The AI Daily Brief: Artificial Intelligence News and Analysis - Tim Cook Is FINALLY Hyping Apple AI

Episode Date: February 2, 2024

Apple AI is coming this fall! At least according to a recent earnings call with CEO Tim Cook. Also, Bard announces image generation, Microsoft brings Copilot to Sales and Service, Amazon introduces Rufus shopping bot, Meta is implementing its own chips and more. ABOUT THE AI BREAKDOWN The AI Breakdown helps you understand the most important news and discussions in AI.  Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown Join the community: bit.ly/aibreakdown Learn more: http://breakdown.network/

Today on the AI breakdown, Apple's Tim Cook is finally talking about generative AI, at least in so far as he's hyping it up. Before that on the brief, Meta, Google, Microsoft, and Amazon all share AI updates. The AI breakdown is a daily podcast and video about the most important news and discussions in AI. Go to Breakdown Network for more information about our Discord, our YouTube, and our newsletter. Welcome back to the AI Breakdown Brief, all the AI headline news you need in around five minutes. Boy, I got to say, closing out the week, everyone decided to announce things in the last little part of the week. Like I said, we have announcements from basically every big tech company, and hey, our main episode is going to be about Apple, so it really is everyone. We're going to kick it off with Google, where Bard has gotten some major updates.

The first has to do with where it is available. Bard is now available in over 40 languages, and people are reporting really good results. Verun Krishnan writes, wow, Google Bard with latest update has gotten quite good. understands even a mix of Tamil in English, tried it with other languages too, and it's incredible. Jack from the Bard team was also really excited about this. He said blind evaluations with our third-party raiders identified Bard with Gemini Pro as one of the top-performing conversational AIs compared to leading paid alternatives. Now, as part of bringing Bard to more languages, Google also expanded what they call their

double-check feature. Basically, whenever you click on the little G-I-con inside Bard, the program goes out and looks to see if there's content on the web that substantiates its response. You can click on highlighted phrases and learn more about either supporting or contradicting information that's found by traditional Google search. And yet, without a doubt, the thing that people are talking about most is that image creation is now integrated in Bard natively as well. Their announcement post writes, for an extra creative boost, you can now generate images

in Bard in English in most countries around the world at no cost. This new capability is powered by our updated Image Gen 2 model, which is designed to balance quality and speed, delivering high-quality photo realistic outputs. They also announced that images created with Bard will use Synth ID to add digital watermarks into the pixels of any images that they generate. One thing that people are noticing is that it seems to perform really well with words. I haven't had a chance to really look at this systematically yet, but some are even suggesting that it does words better than Dolly 3.

Now, given that this is free, you have to think that this just significantly increased Bard's competitiveness in the AI space. Over in Microsoft Land, the company has been slowly rolling out its co-pilot across its suite of 365 products, and copilot for sales and copilot for service is now generally available following the general release of those tools for applications like Outlook, Word, and PowerPoint that came over the last several weeks. They call co-pilot a, quote, AI assistant design for sales teams to maximize productivity and close more deals. It can do things like generating sales meeting preps in Word, summarize emails in ways that surface relevant buying intent, adding leads in updating

CRM records, and a number of other capacities as well. Co-pilot for service, they say, quote, unlocks an organization's trusted knowledge to accelerate onboarding and case resolution, improve efficiency, and automate tasks for agents in their flow of work. Without costly development time, organizations can simply point to their data and, in a few minutes, unlock generative AI powered conversations across their knowledge bases. And so, if the theme of the year is integration of generative AI tools into our existing workflows, this hits that nail pretty well on the head. Over in MetaWorld, we got a report from Reuters that the company is planning to deploy their in-house customized chips this year. Now, their source is an internal company document that was

seen by Reuters, and the goal is, of course, to reduce dependency on Nvidia. Now that said, of course, meta has touted just how many chips they're going to be buying from Nvidia this year, and when it comes to compute, it's really not an either-or, but a both-and, as everyone just needs more computing power. A spokesperson confirmed the story and a statement saying, we see our internally developed accelerators to be highly complementary to commercially available GPUs in delivering the optimal mix of performance and efficiency on meta-specific workloads. Now let's head over to Amazon, where the company has announced Rufus, which they call a new generative AI-powered conversational shopping experience.

This is ChatGPT for shopping on Amazon. As they put it, this is a shopping assistant that they say is trained on Amazon's product catalog that can help customers answer questions on everything from shopping needs to products to comparisons. And again, is one of those sort of boring products on the face of them that will be used by basically everyone. The example they give is a user who's trying to be. to figure out which running shoes to buy. They might start with a question like what to consider when

buying shoes. As they get farther along, they might have different questions like what are the differences between trail and road running shoes. Finally, as they hone in on a specific pair, they might want to compare them or ask about a potential option. For example, a question like, are these durable? Now, Rufus will be available initially through the Amazon shopping app, and customers will be able to type or to talk into it to get Rufus to answer. Initially, it will be rolled out only to a small handful of customers, but that will expand over time. Some think that it doesn't go far enough into a next generation shopping experience. Amber Adtherton writes, Amazon's Rufus AI shopping agent is very Amazon focused on product

comparisons and reviews. Give me the fun shopping agent that sits in my browser, services and buys, all the things for me based on my inputs and browser history. Finally, a fun and very quotable statement from Sam Altman in an interview with investor Alexis O'Hanian, Sam argues that at some time in the not too distant future, a one-person company will be able to hit a billion dollar valuation. That is a heck of a thought and a fun one to think about as we head into the weekend. For now, however, that is going to do it for today's AI breakdown brief. Up next, the main AI breakdown. Welcome back to the AI breakdown. Well, here is one that people have been waiting for of all of the big tech companies, the Magnificent Seven, whatever you want to call them,

the biggest holdout when it comes to the generative AI explosion has of course been Apple. What we're going to do today is look at Apple's latest comments from their earnings call this week, what it suggests about their strategy, and then we'll try to do a little bit more of a what we know so far around what we can expect from Apple and generative AI. So the big news here is that Tim Cook voluntarily and intentionally, and not just in response to a question, brought up AI and more specifically generative AI on this quarter's earnings call. In his prepared remarks, he said, as we look ahead, we will continue to invest in these and other technologies that will shape the future. That includes artificial intelligence, where we continue

to spend a tremendous amount of time and effort, and we're excited to share the details of our ongoing work in that space later this year. Now, of course, analysts saw that as a total opening to try to get more information out. And yet, presses they might, Cook would not get into any more details. At one point he said, our MO, if you will, has always been to do work and then talk about our work and not to get in front of ourselves. And so we're going to hold on to that as well. But we have got some things that we're incredibly excited about that we'll be talking about later this year. On the conclusion of the call, he said, let me just say I think there's a huge opportunity for Apple with generative AI and with AI without getting into many more details or getting out

ahead of myself. So let's talk about this idea that Apple does stuff first and then talks about it. That is certainly been true. Apple is not usually the first mover unless it's an entirely new category, right? Like the iPhone or the iPad or something like that. Instead, they try to get things right. And so to some extent, people haven't been overly surprised that they haven't leapt headlong into the generative AI race in quite the same way as others have. What's more, Apple has some specific constraints that make it potentially a little bit more challenging. The big tech companies that have invested billions of dollars into the AI race make a ton of money. from their cloud products. Amazon Web Services, Microsoft Azure, Google Cloud. There are direct links

right now from a revenue perspective between their generative AI offerings and those existing cloud offerings. Now, Apple, of course, is primarily a device company, a hardware company. They have a preference for software that can run locally that doesn't need to touch the cloud. As we know, these large language models and other heavyweight generative AI applications have a much harder time running on local hardware. And so part of what it seems like Apple's hesitation has been is to try to, on the one hand, wait for or help AI technology move to a place where it can do more with less, and on the other hand, increase the capacity of that local hardware, which is something we've seen, for example, in the new M3 chips, which are very clearly intended to increase Apple device

capacity to run larger models. Now, it's not like we haven't gotten any indications that Apple is doing anything. Back in September, for example, the information wrote a large piece called Apple Boost Spending to develop conversational AI. In that piece, we learned that the budget for training Apple's most advanced AI models had grown to millions of dollars per day, that there were a number of small teams working on generative AI inside the company, and that they were testing a model that they called Ajax, that they believed had capacities at or exceeding GBT3.5. More recently, we've started to get hints at how Apple might be thinking about commercializing its AI. Last year, Bloomberg's Apple tracker Mark German wrote a piece in October called Inside Apple's

Big Plan to bring generative AI to all its devices. Now, that reporting did a couple things. First of all, it did suggest that Apple was caught off guard by how fast AI hit the public consciousness and that it was considered an internal challenge. An Apple insider told German, there's a lot of anxiety about this and it's considered a pretty big miss internally. Now, when it came to what that report said about what Apple was actually doing with generative AI, the message that we got was that Apple was pushing to, quote, add AI to as many apps as possible.

That included music apps, productivity apps, and more. Now, at the time, there was another debate, just as we thought on how to actually deploy generative AI. Could they do it completely on device? Did it require the cloud, or could there be an in-between approach? Now, even before Tim Cook's recent comments, people had expected it to be iOS 18, where we first really start to see AI coming to the iPhone in a big way. Obviously, we've already seen some big updates in the AI. iPhone wars from Google with their pixel and Samsung with their new Galaxy Ultra 24, and 9 to 5 Mac recently found code in the first beta of iOS 17.4, where it appeared to indicate that they were working on a new version of Siri that was powered by LLMs. This is at once the most obvious

and most expected feature, but also one people would welcome, given how Siri has for many underperformed. Interestingly, the code showed that OpenAI's ChatGBTBT API had been used for internal testing for a lot of these features. 9 to 5 Mac writes, in total, iOS 770s, 17.4 code suggests Apple is testing four different AI models. This includes Apple's internal model Ajax, of which there are two versions, including one that is processed on device and one that is not. Other models referenced by iOS 17.4, including ChatGPT, as well as Flon T5. And so that brings us back to Cook's comments. Now, what was so interesting to me is that on the one hand, he did reinforce this idea that Apple's brand is not to talk about things before they happen.

But this whole conversation amounted to an announcement of an announcement. He's said, we're going to do the work and then talk about the work, but they're literally talking about the work. They're hyping it up. The question, of course, is whether that is just a concession to the incredible pressure on Wall Street to at least indicating that they're doing something in this space. In either case, it will have the effect of ramping up already very high expectations, so Apple has its work cut out for it, but hey, it's using the words generative AI for the first time. That's going to do it for today's AI breakdown. Appreciate your listening or watching, and until next time, peace.

Thank you.