Google's Vertex AI: Democratizing AI Model Building

0:00Google has recently announced that they are releasing their advanced AI services to the general public on their platform. So today on the podcast, we're going to break down what this means and what those services are. So recently, Google had a big conference called Google I.O. Typically, this is just a developer conference where they talk to the developers. But this conference really kind of should have been called Google AI because it was them discussing how they were integrating AI into virtually every product they offer. Now, the target audience still was developers. So since this happened, they have now rolled out or announced that they'll be rolling out the ability for developers, the people that are at this conference, to use a lot of the tools and a lot of the AI software that they mentioned on their platform.

0:44So their platform is called Vertex AI, which is part of Google Cloud. So now businesses and organizations can super easily integrate the platform's features into their own software apps. So the new update allows developers to use a really wide variety of new tools and AI models. For one of those, as an example, is the world completion model, which is powered by a system called Palm 2. Palm 2 is kind of this big announcement they made at Google I.O. And there's a feature called the embedded APIs that kind of helps them handle text. So developers can also access a lot of other basic models in an area on the platform called the Magic Garden.

1:23so they can make use of these features within the generative ai studio which is something that google announced at this google io conference as well and they can use that to tweak and launch their own custom models this is massive i think with powerful tools like this coming out of google allowing people to create their own custom models very easily i think we are going to see an explosion of new and really creative ai models with data sets that were not publicly available before coming into the limelight. So Google makes sure that there are really strong security and data handling features built into Vertex AI platform.

2:01They've made that very clear. So let's talk about the model garden. This was something that was recently announced and it's kind of an area on their new platform. And essentially they have access to a huge range of different base models that are created by Google and its partners. And so they have over 60 models available right now. And one of those is Google's Codi model, which essentially helps with completing code and creating chat applications like ChatGPT. And this is, you know, officially open for public preview right now. And so I think this is going to be really impressive just having the ability to start with 60 of these base models, build off of them in your own way on the Vertex AI platform.

2:43So the platform really provides a very comprehensive set of tools to help you adjust, launch, and manage models in your operations. And I think this is the first of its kind, like this platform is really the first of its kind to offer a feature known as reinforcement learning with human feedback, which I think is really needed and impressive. But essentially, it improves the accuracy of these models that you're able to custom train on your own data by using human feedback. So they have that built right into it. And with the Generative AI Studio now widely available, developers can use a really wide range of tools to help build custom AI applications more quickly.

3:26And I think this is going to be especially beneficial to companies that have big data sets, their own in-house data. and now they're able to go and kind of feed that into some of these base models and train them to be really powerful specifically built on their own data and this I think is a really good move for Google you know they have all this AI technology and if they're only using it on their own data I feel like it would be sort of limited where right now where they're kind of sharing it out to the broader community and integrating it into Google Cloud and their other platforms they're now able to you know monetize this AI technology because they know inevitably all of these companies are going to be building AI products are going to be building AI tools.

4:02And so by offering, even though, you know, some people might be like, Oh, man, why doesn't Google just keep that internally, because it's so powerful, you know, for their own purposes, I think the the implications of most of their technology is so wide reaching, it affects so many industries, and it could benefit so many areas. And they know that that would also be incredibly lucrative to build these kind of enterprise software. So I think Google is not just holding this all back, they definitely want to get on the side of sharing this and also profiting from it. So Google has actually shared examples of a number of different companies that are currently using this platform.

4:36One of them is GA Telesis, and they are using the Palm model to build a system that uses emails and email orders to generate quotes for customers automatically, which is really cool, right? People are sending an email saying, hey, we need X, Y, and Z with these specifications for the product, and they automatically can actually generate quotes and just respond back. That would save a ton of time. As someone that has worked in a lot of different industries that do quotes, specifically in travel in the past, I think this would have been an incredible feature. I remember one business in particular where we sold a really high-end luxury resort, and we got just an insane amount of people asking for quotes because there wasn't a very clear pricing model for the resort.

5:20There wasn't a booking engine that specifically said how much each night was. It was very exclusive and expensive so you had to do these kind of like custom quote things and having a product like this would have been really incredible where just the emails asking for quotes where you know we still had them fill out the data about like how many people they wanted to stay in what nights and you know all the info that we needed but then we had to go check that and you know contact people I think that something like this would have been really powerful to be able to integrate and have automatically sending out these kind of email quotes for something a little bit higher end and a little bit more customized than, you know, what you could typically get off of a booking engine, for example.

5:58Another example of a company that is using this is GitLab. GitLab has a feature called Explain This Vulnerability, and essentially it's using what is the coding model to give developers a description of code errors, and it, you know, does it in plain English, and then offers, you know, suggestions for how they can fix those tools. So another example is Canva. You know, that's a tool I have been using for years and years. It's an online design tool, and it uses Google Cloud's AI to translate languages for its users who don't speak English. So another couple companies are Typeface and Datastacks, and they're also using Vertex AI to develop new tools for just a bunch of different AI applications.

6:38So this is, you know, obviously a really powerful software being used by a lot of really big companies. And I think that just the fact that Google really ensures customers that their data is fully protected and remains within their own control at all times when they're using the vertex ai and generative ai building platform i think google i think that's going to make it so that people are a lot more comfortable you know even with my own ai startups talking to investors a question i've gotten a lot is like hey well how do you handle data privacy because you know i wouldn't put my data into generative ai that's going to use it to train and yada yada, which is a very valid concern.

7:17And so I think that the way Google is handling this is completely the right way. And I think it's pretty awesome because essentially, the way that they handle the safety of their user's data is so that they won't ever use the data for training their own models. And oftentimes, it doesn't even leave the user's account. So I think that's really awesome and they can keep it you know safe and encrypted and on that platform so I think in this really rapidly growing field of AI where new content is created using machine learning you know different there's a bunch of different cloud service providers that are trying to become the leader I think Google is making a lot of really good is taking a lot of really good steps in the right direction to maintain market I mean it's hard to say market dominance because it seems like Microsoft and OpenAI are always in the news with powerful tools but I think that and and they probably are dominating at the moment, to be honest, with integrations.

8:15But I think Google has a really powerful set of AI tools. I think they have the distribution and the ability to have the distribution. So I'm going to be very curious to see how Microsoft responds to some of these tools, how OpenAI tries to compete, and who becomes the ultimate AI winner at the end of the day. I feel like this might be an industry with, you know, one major leader. But who knows, it looks like we're shaping up to see a bunch of big players. And I think Google is definitely going to be is not going to let Microsoft take the lead that easily or open AI. And they're definitely going to be a very solid contender in this space, if not the market leader.