego. I need a Pilates studio. I want it to be at walking distance and I want to find one that's got special discounts going on right now and it can actually look at all of that information, hunt it down for you and and present it to you right inside of the search result. So totally different way of searching and I'm really, really excited to see that roll out into the Google search engine soon. Man, there was just so many announcements at this event. I think if OpenAI was trying to impress us with one big announcement, Google was trying to impress us with a whole bunch of little announcements. They also showed off Gemini's real time captioning, the ability to summarize things across multiple emails to save you time. You can even create workflows and that use Gemini and repeat that same workflow over and over and over again in the future. They also showed off something called Gems, which looks to me to be Google's answer to OpenAI's GPTs. They're sort of pre trained chats with like some extra system prompts built in so that you get a similar output every single time. This isn't something that really excites me because the GPTs never really took off. I do have a few GPTs that I use that it just saves me like one step of giving it some extra information to start with. And it kind of seems like gems are going to do this as well. They seem cool and interesting, but nothing that like really blew my mind. There's also something really cool that they showed off where while the guy was on stage he was getting a phone call and as soon as the phone heard that it sounded like it could have potentially been a scam. His phone warned him that you're talking to somebody that could potentially be trying to scam you. That was crazy. It was also a really funny moment of the keynote, but they're building AI into their Android phones that can detect if you're potentially talking to a scammer. That's kind of cool. Hopefully we see Apple do something like that because I personally use an iPhone most of the time and I would love to see that feature in the iPhone. They also talked a little bit about open source, which is interesting because lately we mostly talk about meta releasing stuff, open source. Well, Google's starting to do it now with their Gemma models. They talked about a model called Pali Gemma, which is a multimodal model that's open source, that can actually see images and things like that, but anybody can build off of it because it's open source. They're building Gemma 2, another open source model that's going to be 27 billion parameters this time. And at the very end, Google CEO used AI to count how many times AI was actually said in the keynote. And according to them, it was 120 times until he said it again after that. So overall, my final thoughts on this event. I thought they showed off some really, really impressive stuff. Now, most of it isn't anything that absolutely blew my mind. I was probably a little more impressed by what I saw during OpenAI's keynote the day before. But this had a lot of cool features shown off that I want to get my hands on and I want to use. I want AI agents. I want to play with this new VEO model where I can generate videos similar to Sora. I can't wait to get my hands on the tools that can sort of search all of my Google Drive, all of my Gmail history, the audio notes that I upload, the attachments in my emails, and it can look through all of that as the context for whatever I asked the chatbot that's going to be so valuable to so many people. I am so excited about that. The other feeling that I got is, you know, it's so easy for us to look at big companies like Google, like Meta, like Microsoft, like OpenAI and Go, those are just huge faceless corporations that they don't care about us. But when you're at an event like this, you just don't feel like that's true. An event like this really shows you the sort of human side of these giant mega corporations. These companies are made up of individuals. They're made up of a whole bunch of people that are just as excited and just as sort of nerdy about this tech world as I am and as you are. And they're so excited to show off that little feature that they're working on in their corner of Google. They're so excited and so, you know, pumped up to tell you that this is what we've been working on, and we finally get to show it off on this stage today. I've had so many little side chats with people that work here at Google, and that is one of the things that really clicked in my mind from this event is the sort of human element that a lot of people don't think about when they think of Google. They forget that, like, one person at Google built this really, really cool thing that they're really excited about that's part of the keynote today. But the people watching the keynote, they just see the whole thing. They just see, this is Google as a whole, and here's all the stuff they're throwing at us. But every little announcement that they made today was the hard work, the excitement, the how do we push this harder, make this better? It's that sort of attitude from one or two or a handful of individuals within the company trying to put this out. So while I even fall in the trap of going, oh, this is just Google, this is just Microsoft. This is just a big, faceless, nameless company, you know, trying to collect data or whatever we all say about it, whatever narrative you want to play with, the reality is this is a company of individuals that are doing their best to build what they think is going to be genuinely helpful. And that is what I got out of so many of the side conversations outside of the keynote. That's the one thing I want people to take away from this video, is that it is easy to look at Google as this mega corporation. But it's so satisfying to be at this event and talk to the individuals building this stuff and hear their excitement and their enthusiasm behind what they're building. And that, to me, is what Google I O is all about. And why I want to be here in person is I want to experience that firsthand. And I wanted to talk to the people building this stuff because these people love what they're doing. They're passionate about it. They want to build things that help, and they're excited to be putting it out into the world. And that, to me, is what Google IO was all about. That's all I got for you today. I hope you enjoyed this video. Hopefully you're updated on all of the things that I thought was interesting from Google. I o There was a few things that I skipped over, but they were the less impressive things that were brought up. Hopefully this gave you a good overview. I really appreciate you. If you like this video, give it a thumbs up and maybe consider subscribing to this channel if you want to stay looped in on all of the latest in AI and and be up to date on all of these events and all of the coolest tools and all that kind of stuff. Really, really appreciate you. Thanks for nerding out with me. I'll see you in the next video. Bye Bye.