Loading summary
Transcript1 lines
- [00:00]
A
Welcome to the podcast. Today on the show, the president of Uber says that AI spending is getting harder to justify. I want to break down some of the numbers of how much they're actually spending and the ROI people are expecting and what they're actually getting. At the same time, we have Pope Leo who released this entire AI document and it's kind of targeting tech power concentration. I'm going to talk about some of the things that are mentioned in there, some things that people like and don't like. About 85% of enterprises, according to a new report, want agentic AI, but 76% of them lack the infrastructure to actually deploy it. Sundar Pichai says that Google is currently reshaping search and web with their new AI agents. I want to get into some of these trends because Google has the biggest distribution of any software company in the world. And as they've reached a billion monthly active users on Google Search and Gemini specifically, I believe Google AI search, they have a massive impact on the entire AI market. Universal Music Group and TikTok are renewing a license deal that is going to help fight against unauthorized AI music. There's actually a ton of AI music news going on with Spotify. They released that. They have a deal with Spotify or for. With Universal for licensing so that people can do covers, AI generated covers and remixes. There's a ton going on. We're going to get into all of it. If you want to get a daily breakdown of all of these stories and more as a daily newsletter, go check out my website, aichatdaily.com. you can hit the subscribe tab in the button and I think we'll even have a little link in the description of this show. But we have a newsletter that I send out every single day which covers all of these stories. We send them straight to your inbox. It's all free. And you get a breakdown of all of these stories, links to more and everything going on in the AI industry. If you want to keep up to date on all that's going on, more than you're going to see in the sidebar of LinkedIn where it gives you the news breakdown or wherever else you get your news. All right, let's talk about what's going on with Uber. Their president has said that AI spending is getting, quote, harder to justify. Apparently Uber burned through its entire annual AI budget and they did all of this in four months, the first four months of this year. Now they're saying that they can't actually prove the spending delivered results. Now, I like A big caveat on this whole story. I talk to so many people that are, you know, doing some incredible things. I for one have a ton of projects underway right now. A ton of things that I have launched. You know, One example is aichatdaily.com that I mentioned. I published like 10 articles, sometimes 20 articles a day, doing deep dive breakdowns. Every single story in AI news. This is a new project I've done. There's no way I'd be able to do this if I wasn't actually vibe coding. And, you know, I have tons of other projects like that where AI code has done wonders for everything I've done and has saved me so much time and money. When you get to a big, huge company, sometimes I think that all of the incentives can be a little bit off. We heard from Amazon in particular, they had some sort of AI initiatives and they needed developers to be writing a certain percentage of their code in AI. And so what they started doing if they weren't hitting the benchmarks is they were just like running bots that would just spit out random code for random small unimportant things so they could get their co, their code token usage up, which of course I'm sure Anthropic loved. But it, you know, just costing Amazon no money. And they're like, look, we're hitting all of our, our metrics of what percentage of our code needs to be AI generated. This is exactly the problem I think we're seeing in a lot of companies. If they, if they have, you know, bad incentives, the incentive shouldn't be 75% of your code needs to be written by AI. The incentive needs to be, you know, we're trying to actually make something, get something done. Here's how, how you should do it better and faster. And I think there's a big education piece, there's a big mindset, mindset shift that needs to happen. So with all of that being said, Uber right now is struggling to see what some of their benefits are. And I think a lot of this is a transparency crisis. A lot of this is probably things that other enterprises are seeing. And I think a lot might be they're just not measuring the right things. But let's break down some of the numbers. So uber apparently spent $3.4 billion in R&D in, that's up 9% year over year. And Claude, code tokens costs have been climbing a ton. The executives are saying that this is astronomical trajectory. Basically what they're spending right now on their tokens is keeps going up, not because the cost from Anthropic per se is going up, but because, you know, presumably they're, all of their developers are using it more and more right now. They said that they're kind of framing their AI spending as a direct trade off against hiring. So their CEO said that Uber is hiring fewer humans to offset growing AI investment costs costs. McDonald acknowledged a 25% increase in consumer features from Uber, but admitted that they were drawing a line between token usage and actual deliverables. It was really hard for them, they say, quote, impossible to measure. The reason why I think this is important as far as a conversation goes, Uber is kind of weaponizing the CFO's skepticism across all enterprise AI markets. They've kind of publicly said, like, look, we can't really measure the ROI on our $3.4 billion in AI that we spent year or R and D that we spent last year. And so now I think we're going to see a lot of people that are going to be saying, hey, look, like it's same thing in our company as well. Now I think the, the truth of the matter is they met, they, you know, they admitted it themselves. They're hiring less people to spend more on AI, AKA they're probably getting many more things done with the AI and they just aren't hiring people to do it. Also, I think a lot of projects can take time before getting shipped, so there's some sort of lag there. I would, I would put into account I have a bunch of projects underway. Some projects I finished, some projects are ongoing. And you know, in no means is are my projects as big as Uber, but I do think that there's an important note to be made that you don't always get the benefit of a feature or development work immediately after it happens. However, a year later, I think they should have a better way of actually measuring it. So we'll see where Uber falls in all of this. Pope Leo has just put out a 200 page document and my friend Connor Grennan, who I host the AI Applied podcast with, had a great breakdown of this over on LinkedIn that I'll be using. But by the way, if you're not following the AI Applied podcast, that's another one I would definitely recommend we drop three episodes every single week breaking down how to apply AI into your career. And for enterprise. The first thing that Connor mentioned he got out of this document from the Pope is that it wasn't anti tech and it wasn't saying AI was inherently evil. He kind of frames A type technology as a milestone of human ingenuity. He mentions how it can uplift humanity. And of course, that's assuming we actually give it good ethics and share responsibility with it. The other thing I thought was interesting, the next thing he mentions is that the whole argument is kind of posing AI as not a technical challenge, but an anthropological one. The real question is how humans are going to actually respond to that. Something else that I think was pretty evident in all of this is that it was definitely pro worker and not anti AI. The Pope warns against systems that deskill or surveil employees. He calls mass unemployment from AI, quote, a true social calamity. But I think the framing is also about protecting people. It's not really about blocking progress, right? So it's not saying like, hey, we have to stop AI. It's just like, hey, we need to be aware of the capabilities and make sure that bad outcomes don't happen. They also mentioned that AI is an equalizer. He envisions AI helping developing nations with education, economic development, preserving local cultures and languages. So I think that, you know, to be fair, there was a lot of positives I brought up rather than just, you know, this is going to deepen the gap between the rich and the poor and that kind of stuff. So I do think that, you know, it has the ability to be an AI, to use AI as an equalizer. And the Pope kind of outlined that. The last thing I want to mention is that when he put all of this out, I was actually impressed. He, he showed up with an AI leader. So definitely not a traditional move. But he presented this whole document alongside Chris Ola, who is one of the co founders of Anthropic. And so I think this is, this is a good point, right? Anthropic is evidently working with the Vatican, and the Vatican is working with Anthropic. And why wouldn't they want to have a seat at the table when a lot of this big technology is being rolled out to try to put positive ethics, positive morals, positive insights into what's being rolled out and make sure that this is as beneficial for humanity as possible. So I think that there was tons of nuance in all of this. I like the framing. When you look at most of the news coverage about this, it's like, it's saying things like, Pope Leo compares AI threat to biblical Tower of Babel, Pope warns, blah, blah, blah. It's like, you know, it's all this stuff like Pope warns, like saying all of the, the, the bad side of it. But he, he Put a ton of stuff in there that was also, you know, showing AI as an equalizer and a lot of the, the positives in all this. So definitely important to get the message from both sides of that story. Okay, the next thing I want to talk about is a new report that just came out that said Most enterprises want AI. And it says that specifically 85% of enterprise is plan to adopt it by 2029, but 76% lack the infrastructure to actually deploy it. And the problem, according to this report, is that companies are bolting AI agents onto legacy systems. So, you know, they already have their old system, their old way of doing everything. They're just sticking chatgpt on top of it. They're not actually redesigning their entire organization. They're not looking the tech stack, they're not looking about at kind of how the workforce is using these tools. They're just sticking a chatbot on top and it's not being, you know, nearly as effective when, when you think of how you're probably using AI. People that are listening to this show, like we're all using Cloud Cowork or, or you know, Codex from Chat GPT. We're using this to take control of our computer, to, to fill things out, to, to build code, to, to make websites, to help automate so many things that are happening inside of our roles in our organization. This is not what most enterprises are doing. Most people are. And bolting things on top. So in the report it says that agent develop agents deployed at scale accelerate processes by 30 to 50%. I'm not shocked by that. I've seen those numbers much higher. It says that they cut low value work time by 25 to 40% again. Right. Like all sorts of things I just used to dread like writing titles and descriptions for my podcasts. Day one when ChatGPT came out, that was just not something I ever had to worry about again. But it also says that McKinsey is projecting 3/4 of current jobs are going to need redesign, upskill or redeployment by 2030. Because AI agents are going to basically merge the hierarchical management lines and it's going to eliminate a lot of execution based roles. Which is really interesting. Right? I mean, I can even think of my first job out of college. I was working as a marketing manager in an organization. I was sending the weekly email which I had to go get proofread by, you know, the, by by the, the team that worked on all of our publications. And I had to go and there's just a whole list of things I had to do, I had to copy and paste between spreadsheets and all this kind of stuff that AI100 would just completely automate. Today, when you saw the earlier news in the podcast where we had Uber that was complaining about, you know, hey, look, we're not seeing the ROI from all of our AI spending. I think this is actually a big reason why. I think the winners are going to be the people that actually rebuild from scratch. Everyone else that just bolting AI on top of what they already have, maybe broken systems, old legacy code. They're not going to see the ROI materialize. And so I think you're going to watch what industries are moving first and that's where the real competitive moats are going to start forming. We're already seeing that with a lot of knowledge work and with developers really crushing it and getting a lot more out of it. Okay, Sundar Pichai said that Google is reshaping search and the web with AI agents. What is he talking about? Apparently Google's 13 billion user products now run on unified Gemini infrastructure for the first time. And we actually have recently talking to Logan Kilpatrick from Google on the AI Applied podcast, another shout out to AI Applied and he was mentioning how it's really nice that we're really starting to pull a lot of things together. We see a lot of criticism on on X I've seen a lot of posts saying like, there's so many different Google and Gemini products all over the place with different logins and it's hard to keep them all, keep them all in your mind. And I guess over at Google it was a real problem because they had different versions of Gemini, they had all sorts of different ways that things were being connected. They're actually standardizing a lot more things inside of Google. And on the one hand, yeah, I can see the criticism and it is kind of maybe branding confusing. But at the same time, I think everyone inside of Google, you got to give it to them. They're firing on all cylinders and pumping down as many products, as many integrations as possible. They're seeing what sticks, they're cutting what doesn't. It feels a lot more like a startup than like a massive corporation. I can think of a lot of corporations that are not doing that and are. It feels like they're getting left behind. So if anything, I think we have a whole bunch of random, you know, smash hit runaway successes from Google like Notebook lm, which I never would have thought was going to be like one of Google's top products. But I'm seeing Notebook LM rolled into everything, where you can turn any document into a podcast and listen to it. People absolutely love it. The usage went crazy, it went very viral and it has a lot of sticky factors. So I think Google's kind of playing this game where they're throwing a lot out there and they're seeing what sticks. Apparently everything from personal intelligence, ask maps to search. YouTube and Chrome is all using the same version of Gemini. Now Google is training AI models on YouTube videos and they're also redesigning YouTube search to summarize content and skip users directly to the relevant timestamps. Which personally, I mean, who can complain about having to skip someone on a YouTube video saying, hey, welcome to the channel. If you like this channel, go. Make sure to like and subscribe and yada da da da. Today we're going to talk about xyz and they give the whole like intro to their life. It's basically the equivalent of if you've ever been on Google and you've searched and found a recipe and at the beginning of the recipe you have to read the person's entire life story because they're trying to put more text in there for SEO value and you're just trying to figure out how to scroll down on the recipe page to see the actual recipe at the bottom. That is basically what half of YouTube videos are. And I think it's kind of interesting that for whatever reason, Google made it so that a YouTube video has to be eight minutes long to get monetized. So now everyone's got to make 8 minute long YouTube videos, even if they could have given you the information in two minutes. But so yeah, anyways, it's funny, we make these systems that cause these problems and now when you go to YouTube, the AI is going to help you skip to the exact part that you need to and move on with your life. So yeah, anyways, it's just funny the problems we create for ourself. Conde Nast CEO publicly said that their company is planning for zero search traffic from Google. And Sundar previously dismissed that scenario, but he now says that it is a possible strategy. We're seeing that Google search is taking up more and more or Google AI search is taking up more and more of the results. And you're, you're really, when you search Google now you're just seeing the AI result at the top and yes, you can scroll down and in, in rare long form cases you're, you know, you're going to maybe see some links, but really they're, they're pushing heavily into AI and this is the AI results on Google. And that's what we're seeing more. Okay. Universal Music Group and TikTok just renewed their licensing deal, committing to removing unauthorized AI generated music. And they're also trying to improve artist attribution on the platform. This is interesting. I've actually seen a ton of stories about viral AI generated content and music. I think specifically when it says unauthorized, I, you know, I had a friend recently that tagged a story and on their Instagram story there was like a song playing in the background. I was like, man, I like that song. I clicked on it. Anyways, turns out it's AI generated, which is, which would be fine, except that they. The artist's name on that song was Lana Del Rey. It was obviously not Lana Del Rey. It wasn't on her main account, but it was like its own account. I'm assuming something like that is gonna get swept up and banned and deleted, but you gotta have these like licensing deals between Universal Music and TikTok. I'm not really sure who's paying who in this case, but in any case it's kind of there. This is an agreement that is ending a 2024 kind of spat that UMG, you know, they pulled their entire catalog off of TikTok over AI content moderation failures. So anyways, there's this whole beef and now there's, you know, millions of dollars in lost streams and, and money from creators once they're. Once all of their music got pulled off of TikTok because of UMG. So it's been a whole, it's been a whole thing in the industry and it seems like they're trying to work out a licensing deal to, to make it so that this AI content moderation is actually going to be happening. Drake and the weekend AI generated tracks accumulated millions of streams before those got removed. Anyways, I think we're probably coming to the end of that long Sega because Spotify and Universal have a deal where they're going to have licensing agreements for people that are trying to make AI generated covers and artists can opt in and they're going to get streaming royalties if someone makes a cover of their song or uses their voice to make something. So anyways, I think we're going to get to, to somewhere where you can use some of those famous voices. They will be opted in. You're going to. There's going to be some sort of royalty paid out to them, but it's definitely an interesting shift in the music world. Someone recently left a review on Apple that says Leaving a review on Apple Podcast sucks. I listen here, but leaving a review is totally unintuitive. It's so I actually had to use the help to figure out where to do this do better Apple and enjoying the new format of the show. Keep up the good work. Okay, big thank you so much for the review. If you haven't left a review because it is unintuitive, I guess you can push the help button on Apple. I really hope it doesn't come to that. But if you wouldn't mind leaving a review on the show if you haven't already, it helps the show out so much and I would be truly, truly grateful. It's probably the reason why we have hundreds more reviews over on Spotify than Apple, even though 80% of the listeners are from Apple. But if you are one of those Apple listeners and you haven't left a review yet, if you wouldn't mind doing it, I would appreciate it so much. All right, catch you guys all in the next episode.