Tech Brew Ride Home - Tue. 10/01 – Running This Pod Through NotebookLM

Episode Date: October 1, 2024

Do we have the first IPO of the AI era? Do we have the first AI model beyond the transformer architecture? Microsoft has a bunch of new AI tools inside Windows. We try to explain that whole controversy around PearAI. And what about that NotebookLM feature that lets you create a two-hander podcast out of any text. Links: AI chipmaker Cerebras files for IPO to take on Nvidia (CNBC) MIT spinoff Liquid debuts non-transformer AI models and they’re already state-of-the-art (VentureBeat) Microsoft Copilot can now read your screen, think deeply, and speak aloud to you (TechCrunch) Oura Nears $500 Million in Annual Revenue and Readies New Ring (Bloomberg) Y Combinator is being criticized after it backed an AI startup that admits it basically cloned another AI startup (TechCrunch) NotebookLM’s automatically generated podcasts are surprisingly effective (Simon Willison's Blog) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the TechMeme right home for Tuesday, October 1st, 2024. I'm Brian McCullough today. Do we have the first IPO of the AI era? Do we have the first AI model beyond the transformer architecture? Microsoft has a bunch of new AI tools inside Windows. I try to explain that whole controversy around Pair AI. And what about that notebook LM feature that lets you create a two-hander podcast out of any text? You're going to want to listen to the end because, well, of course, I had to try it out. Here's what you missed today in the world of tech. AI chip startup cerebrus systems has filed for a US IPO with plans to trade on the NASDAQ with the ticker symbol CBRS. Also, it reported $66.6 million of a net loss on $136.4 million in sales in the first half of 2024.

Now, I'm suggesting this is the first IPO of the current AI era because I couldn't think of anybody else that this would apply to. But feel free to correct me if I just am wrong and not. remembering. Quoting CnBC. Cerebrus competes with NVIDIA whose graphics processing units are the industry's choice for training and running AI models. Cerebrus says on its website that its WSZ3 chip comes with more cores and memory than NVIDIA's popular H-100. It's also a physically larger chip. In addition to selling chips, Cerebus offers cloud-based services that rely on its own computing clusters. Operating expenses have increased this year, in part because of higher personnel costs to support revenue growth, the company said. In addition to NVIDIA, Cerebrus cites AMD, Intel, Microsoft,

and Google as competitors, quote, as well as internally developed custom application-specific integrated circuits and a variety of private companies, end quote. TSM makes the Cerebris chips. Cerebris warned investors that any possible supply chain disruptions may hurt the company. Cerebris was founded in 2016 and is based in Sunnyvale, California. Andrew Feldman, the startup's co-founder and CEO, sold server startup. C Micro to AMD for $355 million in 2012. The company said in 2021 that it was valued at over $4 billion in a $250 million funding round. In May, G42 committed to purchasing $1.43 billion in orders from Cerebrus before March 2025, according to the filing. G42 currently owns under 5% of Cerebris's Class A shares, and the firm has an option to purchase more depending on how much Cerebrus product it buys, end quote. Well, that's interesting because in the filing to go public, Cerebrus, as you heard, had to reveal that this is UAE-based G-42 accounted for 87% of Cerebris' first half 2024 revenue and has committed to

buying $335 million worth of its stock by April 2025. That would give, as mentioned, G-42 greater than a 5% stake in the company. So is that a good sign? One customer is almost 90% of Cerebus' revenue, and they're committing to buying their stock at IPO to keep it propped up. For a while now, we've been wondering if something was going to come along after the Transformer. Most of the LLMs that we talk about on the show are built using the same architecture. The Transformer model outlined in the Seminole 2017 paper attention is all you need. But is there something beyond that? Well, MIT spin-off Liquid AI has debuted its non-transformer AI models

LFM 1B, LFM3B, and LFM40B MOE, claiming they achieve, quote, state-of-the-art performance at every scale. Quoting Venture Beat. Liquid states that its goal, quote, is to explore ways to build foundation models beyond generative pre-trained transformers or GPTs, and with these new LFMs, specifically building from, quote, first principles, the same way engineers built engines, cars, and airplanes. It seems they've done just that, as the new LFMs,

models already boasts superior performance to other transformer-based ones of comparable size, such as Meta's Lama 3.18B, and Microsoft's Phi3.5-3.8B. Maxime Labone, head of post-training at Liquid AI, took to his account on X to say the LFMs were, quote, the proudest release of my career, and to clarify that the core advantage of LFMs is their ability to outperform transformer-based models while using significantly less memory. The models are engineered to be competitive not only on raw performance, benchmarks, but also in terms of operational efficiency, making them ideal for a variety of use cases from enterprise-level applications, specifically in the fields of financial services,

biotechnology, and consumer electronics, to deployment on edge devices. However, importantly for prospective users and customers, the models are not open source. Instead, users will need to access them through Liquid's inference playground, Lambda chat, or perplexity AI. Liquid says it used a blend of, quote, computational units deeply rooted in the theory of dynamical systems, signal processing, and numerical linear algebra, and that the result is, quote, general purpose AI models that can be used to model any kind of sequential data, including video, audio, text, time series, and signals to train its new LFMs. Last year, Venture Beat covered more about Liquid's approach to training post-transformer AI models,

noting at the time that it was using liquid neural neural. networks, or LNNs, an architecture developer at C-Sail that seeks to make the artificial neurons or nodes for transforming data more efficient and adaptable. Unlike traditional deep learning models, which require thousands of neurons to perform complex tasks, LNNs demonstrated that fewer neurons, combined with innovative mathematical formulations, could achieve the same results. Liquid AI's new models retain the core benefits of this adaptability, allowing for real-time adjustments during inference without the computational overhead associated with.

with traditional models, handling up to 1 million tokens efficiently while keeping memory usage to a minimum. While other models experience a sharp increase in memory usage for long context processing, LFM3B maintains a significantly smaller footprint, making it highly suitable for applications requiring large volumes of sequential data processing, such as document analysis or chatbots. Liquid AI has built its foundation models to be versatile across multiple data modalities, including audio, video, and text. With this multimodal capability, Liquid aimed.

to address a wide range of industry-specific challenges from financial services to biotechnology and consumer electronics, end quote. New AI stuff you can use right now, even if you're not an AI developer, Microsoft has unveiled copilot vision to analyze text and images on web pages. Also, think deeper to reason through more complex problems, and also co-pilot voice, quoting TechCrunch. Co-pilot vision has a view of what you're viewing on your PC, more specifically a lens into the sites your visiting with Microsoft Edge. Gated behind Copilot Labs, a new Copilot Pro exclusive opt-in program for experimental co-pilot capabilities. Copilot Vision can analyze text and images on web pages and answer queries, e.g., what's the recipe for the food in this picture, about them. Vision, which can be

pulled up by typing at Copilot in Edge's address bar isn't exactly a technical Marvel. Google offers similar search technology on Android and recently brought bits and pieces of that tech to Chrome as well. But Microsoft suggests that Copilot Vision is more powerful and conscious of privacy than previous screen analyzing features. Co-Pilot Vision can suggest next steps, answer questions, help navigate whatever it is you want to do and assist with tasks all while you simply speak to it in natural language. Microsoft wrote in a blog post shared with TechCrunch. Imagine you're trying to furnish a new apartment. Copilot Vision can help you search for furniture, find the right color palette, think through your options on everything from rugs to throws, and even suggest ways

of arranging what you're looking at. No doubt, eager to avoid another round of bad press from AI privacy fumbles, Microsoft is stressing that Copilot Vision was designed to delete data immediately following conversations. Processed audio, images, or texts aren't stored or used to train models of the company claims, at least not in this preview version. Copilot vision is also limited in the types of websites it can interpret. For the time being, Microsoft's blocking the feature from working on paywalled and sensitive content, limiting Vision to a pre-approved list of popular web properties. What does sensitive content entail exactly? Porn? Violence at this juncture, Microsoft wouldn't say. As with Vision, co-pilot's new Think Deeper feature is an attempt to make

Microsoft's assistant more versatile. Think Deeper gives Copilot the ability to reason through more complex problems, Microsoft said, thanks to reasoning models that take more time before responding with step-by-step answers. which reasoning models? Microsoft was a bit cagey again when I asked, saying only that Think Deeper uses, quote, the latest models from OpenAI fine-tuned by Microsoft. Reading between the lines, it's safe to say they're a customized version of OpenAI's 01 model. We've designed Think Deeper to be helpful for all kinds of practical everyday challenges, like comparing two complex options side-by-side, Microsoft wrote in a blog post. Think Deeper can help with anything from solving tough math problems to weighing up the cost of managing

home projects, end quote. And a new copilot feature generally available today as well is co-pilot voice. Not to be confused with GitHub's co-pilot voice, launching in English in New Zealand, Canada, Australia, the UK, and the U.S. to start, voice adds for synthetic voices letting you talk to co-pilot and have its responses be spoken aloud, end quote. Oura says it expects annual sales to double to roughly $500 million this year, 2024. Sources say Aura plans a new thinner fitness ring with better battery life, maybe coming as soon as this month, which I would say they should work on this fast because I feel like more competition is coming to the space Tutsweet. Quoting Bloomberg, Chief Executive Officer Tom Hale, speaking in an interview,

said that ORA is building a loyal following after selling more than 2.5 million rings. Still, the company isn't yet at the stage of planning an initial public offering, he said. The 11-year-old business, which pioneered the concept of finger-worn activity trackers, makes its money by selling rings for $299 and subscriptions priced at $6 a month. It's more of a niche market than smartwatches or earbuds, but the field is getting more crowded. Samsung recently launched a $400 product called the Galaxy Ring. Hale is upbeat about expanding the business. The company's profit margins are closer to that of a software company than a hardware maker, he said,

and or as subscribers have been sticking with the product. retention is better than any other subscription model I've seen, Hale said. To double this business, we don't have to do that much. He said that the company's roughly half a billion dollars in revenue for calendar 2024 would be twice what it recorded in 2023. Though or a decline to discuss future products, people with knowledge of its plans say the company is introducing a fourth generation ring in October. The device will have a thinner design and better battery life, as well as more accurate

activity tracking, they said. It's set to be the company's biggest product overhaul. in three years, end quote. I used to do this more. I used to provide explainers of controversies happening online in tech circles, even if they weren't exactly newsworthy or timely. So, from the controversy explainer file, here's why some folks are criticizing why Combinator-backed startup pair AI for forking AI code editor continue, which has the Apache open source license, but then they slapped a chat GPT-written closed license on top of that. Quoting TechCrunch, Pair AI offers an AI coding editor.

The startup's founder, Duke Pan, has openly said that it's a cloned copy of another AI editor called Continue, which was covered under the Apache Open Source license. But Pair AI made a major misstep. Pear AI originally slapped its own made-up closed license on it called Pair Enterprise License, which Pan admitted was written by Chat-GPT. Changing a license like this is a big deal in the open source world. Not only are there legalities involved in violating a software license, but it defeats the whole purpose of open source, which is about community building, sharing, and contributing. In an apology, Pair AI's pan posted on

Monday, he said that the project has now been released under the same Apache Open Source license as the original project. The launch thread went viral with thousands of comments by Sunday. Some were congratulatory, but others were vicious in pointing out the licensing and the fact that Pair AI wasn't so much a fork with new stuff added, but a replica with a new name. Pan admitted as much in his apology. So many angry comments were made on PAN's original thread that X put a community note on it that read, Pair is a fork of continue. Dev, an open-source AI code editor. Pair AI used continue. Dev's code and mass replaced all references to continue to pair AI to mislead people into believing that they built this product on their own, end quote.

This note wasn't accurate either, though. Pair AI did say in some of its materials that the project was a clone, also known as a fork of Continue, continue, as well as the original project that continue used, VS code. X subsequently removed that note. Pan apologized for how hard it was to find that information, too. He said that one way he and his co-founder, Nangang, quote, screwed up, critically, was not being clear enough about this, doing so upon a fork of others' work without many new features, and, talking about it so publicly online made it look like we were stealing the work of others as our own, end quote. On Sunday, continue jumped in by posting a subtle threat that it was, quote, ecstatic to see the ecosystem that has formed around us, but open source can't be taken for granted. It's a movement built on trust and on respect for contributions, licenses, and intellectual property, end quote. Why Combinator CEO Gary Tan got involved too, he defended Pair AI with several tweets, quote, don't understand why people are dragging a new project when literally it's open source Apache license, and that's the reason why open source is awesome, one read.

As you might imagine, people pointed out that it was changed to an Apache license after the uproar. There were other reasons this project caught Eyre Pan boasted how he, quote, just quit my $270,000 job at Coinbase to do this startup, even though this was about as far from an original idea as a startup can get. In addition to continue, another big competitor is Cursor. On top of that, YC has funded two other AI co-execor. code editors already, void and melty, as the mob was quick to point out. To which Tan replied on X, more choice is good, people building is good, if you don't like it, don't use it, end quote.

Others, though, criticize YC for selecting Pair AI in its cohort at all. Blogger Sven Schneiders wrote that Pair AI is an example of the, quote, the decline of YC because it accepted a company that is nothing more than a code base copied from another YC-backed company, end quote. On Hacker News, the site for programmers owned by YC, a commenter wrote that the debacle quote says more about YC than this particular founder, lots of these types nowadays, i.e., their process, their due diligence. Another wrote, is it typical for VC to just throw money at projects without any sort of oversight auditing of, or, geez, I don't know, licensing legal issues, end quote. Finally, today, you might have seen the post going around about Google's Notebook LM and its

audio overview feature, which can generate a podcast with two AI hosts after you upload, say, your notes or links to a bunch of stuff. On his blog, Simon Willison had a breakdown, quote, notebook LM is effectively an end-user customizable rag product. It lets you gather together multiple sources, documents, pasted text, links to web pages and YouTube videos into a single interface where you can then use chat to ask questions of them. Under the hood, it's powered by their long-context Gemini 1.5 Pro LLM. Once you've loaded in some sources, the notebook guide menu provides an option to create an audio overview. A key clue to why audio overview sounds so good looks to be Soundstorm, a Google Research project which can take a script and a short audio

example of two different voices, and turn that into an engaging full audio conversation, end quote. Now, especially interesting was the post going around over the weekend where somebody created a notebook LM podcast where the hosts discover they're not real. Quoting from that, this is host number one. I tried. I tried calling my wife, you know, after they told us, I just, I needed to hear her voice to know that she was real. Host number two. What happened? Host number one, the number. It wasn't even real. And there was no one at the other end. It was like she, she never existed, end quote. Okay, obviously, I had to give this a try. So, after a quick additional commercial break, I'm going to play for you what Notebook LM did with my script for today's episode.

Note that I did not feed it this part, of course. I didn't want to overly confuse it. So what I'm saying now, it didn't hear or read or whatever. Also, I took out the things like when I say quoting CNBC or whatever. I don't think leaving that out is putting my finger on the scale overly, but I don't know. Maybe I'll run it through with all of everything, too, just to see what happens. Or maybe tomorrow I'll just feed it the MP3 file of tomorrow's episode. But anyway, listen to this.

Listen to how it picked a theme out of what I fed it and basically alighted over some of the stuff I just spoke about, but then added things that I didn't. Remember, all I fed it was my script for today's show. I just cut and pasted the text. Anyway, listen to this one ad and then listen to what AI produced with the content you just listen to. AI, right? It's everywhere you look these days. Every day there's some new headlines, some new breakthrough. It can be tough to keep up, you know? It really is moving at an incredible pace. It is. So today we're doing a deep dive. Cutting through the hype. We're looking at the TechMeme Ride Home podcast from October 1st.

And they had some really interesting AI stories. We're talking chips. We're talking models. the whole shebang. It's a great episode. Lots to unpack. For sure. Let's start with Cerebra Systems. They're taking on the big dog, Nvidia, right in the AI chip market.

They just filed for an IPO, and things are looking up for them. But, and there's always a, but there's this one thing that kind of jumped out at me. Yeah, their IPO filing. It shows that almost all of their revenue, like close to 90%, it comes from a single customer, G42. G42.

They're based in the UAE, specializing in AI and cloud computing. Okay. Okay, so one customer makes up basically their entire business right now. Is that a good thing or a bad thing? I'm not sure how to feel about that. It's a double-edged sword, isn't it? Mm-hmm.

On one hand, relying so heavily on one buyer, it's risky, right? If that relationship goes south, so does a huge chunk of their revenue. But on the flip side, the fact that G-42 is in so deep, it could be a sign that they really believe in Cerebra's technology. And not just believing, they're putting their money where their mouth is, committing to buy $335 million of Cerebris' $3. stock on top of their product orders. That's a big deal. $335 million. Wow. So is a sign of things to come? Are we going to see more specialized AI

companies looking for custom-made chips instead of just going with the big players like NVIDIA? It's definitely possible. G42's approach could be a signal that we're moving away from this one-size-fits-all model for AI hardware. And that could lead to a whole new kind of arms race. An AI chip arms race. Now that's a headline. Okay, but moving on from the hardware, Let's talk about the brains of the operation, the software, the models, all of that. And that leads us to Liquid AI, am I right? They're saying they've got something even better than the Transformer model, which, let's be honest, is pretty much running the show in AI right now. Right. The Transformer model is behind a lot of the AI we see today.

But Liquid AI, they're claiming their LFM series. It's not just different. It's more efficient and more powerful at the same size compared to what we're used to. Hold on, hold on. More efficient and more powerful. How is that even possible? Well, it comes down to their approach. It's kind of radical, actually. Traditional transformer models, they use tons of neurons, thousands of them.

Liquid AI streamlined it using fewer neurons, but they made them much more adaptable. They're taking cues from the human brain, emphasizing flexibility and dynamic processing. So fewer but smarter neurons, kind of like a small team of elite athletes versus a huge team of, well, less experienced players. Exactly. And that allowed their models to handle much more data with less computational. computational overhead. We're talking processing up to a million tokens. A million. Okay, now I need the AI to explain that. What does a million tokens even mean in the real world? Think of it this way. You could analyze an entire novel in one go.

We're talking about AI potentially understanding and manipulating language at a whole other level. Imagine more accurate translations, better text summarization, even AI generated content that's actually, you know, good. Okay, I'm getting a little bit of a black mirror vibe here, but I'll bite. This all sounds, pretty revolutionary, but there has to be a catch, right? You're catching on. Here's the thing. Liquid AI's models. They're not open source, unlike a lot of AI models out there.

So they're keeping it under wraps, proprietary and all that. Exactly. It's a gamble. Will developers and companies pay a premium for access to potentially game-changing technology? We'll have to wait and see. But it's going to be interesting, that's for sure. It's like choosing between the familiar, well-trodden path and this mysterious, potentially amazing but locked door. Speaking of different approaches, let's talk about Microsoft. They're not

reinventing the wheel, but their latest Windows updates, they're all about AI integration. Right. They're weaving AI into the fabric of the operating system, making it more accessible, more user-friendly. And sometimes those subtle changes, they can have the biggest impact. Let's talk about this co-pilot vision thing. It sounds straight out of a sci-fi film. It is, doesn't it? Your AI assistant can actually see what's on your screen now. It analyzes images directly in the edge browser and answers questions about them. Wait, so if I'm looking at a picture of, I don't know, a flower online and I want to know what kind it is, I can ask co-pilot. Exactly.

Or imagine being on a video call and there's this cool piece of art behind your colleague. Co-pilot could analyze it, tell you the artist, the title, maybe even some historical context. Okay, that's seriously cool. But also a little creepy. AI looking at my screen. There have to be some privacy concerns there, right? You're absolutely right. Microsoft's going to have to be super transparent about how they use this data

and make sure users have control over their privacy settings. But the potential here in terms of accessibility and just getting information quickly, it's huge. Yeah, it feels like a whole new way of interacting with our computers. And speaking of pushing boundaries, Microsoft is trying to make co-pilot even smarter with this think-deeper feature. What's the deal with that? So they're basically boosting co-pilot's ability to solve problems. and, well, think. Instead of just reacting to command,

they're aiming for a back and forth. Imagine asking co-pilot to plan your next vacation. That's ambitious. With think deeper. It could break down the entire process. It could research destinations, find flights, suggest activities, and then hand you a personalized itinerary.

So it's less of a search engine and more of a, I don't know, a thought partner. Exactly. And as AI becomes more and more integrated into our lives, those are the kind of features that will make a real difference, making it feel seamless, intuitive. It's pretty mind-blowing when you think about it, but it also makes you think about the ethical side of things, like with that pure-AI controversy,

we'll get into that after the break, because as AI gets more sophisticated, we have to make sure we're developing it responsibly. There's a lot to consider. So before the break, we were talking about Microsoft and their approach to AI. But let's shift gears for a sec. The pure-AI thing, it's a bit of a cautionary tale,

right, especially when we're talking about open source and AI development. Yeah, it really brings up some important points about ethics and transparency. Yeah. To recap, you had pure AI. They were working on this AI code editor. The problem was it was basically a clone of another open source project, continue. And to make matters worse, they slapped their own closed source license on it. So they took something that was freely available, copied it, and then tried to lock it down.

Pretty much. On the developer community, they were not happy about it. I can imagine. And Y Combinator, they're a big deal in the startup world. They backed pure AI. So this wasn't some small time operation. It makes you wonder, what were they thinking? Well, it just goes to show.

Even with all the excitement around AI, we can't forget the importance of ethics, especially in a field that's all about collaboration and open innovation. It's like the old saying, move fast and break things. But maybe in this case, they should have been a little more careful about what they were breaking. Right, because trust and reputation, they're everything. And once those are gone, it's tough to get them back. For sure. Okay, let's lighten things up a bit.

We talked about AI chips, AI models, all that. But what about AI that can create a podcast? Like a whole podcast complete with different hosts just from your notes. Crazy, right? You're talking about Google's notebook L.M. It's pretty mind-blowing what it can do. So it takes your ideas, your written stuff, and turns it into a full-blown podcast episode with different voices and everything.

Yeah. And it uses Google's. soundstorm technology. That's what makes it sound so realistic. It captures all the nuances of human speech, the pauses, the inflections, even the ums and ahs. It's a bit eerie, isn't it, how good it's getting. There's this blogger Simon Willisand, he used Notebook L.M to make this podcast where the AI hosts, they become self-aware. Yeah, I saw that. It was pretty wild. They start questioning their own existence. It's a funny thought experiment, but it also makes you think, are we really that far off from AI that can truly mimic human consciousness?

Okay, now that's a whole other deep dive for another day. But speaking of Microsoft, remember we were talking about their approach to AI? They're also doing some interesting stuff with making AI more integrated into our everyday lives. Right, like with Windows. It's all about making AI more accessible, more user-friendly. Yeah, and they have this new feature. Think deeper. It's part of their co-pilot.

Apparently, it makes co-pilot a lot more, well, thoughtful. It's a step up from just giving commands. With think deeper, co-pilot can actually help you solve complex problems. Can you give me an example? Sure. Imagine you're planning a trip. You could ask co-pilot, not just to book a flight, but to help you plan the entire trip, considering your budget, what you're interested in, all of that. So it's like having a personal travel agent right there on your computer.

Exactly. And it's not just limited to travel. They're aiming to make think deeper. work across all range of tasks, from writing to brainstorming. It's pretty wild. It's like we're at this point where AI is becoming less of a separate thing and more of an extension of our own capabilities. It's true. But, and I think this is important to remember, we need to make sure we're developing AI responsibly. We've talked about the ethical dilemmas. Transparency is crucial.

Absolutely. We need to understand how AI is being used, how our data is being collected, and we need to make sure we have a say in all of it. It's about finding the right balance between innovation and responsibility. Well said. Now, let's talk about something that's both incredible and a little bit unsettling. Remember how we were discussing Google's notebook, LM, and its ability to generate podcasts from text? Well, Google's also been doing some mind-blowing work with AI and music. Ah, you mean music, LM.

It's incredible what they've achieved. It's an AI that can compose music, real music, and not just simple melodies. We're talking complex compositions, different styles. styles, the whole nine yards. It's remarkable. You can give it a text prompt, like create a piece of music for a film noir scene, and it'll generate something that perfectly captures that mood. It's like having an orchestra at your fingertips. But it also begs the question, what does this mean for musicians, for composers?

It's disruptive, that's for sure. It challenges our traditional notions of creativity and authorship, but I also think it presents incredible opportunities for collaboration and exploration. Imagine musicians using AI as a tool to break through creative blocks or explore new sonic territories. It's like having a creative partner that can tap into an infinite well of musical ideas. Exactly. And that's what's so excited about AI. It has the potential to unlock creative possibilities we never even imagine. But it also comes with a responsibility to use it ethically and thoughtfully. And that's something we need to keep in mind as we move forward because the future of

AI, it's not something that's being decided for us. It's something we're all shaping together. It's amazing to think about all the different ways AI is popping up, like we were talking about before, Google's doing some seriously cool things with AI and music. Yeah, Music L.M. is a game changer. It really is. I mean, an AI that can compose music. It sounds like something out of a movie. I know, right? But it's real, and it's already creating some pretty incredible stuff. It makes you wonder, what does it mean to be a musician or a composer in a world where AI can just whip up a symphony in minutes? It's a valid question.

But instead of viewing it as a threat, I think it's more helpful to see it as an opportunity. Like a new kind of instrument, almost. Exactly. One that has the potential to push the boundaries of music and art in ways we can't even imagine yet. And that's what's so exciting about AI in general, right? It's not just about making our lives easier or automating tasks. It's about expanding our creative horizons. Absolutely.

It's about augmenting our own abilities and exploring new frontiers. It's like we're entering this undertaking. charted territory and AI is our compass and our map all rolled into one. And the possibilities, they're pretty much endless. Yeah. But like we've been talking out throughout this whole deep dive, we need to make sure we're holding onto the compass and the map with a sense of responsibility. You're right.

Because AI is a powerful tool. And like any tool, it can be used for good or for bad. It all comes down to the choices we make, the values we prioritize, and the future we envision. And that future, it's not something that's going to be decided. for us. It's something we all have a hand in shaping. I couldn't agree more. It's a collective effort, a global conversation that needs to include everyone. Because the impact of AI, it's going to be felt by everyone, everywhere. Precisely. And it's already happening, whether we realize it or not. So let's make sure we're paying attention. Let's stay informed, stay engaged, and most importantly,

stay curious. Curiosity. That's what drives progress. That's what pushes us to ask the tough questions and seek out new solutions. And that's what will ultimately lead us to a future where AI is a force for good in the world. A future where technology serves humanity, not the other way around. Well said. Well, that about wraps up our AI deep dive for today. We covered a lot of ground from the nuts and bolts of AI chips to the philosophical

implications of AI generated art and music. It's been quite a journey, wouldn't you say? It has. And it's just the beginning. I can't wait to see what the future holds. Me neither. Thanks for joining us on this event. and we'll see you next time for another deep dive into the world of tech and innovation.