The AI Daily Brief: Artificial Intelligence News and Analysis - The AI Arms Race: Where the World's Biggest Companies Stand

Episode Date: April 17, 2023

Google is working on a top secret search update called "Magi." Amazon announced Bedrock last week. This episode covers the AI strategies of Google, Amazon, Microsoft, Meta, Adobe, and Apple.

The podcast you're about to hear was originally released as a YouTube video on Monday, April 17th. In this show, we look at what all of the different tech giants from Facebook to Google to Microsoft to Apple are doing in the AI space, where, in other words, they fit in the AI arms race. Welcome back to the AI breakdown. Today we are checking in on where the biggest companies in the world are with regards to the AI arms race that is right in. now just eating everything in big tech and beyond. And the specific context or prompt for this was news that Google is finally trying to catch up after being really behind, at least it appears to be. And so to kick off, before we get into what exactly Google is doing, I thought that this piece in the New York Times gave some interesting background. It's called Google devising radical search

changes to beat back AI rivals. And where the piece starts is, is with a concern. It reads, Google's employees were shocked when they learned in March that the South Korean consumer electronics giant Samsung was considering replacing Google with Microsoft's Bing as the default search engine on its devices. AI competitors like the new Bing are quickly becoming the most serious threat to Google's search business in 25 years, and in response, Google is racing to build an all-new search engine powered by the technology. Now, this comes from internal documents that the time saw, and it appears that this is called project name Magi. So what is going into Magi? Let's get a few of the details.

Currently, there are something like 160 engineers working on the project, which seems small in terms of Google scale, but you don't necessarily need a huge number of people to make a big difference. This tool will apparently have a chat-like interface, so taking a queue from chat GPT. And that's going to be different, obviously, than Google's existing method of displaying 10 results per page. Instead, there's going to be a chatbot that provides instant answers. Now, if this is true, one of the implications is potentially a pretty fundamental shift in how we think about search, right? Right now, our paradigm of search just is Google. That's what we think when we think of search. If everything shifts to this sort of more chat-based interfaces, that's a new type of UI, UX experience that human,

will have to adapt to. And maybe that's better in the long run, but it will represent a change. What else is going on with this new potential AI-powered search? One of the ideas is that transactions could happen directly in the search experience. So instead of Googling for a flight and then going off-site to buy it, why can't you just buy it from within the search experience? That's the idea here. Now, public release of this is apparently coming in a month. It'll be rolled out incrementally initially to about a million users and then 30 million by the end of the year will have access. Now, this isn't the only AI tool that Google is working on. There are other products like GIFI or GIFI, which is an AI to generate images in Google image search results. There's a tutoring platform that

teaches users new languages using AI conversations and probably a lot more that we don't know about yet. So this is the breaking news, I guess you could call it that. Google is Trump. trying to catch up and at least recognizing that they have an issue. But for me, when I was reading this morning, it brought up the question of where is everyone else? Last week, one of the biggest pieces of news came from Amazon. They announced Bedrock, which is their platform for building and scaling generative AI applications.

And the way that it does so is by giving people access to other foundational models. So this is very much a builder tool. Basically, the idea is that Bedrock will allow enterprises and individual developers to use and fine-tune base models from companies like Anthropic, like Stability AI, Amazon Titan, which is Amazon's internal LM, to build new types of applications. So this is integrated with the AWS product group, and it seems pretty clear that Amazon is going after that developer set. They also have something called Code Whisperer, which is basically their version of the very popular co-pilot that lives inside GitHub, which uses AI to help developers be more efficient in their coding. So

Amazon at least so far appears to be pursuing a strategy focused on developers, or at least that's the niche that they're trying to carve out initially. Now, when it comes to Microsoft, obviously Microsoft has made a huge investment in OpenAI. In fact, they made a couple of investments. The most recent came in January, but that followed investments from 2019 and 2021. This most recent investment was something like $10 billion on a $30 billion valuation. Now, of course, when it comes to what Microsoft is doing with OpenAI, in addition to just reaping the benefits from being invested in that platform, they're also using the breakthroughs that OpenAI is making to change and soup up and power their suite of tools. Their productivity suite, their browser,

obviously Bing being a real contender for the first time is exactly what Google is recognizing as a threat. So Microsoft's big strategy, I think you can say for now is Open AI integration and everything that brings with it. And what? What about meta? I think in some ways this headline really captures it all. Mark Zuckerberg bows to pure pressure, announces Pivot to AI. How many Pivot's has Facebook done now? Now what they're poking fun of is the fact that Facebook very famously change its name to meta, to reflect its commitment to the Metaverse, only to less than a couple years later, basically abandon that or seemingly abandoned that and move into this new, hotter area. Now, of course,

Anyone who's watching this probably has a sense that AI is not just another hype sector, but is something that is going to pretty fundamentally change almost all of our experiences, certainly on the internet, and pretty far beyond that as well. So it's a necessity for companies to get in the game. But it still is notable that Facebook is coming off of this reputation of being flippant, let's say, or maybe moving around and pivoting a little bit more companies. Now when it comes to what they're actually doing, there's a few different things. In February 27th, Mark Zuckerberg announced that there was a new AI group that was working on experiences that could connect to their existing tools, such as Messenger and WhatsApp.

But a few days before that, they also released or introduced Lama, which is their large language model. That's their sort of version of these LLMs that other companies like ChatGPT, OpenAI are building upon. Then they also are, as time goes on, releasing these sort of little patches and open sourcing different tools. in and around AI. So last week, for example, they announced that they were open sourcing the code for a drawing animation tool that was built on 180,000 annotated amateur drawings as something that the community could start to build on. They also just this morning announced something called Dyno v2, which they say is a, quote, state-of-the-art AI model that trains itself on vision tasks, like estimating depth, segmenting objects, and comparing image similarity. For example, we can use it to map forest canopy height, across, different continents using satellite imagery. In the future, this could help improve medical imaging, food crop growth, and help make the Metaverse more immersive. My feeling looking at Facebook relative to some of these other companies is that it just doesn't yet know exactly what its meta, pun intended, AI strategy is. Sure, they're building all these tools,

but what is going to be Facebook stamp? Where do they want to play in this space? I'm not sure that they even know yet. Another company that I think, even though it's maybe not as big as the Microsoft's and Facebook and Googles of the world, but that is worth noting is Adobe. Adobe in March announced Firefly, which is basically their family of new generative AI tools that will sit across their suite of products.

We've already seen some really cool things built on with Firefly. Chris Kashtanova, who does tutorials on all of these different image tools, made a comic book using Adobe Firefly. It really showed off some of the features of that platform. And then just this morning, they announced that they were adding these

generative AI tools into Premiere Pro, which is, of course, their premium video editing software. So I think in contrast to Facebook slash meta, which is experimenting with a ton of things but doesn't have a clear strategy, Adobe is an example of a company that has a super clear strategy. They are using AI to change the suite of creative tools that they have always built their business around. And they want to be first movers when it comes to how those tools can transform, how they're people create content, how they create documents, how they create movies, how they create art, and so on and so forth.

And I think one thing that's also worth noting with Adobe Firefly is that they're positioning it as a tool that is built and trained entirely on rights approved images. So no weird questions of licensing, of pulling images that they shouldn't have access to, which probably means that they think that will be a bigger issue, legally speaking, going forward. Okay, so we've gone now through the vast majority of the big tech giants, the fan companies, and even into some others like Adobe, there's one obvious exception or one obvious company that's missing, and it's glaring in its absence. Fast Company wrote a piece very recently called Apple's silence on generative AI grows louder. We don't have any idea what Apple is doing in AI right now. We have lots and lots of speculation and lots of theories,

and we have Tim Cook saying that it's a quote-unquote major focus without going into what that means, but we just don't know what Apple is doing on AI. Now, one reason why they might have a bit more of a challenge, Fast Company writes about here. They call the section the privacy problem, and they say adding generative AI to its assistant may be harder for Apple than for its big tech peers.

The company's dogged focus on privacy has been an effective. effective cudgel against meta and Google in the social networking and advertising realms, but it's a serious impediment to building an assistant that harnesses the goodness of generative AI. In order to transform Siri, to make it smarter and more human feeling via a large language model, Apple would have to open up access to a broad swath of public information from the web, and more importantly, personal user information, their communication style, plans, priorities, preferences, health stats, tastes, and relationships. Apple has so far been willing to train models this type of data only if the model can run solely on the user's device versus in the cloud,

and if the data remains invisible to Apple itself. So the point here is that Apple, because of its A, position on privacy issues, but B, its publicness of that position is maybe boxed in a little bit. It is definitely the company that people are most anticipating getting involved. Smokeaway writes, Amazon making AI moves with AWS. Apple next week, question mark. Now, one last thing before we close out, I did a poll earlier today where I basically asked folks on Twitter if they thought that AI favored incumbents. And the logic would be around the idea that these models require an incredible amount of data, which might not be accessible to startups, that the cost of powering these big AI models is extraordinarily expensive. And it's really interesting

how little consensus there is about this. Now, obviously, it's a very small sample size, just over 100 votes, but about a third said that AI favors incumbents, 25% said that it's still open to disruptors, and 41% said it's still too early to tell. In fact, a number of people said that they would have said incumbents about a year ago, but the way that Open AI has really set the tone for the entire space makes them think differently, and I think that's a valid point. Anyway, I think we're going to spend more time on how much room to run startups have and what the challenges of that might be, how the regulatory sphere could impact that. One of the things that regulations and regulatory compliance regimes tend to do is

prioritize companies that have budgets for compliance over startups that don't. But there's lots and lots of questions that will go into this. For now, that is where I'm seeing the big company stand in this AI arms race. Hope this was useful. Go please check out and subscribe to the AI breakdown wherever you listen to podcasts. That's the audio version of this show. And if you haven't yet subscribe to this channel, And lastly, let me know what you think about Apple and what they're going to do in the comments.

I'm super curious. I've got my own theories, but I want to hear from you. All right, guys, until next time, peace.