Join hosts Mike Kaput and Paul Roetzer in this informative episode as they explore the latest advancements in AI. They discuss Microsoft’s innovative release of Copilot Pro, Meta’s ambitious project on ‘Open Source’ Artificial General Intelligence, and the IMF’s significant prediction that AI will affect 40% of jobs worldwide. Prepare for a comprehensive analysis of how these developments are revolutionizing the tech world and what it means for the future of work. 

Listen or watch below—and see below for show notes and the transcript.

This episode is brought to you by our sponsors:

Many marketers use ChatGPT to create marketing content, but that’s just the beginning. When we sat down with the BrandOps team, we were impressed by their complete views of brand marketing performance across channels. Now you can bring BrandOps data into ChatGPT to answer your toughest marketing questions. Use BrandOps data to drive unique AI content based on what works in your industry. Visit brandops.io/marketingaishow to learn more and see BrandOps in action.

Today’s episode is also brought to you by Marketing AI Institute’s AI for Writers Summit, happening virtually on Wednesday, March 6 from 12pm – 4pm Eastern Time.

Following the tremendous success of the inaugural AI for Writers Summit in March 2023, which drew in 4,000 writers, editors, and content marketers, we are excited to present the second edition of the event, featuring expanded topics and even more valuable insights.

During this year’s Summit, you’ll:

• Discover the current state of AI writing technologies.
• Uncover how generative AI can make writers and content teams more efficient and creative.
• Learn about dozens of AI writing use cases and tools.
• Consider emerging career paths that blend human + machine capabilities.
• Explore the potential negative effects of AI on writers.
• Plan for how you and your company will evolve in 2024 and beyond.  

The best part? Thanks to our sponsors, there are free ticket options available!

To register, go to AIwritersummit.com

Listen Now

Watch the Video

Timestamps

00:03:35 — Microsoft’s new Copilot Pro brings AI-powered Office features to everyone

00:11:34 — Meta announces it’s building open source AGI.

00:25:57 — IMF warns AI to hit almost 40% of jobs worldwide and worsen overall inequality

00:36:35 — Altman says ChatGPT will have to evolve in “uncomfortable” ways

00:43:40 — How OpenAI, Meta, Google are planning for 2024 elections

00:47:22 — OpenAI says it’s “impossible” to create useful AI models w/o copyrighted material

00:49:40 — The AI phones are coming

00:52:30 — Elon Musk gives Tesla ultimatum: Another 12% of shares or no AI, robotics

Summary

Microsoft’s new Copilot Pro brings AI-powered Office features to the rest of us

Every Microsoft customer now has powerful AI right at their fingertips: The company recently announced Copilot Pro, a new subscription that gives individuals and creators access to Microsoft’s AI assistant in Word, Excel, PowerPoint, and Outlook.

For $20/month, you can turn on Copilot Pro within your existing Microsoft 365 Personal or Family subscription.

This instantly gives anyone the ability to use AI to do things like: generate entire PowerPoint decks using only text prompts, write and rewrite content automatically in Word, write and reply to emails automatically in Outlook, and start analyzing data and generating graphs in Excel.

The release of Copilot Pro means there are now three versions of Copilot:

1. A free version of Copilot that functions as a ChatGPT-like AI assistant.
2. Copilot Pro, which is specifically for individuals.
3. Copilot for Microsoft 365, the existing business license for companies and teams.

That last one is important…Because Microsoft just made a big change to its business licenses, too. Previously, you had to purchase a minimum of 300 seats to get Copilot access for your company. However, that minimum is now gone, meaning you can purchase any number of seats under 300.

Meta goes all-in on AI with the announcement it’s building open source AGI.

Meta just surprised the AI world with some big announcements: in a recent Instagram Reel, CEO Mark Zuckerberg said the company is committed to building open source artificial general intelligence (AGI).

That means building AI systems that are as smart as people at a lot of different tasks—and making those systems available to anyone online to use and remix as they see fit.

Said Zuckerberg: “It’s become clear that the next generation of services required is building full general intelligence, building the best AI assistants, AIs for creators, AIs for businesses and more that needs advances in every area of AI from reasoning to planning to coding to memory and other cognitive abilities.”

To do that, the company is taking some significant steps, including:

• Bringing its AI research teams (FAIR and GenAI) closer together to align on the goal of building AGI.
• Training LLaMA 3, the next version of its powerful open source AI model.
• And building a breathtaking amount of computing infrastructure by the end of the year, including acquiring a whopping 350,000 H100s (powerful processors built for AI applications).

Interestingly, Zuck also gave a shout out to Meta’s AI-powered glasses, created in partnership with RayBan, saying “I think a lot of us are going to talk to AI as frequently throughout the day. And I think a lot of us are going to do that using glasses. These glasses are the ideal form factor for letting an AI see what you see and hear what you hear.”

IMF warns AI to hit almost 40% of jobs worldwide and worsen overall inequality

These new stats on AI’s impact on jobs made me sit up and pay attention:

40% of jobs across the globe could be affected by the rise of AI, according to new research from the International Monetary Fund (IMF).

60% of the jobs impacted will be in “high-income” nations.

The researchers adapted a commonly used conceptual framework from past studies to measure what human work will be exposed to AI. When looking across all jobs, high-income nations were most exposed to AI transformation.

Why? They theorize that emerging markets and low-income countries don’t have the infrastructure of skilled workers to harness the immediate benefits of AI.

Overall, the research paints a picture of big winners and big losers—with little space in-between:

Some workers in high-income countries will capture huge productivity gains and benefit financially from the technology. Others will see AI lower demand for their labor and possibly even eliminate their jobs.

And lower-income countries unable to fully leverage AI may lose out to higher-income countries that can.

Links Referenced in the Show

• Microsoft’s new Copilot Pro brings AI-powered Office features to the rest of us
• Meta goes all-in on AI with the announcement it’s building open source AGI.
• IMF warns AI to hit almost 40% of jobs worldwide and worsen overall inequality
• Exclusive: Altman says ChatGPT will have to evolve in “uncomfortable” ways
• How OpenAI, Meta, Google are planning for 2024 elections
• OpenAI says it’s “impossible” to create useful AI models without copyrighted material
• The AI phones are coming
• Elon Musk gives Tesla ultimatum: Another 12% of shares or no AI, robotics

Read the Transcription

Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.

[00:00:00] Paul Roetzer: we as Humanity have gone through disruptive general purpose technology changes before, but we’ve never gone through one that’s going to move so quickly.

[00:00:11] Paul Roetzer: Welcome to the Marketing AI Show, the podcast that helps your business grow smarter by making artificial intelligence approachable and actionable. You’ll hear from top authors, entrepreneurs, researchers, and executives as they share case studies, strategies, and technologies that have the power to transform your business and your career.

[00:00:31] Paul Roetzer: My name is Paul Roetzer. I’m the founder of Marketing AI Institute, and I’m your host.

[00:00:40] Paul Roetzer: Welcome to episode 80. Wow. 80. It sounds like a big number. episode 80 of the Marketing AI Show.

[00:00:47] Paul Roetzer: I am your host, Paul Roetzer, along with my co host, Mike Kaput. Hey, Mike, we swapped. You were out of town last week. Where were you last week?

[00:00:56] Mike Kaput: I was in California.

[00:00:58] Paul Roetzer: Okay. Right outside LA. [00:01:00] Alright, I am in Florida, Orlando, I think, so yeah, we are, we’re trading trips and then I think we’ll be back together actually on, on Wednesday this week, right?

[00:01:12] Paul Roetzer: So it is Monday, January 22nd. We are doing this in the afternoon due to my flights. So it’s about 3 20 PM Eastern time. on Monday afternoon. we have a lot to get into today as usual. some pretty cool topics, some bigger topics about some trends that are happening. But first let’s touch on our episode sponsors.

[00:01:35] Paul Roetzer: BrandOps is back again as a sponsor of this episode. Many marketers use ChatGPT to create marketing content, but that’s just the beginning. When we sat down with the BrandOps team, we were impressed by their complete views of brand marketing performance across channels. Now you can bring BrandOps data into ChatGPT to answer your toughest marketing questions.

[00:01:57] Paul Roetzer: Use BrandOps data to drive unique AI [00:02:00] content based on what works in your industry. Visit brandops. io slash marketing AI show to learn more and see BrandOps in action. And this episode is also brought to us by the Marketing AI Institute AI for Writers Summit, which is happening virtually on Wednesday, March 6th.

[00:02:21] Paul Roetzer: From noon to 4 p. m. Eastern time, the lineup is almost complete. I have one more speaker to add to the final panel, which is going to be awesome. It’s about AI writing in the enterprise, everything that goes along with adoption and integration of AI writing tools and platforms. at that event last year in its inaugural year, we had over 4, 000 writers, editors, and marketers join us.

[00:02:47] Paul Roetzer: for that event. So it is a free event. There’s a paid option where you can get on demand, as an upgrade, and there’s a private registration where contact information isn’t passed along to the event sponsor, but it’s free [00:03:00] otherwise. So there’s no reason not to join us if you are involved in content creation in any way, whether you’re at a publisher, a media company, brand side, freelance, whatever it is.

[00:03:11] Paul Roetzer: So join us again for that event, March 6th. You can go to aiwriters. com. or AIWriterSummit. com, that’s AIWriterSummit. com to learn more about that event. It is coming up fast. That’s like six weeks away or something. So, definitely check that out. All right, Mike, let’s get rolling.

Microsoft’s new Copilot Pro brings AI-powered Office features to everyone

[00:03:32] Mike Kaput: All right, Paul. First up.

[00:03:35] Mike Kaput: Every Microsoft customer, even if you don’t work at a big enterprise, now has powerful AI right at their fingertips. And that’s because of a few announcements from Microsoft this past week. First up is that the company recently announced Copilot Pro. which is a new subscription that gives individuals and creators access to Microsoft’s AI assistant in [00:04:00] Word, Excel, PowerPoint, and Outlook.

[00:04:02] Mike Kaput: So for 20 bucks a month, you can turn on Copilot Pro within your existing Microsoft 365 personal or family subscription. So this is for individuals. And it instantly gives you the ability to use AI to do things like generate PowerPoint text using text prompts, write and rewrite content automatically in Word, write and reply to emails automatically in Outlook, and start analyzing data and generating graphs.

[00:04:33] Mike Kaput: in Excel. So now there are three tiers or versions of Copilot now that we have this announcement of Copilot Pro. One is a free version of Copilot that functions like a ChatGPT like AI assistant. The second is Copilot Pro, which we just talked about, which is specifically for individuals. And then the third is kind of what we’ve referenced in the past when it comes to Copilot, which is called [00:05:00] Copilot for Microsoft 365.

[00:05:02] Mike Kaput: This is the existing business license for companies and teams. And this last point is important because in addition to announcing Copilot Pro, Microsoft made a big change. to business licenses. Previously, you had to purchase a minimum of 300 seats to get co pilot access for your company. However, that minimum is now gone.

[00:05:27] Mike Kaput: That means you can purchase any number of seats now under 300. Opening this up to any business that’s interested. So Paul, I want to kick things off and ask if you can kind of give us an overview of what your thoughts are on Copilot Pro specifically and kind of how you see this fitting into the overall picture of Microsoft’s Copilot offerings.

[00:05:51] Paul Roetzer: Yeah, I mean, it seems like the biggest, news here is really that they democratized it for everybody else that isn’t a big enterprise. So, you know, [00:06:00] now if you’re an individual, you know, personal Microsoft 365 user, you have it for your family, or you’re a small business, you can go get it. it certainly seems like the timing of this.

[00:06:12] Paul Roetzer: was very closely tied to ChatGPT team coming out, which also opened it up to businesses under 150 seats. So yeah, I mean, I think it’s the kind of technology we’ve been talking about for months. I mean, they announced that. This was coming, I think it was March of 2023, they sort of, previewed what was coming and then in November, the technology, the co pilot technology rolled out into 365 for companies with 300 or more seats.

[00:06:42] Paul Roetzer: And now everyone else is starting to get the technology. So I think the biggest news is really just the availability now for everyone to have this capability for 20 per month. So for some

[00:06:55] Mike Kaput: of the businesses that are kind of just getting up to speed here now that they have [00:07:00] access to Copilot, which of the features in Copilot for businesses specifically kind of jump out to you as particularly interesting to maybe explore further to enhance productivity?

[00:07:12] Paul Roetzer: For me, I’ve always been waiting to see how the integration with Excel and PowerPoint work. And then the same on the Google side, how it works with Sheets and Slides. Because I think we all, you know, we can experience ChatGPT and get a general sense of how it’s going to work in Microsoft Word or Google Docs.

[00:07:30] Paul Roetzer: We, we, we understand how it creates and edits and simplifies and summarizes and all the things it does there. But I think for a lot of people, especially people who haven’t played around with Code Interpreter, or now known as Advanced Data Analysis in ChatGPT, you don’t really realize what its capabilities are going to be within spreadsheets.

[00:07:48] Paul Roetzer: And a lot of marketers, a lot of business people spend a lot of time in Excel and in Google Sheets. And the reality is that many of us. probably just sort of like hack our way through and [00:08:00] don’t ever really take the time to become extremely adept power users. even though there’s probably tons of efficiencies to be gained.

[00:08:08] Paul Roetzer: Like I always joke about like, if I never have to build a pivot table again, I will be happy. Like if I can just. from what I’ve seen online so far, responses are sort of like lukewarm about, you know, how people perceive the value of the current version of CoPilot to be within Excel and within PowerPoint.

[00:08:32] Paul Roetzer: But, you know, I think the key is like, it. It doesn’t have to go from 0 to 100 in terms of its capabilities or, you know, in terms of how well it automates a process for you. I think for a lot of business and even personal users, it’s just going to be those incremental gains where you ask it to create a PowerPoint for you based on a document.

[00:08:52] Paul Roetzer: And it does like 60 percent of the work. Maybe the design’s not beautiful. Maybe you need to like go in and edit all the notes, but maybe [00:09:00] you, you save yourself the front end two hours of just getting the information in there. And as we’ve talked about before, just that where it gives you a really reasonable first draft, or it does an initial analysis of a dataset that saves you two, three, four, five hours, or does the work you would have had to have gotten someone else in your company to help you with.

[00:09:22] Paul Roetzer: That’s, that can be transformational without being insanely high quality AI.

[00:09:31] Mike Kaput: So if I’m a company that’s now interested in kind of exploring Copilot, but I haven’t done so yet, haven’t really looked too much into it, like, what should I be thinking about next? how should I be approaching this?

[00:09:43] Paul Roetzer: I think the, you know, again, you and I, when we were even prepping for this, it’s like, what, what is what are everything just called co pilot now at Microsoft.

[00:09:52] Paul Roetzer: And we’re, we’re, I mean, we use Microsoft, we, we, we use docs and Excel at times, but we’re largely a Google shop. [00:10:00] so I’m not as aware of like all the different nomenclatures for everything. It’s like, is Google suite or is Microsoft suite? Microsoft 365. Is that also Office? And I know they changed the naming conventions years ago.

[00:10:13] Paul Roetzer: but anyway, so my feeling is, I think the thing that we all have to decide is, Okay, I’m already paying for ChatGPT team. Like, you know, as the institute I’ve talked about this before, I got that for us as soon as it came out. So that’s 30 a month. We experimented with Duet AI from Google for 25 or 30 a month, which I put on hold until the updated version comes out and it improves, but we do also use Microsoft.

[00:10:40] Paul Roetzer: So do we need Microsoft 365 Copilot if we already have ChatGPT team? And then if you’re a bigger company, maybe you have Jasper or Writer and it’s like. It starts to get really complex and there’s going to be a lot of overlap of capabilities. And so I think for most business users, the question is going to [00:11:00] become if we go all in on Copilot, do we, do we need these other tools anymore?

[00:11:05] Paul Roetzer: Or is this actually going to consolidate our tech stack a little bit? And it’s probably too early to tell, honestly. Like I think 2024 is going to be a big year of probably experimenting with a collection of tools and not settling on anything and thinking you’re set. because it’s just going to keep evolving.

[00:11:22] Paul Roetzer: And I think a lot of companies and a lot of leaders I’ve talked to aren’t ready to commit to a platform for the next 12 months. They really just want to keep experimenting with the different solutions.

Meta announces it’s building open source AGI.

[00:11:34] Mike Kaput: So in another big piece of news, Meta just surprised a lot of people in the AI world by making some significant announcements.

[00:11:43] Mike Kaput: In a recent Instagram Reel, CEO Mark Zuckerberg said that the company is committed to building open source Artificial General Intelligence, or AGI. That means building AI systems that are as smart as people at a lot of different tasks, and then [00:12:00] making those systems available to anyone online to use and remix as they see fit.

[00:12:05] Mike Kaput: Zuckerberg said that, quote, It’s become clear that the next generation of services required is building full general intelligence. Building the best AI assistants, AIs for creators, AIs for businesses, and more that needs advances in every area of AI from reasoning to planning to coding to memory and other cognitive AI cognitive abilities.

[00:12:26] Mike Kaput: Now to do that, the company is taking some specific steps and they include bringing together its AI research teams, which are called FAIR and GenAI. bringing them closer together to align on this goal of building AGI. Zuck said they’re training LLAMA3, which is the next version of its open source AI model, one of the more powerful models out there, and they’re also, and this kind of turned heads.

[00:12:53] Mike Kaput: building a breathtaking amount of computing infrastructure by the end of the year. Zuckerberg quoted [00:13:00] acquiring a whopping 350, 000 H100s, which are powerful processors that are built for AI applications. Last but not least, Zuck also gave a shout out to Meta’s AI powered glasses that are created in partnership with Ray Ban.

[00:13:17] Mike Kaput: He mentioned that Quote, I think a lot of us are going to talk to AI as frequently throughout the day, and I think a lot of us are going to do that using glasses. These glasses are the ideal form factor for letting an AI see what you see and hear what you hear. So Paul, first up, can you put this into some context for us?

[00:13:37] Mike Kaput: Like, why is it such a big deal that Meta is making an announcement like this now, compared to, and why is what they’re doing important compared to the approach being taken by people like OpenAI, Google, Microsoft, etc.

[00:13:52] Paul Roetzer: Yeah, honestly, like it was a lot to unpack for a one minute, 42 second video from Zuckerberg.

[00:13:58] Mike Kaput: Yeah. I basically [00:14:00] read you the whole statement. It was not long.

[00:14:02] Paul Roetzer: No, this wasn’t like some big paper that came out. It was a video that he posted on his Facebook page and I’m sure on threads and wherever else. I had to reinstall threads on my phone just to like, go see it. but anyway, but the accompanying.

[00:14:17] Paul Roetzer: host was, I mean, maybe like 120 words. So he basically packed all of those announcements into a one minute and 40 second bit that he just recorded. Like, it looks like in a hotel room. Yeah. but again, there was a lot to unpack there and you hit all the highlights, but Just to expand on a little bit. So the research lab one definitely jumped out to me right away.

[00:14:40] Paul Roetzer: We talk about JaYann LeCun lot on this show. He’s one of the foremost experts in ai, one of kind of the godfathers of modern ai, and really been playing a critical role in it since the 1980s and 1990s, back at Bell Labs. so the fact that his area fair, which used to stand for Facebook AI research, and now [00:15:00] I don’t know what the F stands for anymore within it, but, um.

[00:15:04] Paul Roetzer: When they renamed it meta, they chose not to rename the AI research lab. But the fact that they moved the research lab under the product team was the first thing that jumped out to me because it sounded real similar to like what happened at Google when the brain team and the deep mind team combined.

[00:15:20] Paul Roetzer: In essence, what’s been going on for the last decade at these big companies was they were spending billions on AI research that didn’t have immediate implications to the products. It wasn’t like they had some, you know, KPI that said, like, this is what you have to deliver in terms of product impact. They were just doing the research for the next frontier.

[00:15:38] Paul Roetzer: We’ve now arrived at that frontier, and now there’s an urgency to productize what these teams have been building. So the fact that they kind of brought this under, and so in Meta’s case, they put it under, Chris Cox, the Chief Product Officer. He did an, there was an internal memo, I think this was, this was the Verge that had the [00:16:00] internal memo.

[00:16:01] Paul Roetzer: he said, with this change, we elevate the importance of AI. This is from Chris Cox, the Chief Product Officer. With this change, we elevate the importance of AI research as an essential ingredient to the long term success of the company and our products alongside the major infrastructure investments Mark mentions, which we’ll talk about again in a minute.

[00:16:20] Paul Roetzer: Moving FAIR and GEN AI closer together will mean a more coherent AI research portfolio and roadmap. This is the part I boldfaced. With LLAMA becoming the primary launch vehicle for progress toward AGI. So, the research lab thing was huge. The AGI thing was fascinating because I don’t remember Zuckerberg stating that as a goal for meta previously.

[00:16:45] Paul Roetzer: Do you recall

[00:16:46] Mike Kaput: that? Not, no. I was actually, I had to double check a couple times preparing for this. I was like, did he, he for sure said general intelligence, right? Oh yeah. I have not heard it in the past and that’s why I was double checking. It’s [00:17:00] not common language for them to use.

[00:17:02] Paul Roetzer: So now, interestingly related to that, it ends up that, LeCun was at the World Economic Forum last week and he was doing an interview about this and he started talking about human level intelligence.

[00:17:19] Paul Roetzer: And so I’ll just read a couple of interests, because again, this is like, this isn’t specifically tied to what Zuckerberg said, but it’s interesting that LeCun, who is the AI researcher, the head AI guy at Facebook, He said, human level AI is not just around the corner. This is going to take a long time and it’s going to require new scientific breakthroughs that we just don’t know yet.

[00:17:42] Paul Roetzer: The systems are intelligent in the relatively narrow domain where they’ve been trained. They are fluent in language, and that fools us into thinking that they are intelligent, but they are not intelligent. It’s not as if we’re going to scale them up and train them with more data, with bigger computers, [00:18:00] and reach human intelligence.

[00:18:01] Paul Roetzer: This is not going to happen. what’s going to happen is that we’re going to have to discover new technology, new architectures of those systems. So now if you follow Yann, this is not a surprising comment at all. He is very much of the belief that we don’t get to AGI through language models. He believes that these things need worldviews.

[00:18:20] Paul Roetzer: They need to understand the world around them and learn like a child does, where it’s observing things and making connections. And so that’s long been his belief, but just the fact that the week. Zuckerberg states they’re going after AGI. LeCun’s like, yeah, we’re, no, we’re close to this because you, you wouldn’t think they work for the same company in that moment when you were looking at those two things, the GPU thing, I get that for, for a lot of people who listen to GPU things, probably kind of an abstract thing, like what the hell are GPUs and what does that mean?

[00:18:49] Paul Roetzer: Like 600, 000 equivalent that I was trying to find some context to give people, where it’s. Kind of obvious. So here’s the only [00:19:00] one I could find that’s like, we know for sure, because a lot of labs don’t disclose this. Like OpenAI has never actually disclosed how many GPUs they train GPT4 on. It’s believed it was about 20 to 25, 000 A100s, which is another type of Nvidia chip.

[00:19:16] Paul Roetzer: So 20 to 25, 000 versus 300 to 600, 000, you can do that math. But the other one we know for sure. We talked about this, I don’t know, 10, 20 episodes ago, is, Inflection, which is training Pi, pi. ai, and they said when they raised a bunch of money in June, that they were, their goal was 22, 000 chips, H100 chips by the end of 2023.

[00:19:42] Paul Roetzer: So 22, 000 was their goal, and they, they said. Inflection is building the largest AI cluster in the world, NVIDIA H100 GPUs. Well, I’m guessing it isn’t the largest anymore, if that was the goal. [00:20:00] So, that just gives you a sense. And then the only other one I saw was actually ARK Invest. The director of research there tweeted, 600, 000 H100 equivalents at Meta is roughly eight times Tesla’s roadmap by the end of 2024.

[00:20:15] Paul Roetzer: So it is a lot of computing power is basically what we’re saying. I have, I tweeted like, wow, talking about like the ultimate Silicon Valley flex is like 600, 000 GPU, just drop the mic and leave. I heard one estimate. It was about 20 billion worth of chips. I don’t, I don’t know exactly how much they go for.

[00:20:35] Paul Roetzer: So, and then the fact that they’re training Lama three is like, Oh, by the way, we’re probably training a model more powerful than GPT 4 and that was like the side note within this. So yeah, it was wild, man. That was, I don’t know. And again, you gotta love, like, you and I came up working in an agency. I came up in the PR industry where, like, you work for months to try and get, like, [00:21:00] a headline for your client.

[00:21:01] Paul Roetzer: Like, a big news headline. You plan product launches forever, major announcements. And now it’s just like Zuckerberg just drops a minute and a half video of him, you know, rolling out of bed in the morning saying they’re doing all this stuff and it’s everywhere. Like you don’t even need a PR agency anymore.

[00:21:18] Mike Kaput: Yeah. So I want to kind of unpack a little bit. What kind of, you know, we talk about OpenAI, Google, Microsoft quite a bit, like what kind of unique advantages does Meta have in the AI space? I mean, they’re doing all these very interesting things, but how are they differentiating and competing against some of these other giants?

[00:21:39] Mike Kaput: I mean, what strikes me is we’ve seen some recent articles kind of come out about the sheer amount of data. that they’re collecting on each of their users. it seems like maybe this is something that Meta has that others might not. One article from The Verge reported that a stunning 48, 000 companies sent [00:22:00] Meta data on a single person.

[00:22:02] Mike Kaput: And it also found that on average, Meta received data from more than 2, 200 companies. on each individual that they analyzed in this study. So like, is that the secret sauce that Meta has that others don’t?

[00:22:18] Paul Roetzer: Yeah, I think you could do a trilogy. They have compute, obviously, that probably, you know, I would imagine Google might be the only one that, is surpassing them at the moment.

[00:22:29] Paul Roetzer: I don’t, I don’t know that to be fact, but I would assume, they have immense amounts of data, not only the stuff that they’re getting through all these third parties, But the stuff that they get from Facebook and wherever else that they’re taking their data from. And then the other one that, Yann LeCun tweeted last week.

[00:22:49] Paul Roetzer: Was actually, it was an Ethan Mollick tweet that Yann retweeted. they spend more on R& D when you look at percentage of revenue than any company in the world. [00:23:00] So they’re at 30 percent and this has been going on for the last, you know, 10 plus years. So they’ve been investing in AI research for a very long time.

[00:23:08] Paul Roetzer: Um. The chart, which we’ll put a link to 30% of meta’s revenue goes to r and d. The next closest is Nvidia at 27%. And then, A SML at 15%. Amazon alphabet at 14%. Just to give you context, apple, which spends billions, is at 7%. So meta doesn’t spend the most in terms of dollars, but percentage of revenue.

[00:23:34] Paul Roetzer: They are the largest spender. So they have been committed, no, they blew. 10 billion of that on the metaverse, but that’s like a totally separate topic. So that’ll eventually pay off. I’m not, I’m not saying it was a total waste of money.

[00:23:50] Mike Kaput: So to wrap this up, there’s kind of one big element of this that we haven’t talked about yet.

[00:23:56] Mike Kaput: you know, on one hand, this can be kind of really exciting. Like we’re putting [00:24:00] powerful AI in the hands of everyone that can go create and innovate. But on the other, this whole open source thing seems like it could be a potential nightmare. for responsible AI usage. Like what happens once powerful AI, even we’re talking now general artificial intelligence or artificial general intelligence rather, is open source to anyone.

[00:24:22] Mike Kaput: Like, is this safe? Is this the responsible way to Do this. It feels a bit like playing with fire from one perspective.

[00:24:29] Paul Roetzer: Yeah, I don’t know. I mean, I wish I had the answer. There are a lot of extremely smart people, many of whom you and I both respect and follow closely who think open source is the only way forward that we can’t trust, you know, three, four organizations in the world to control AGI and all the power.

[00:24:48] Paul Roetzer: And so they see open source as the only way to democratize this. And then you accept that there’s going to be bad actors and ramifications that come along with it. The alternative is, you know, more in the Google [00:25:00] OpenAI camp where they want it controlled. They want the regulation because they believe it’s dangerous.

[00:25:05] Paul Roetzer: And the open source people would say you’re only doing that to try and prevent. competition, the regulatory capture concept we’ve talked about a number of times. I don’t, I don’t know where I fall, honestly, on what I really believe. All I know is it’s irrelevant because we’re done. There’s no turning back.

[00:25:23] Paul Roetzer: The open models are out there. They’re not only in the U. S., they’re in other countries. and It’s just where it’s, it’s going. And, I think we just have to do our best to prepare for the fact that this is, the world we’re going to live in. There will be very, very powerful open source models. Meta is intent on that happening.

[00:25:43] Paul Roetzer: And even if they put regulation in place today, we’re, it’s already. Too late to turn back. I think we have to figure out how to live in this world of, of both models existing. All right.

IMF warns AI to hit almost 40% of jobs worldwide and worsen overall inequality

[00:25:57] Mike Kaput: Our third big topic today is about [00:26:00] some new stats on AI’s impact on jobs. And a couple of these kind of made me sit up and start paying some attention.

[00:26:08] Mike Kaput: 40 percent of jobs across the globe could be affected by the rise of AI, according to some new research from the International Monetary Fund, the IMF. And 60 percent of the jobs that are going to be impacted, they say, will be in, quote, high income nations of the developed world. These researchers at the IMF kind of adapted a commonly used conceptual framework from a bunch of past studies to measure which types of human work will be exposed to AI.

[00:26:38] Mike Kaput: And when they looked across all these different types of jobs, high income nations ended up being most exposed to AI transformation. Now, why is this the case? Well, they theorized that emerging markets in low income countries don’t have as much of the infrastructure of skilled workers to harness the immediate benefits of the [00:27:00] AI that we have today.

[00:27:02] Mike Kaput: Now, overall, there’s a ton to unpack in this. We’ll link to all of the relevant research. There’s a lot to go through. I’d encourage you. to explore it if you’re interested in it. But the overall picture here is that the research is saying AI is going to create some really big winners and some really big losers.

[00:27:21] Mike Kaput: And there’s not a ton of space it sounds like in between. Some workers in high income countries are going to capture huge productivity gains and benefit financially from the technology. Others are going to see AI lower demand for their labor and possibly even eliminate their jobs. And the lower income countries generally sounds like they will be unable to fully leverage AI and may lose out to higher income countries that can use it to create macroeconomic benefits.

[00:27:53] Mike Kaput: So first I kind of want to take a step back here, Paul, before we dive into the specifics [00:28:00] of this study and kind of frame the AI employment discussion for anyone who is either new to it or hasn’t been following it that closely. Like, Why exactly are we so worried as a society about AI taking jobs in the first place?

[00:28:16] Mike Kaput: Like, it seems like we’re way more worried about AI and its impact on human employment than we have been at any other time when we talk about, like, traditional automation, say, in factories or manufacturing, things like that.

[00:28:31] Paul Roetzer: Yeah, I think to kind of summarize the challenge here is we as Humanity have gone through disruptive general purpose technology changes before, but we’ve never gone through one that’s going to move so quickly.

[00:28:46] Paul Roetzer: usually you have years or decades to adapt. This one probably isn’t going to give us that much time. In some industries, it can be months or, you know, one to two years where you have to kind of evolve. And so I think a lot of [00:29:00] people. just assume we’ll figure it out. You know, a large percentage of the U.

[00:29:05] Paul Roetzer: S. population, and really around the world, were farmers at one point. And now they’re not. Now it’s a very small percentage of populations. People say, well, we, we evolved. We found other jobs to do. horses were a pretty big thing for a while in, in, in city streets. And now you don’t see them as often on city streets.

[00:29:24] Paul Roetzer: And so, you know, you have this whole evolution that happens. But the thing we look at is You can look at the replacement of roles, and I think this is what where the flaw happens a lot of times is people think about you will or will not need lawyers. . You will or will not need marketing consultants.

[00:29:42] Paul Roetzer: You will, will not need customer service representatives. it’s not so much that the AI is going to replace at a one-to-one level that it’s going to take a job or a role and just eliminate it and do 100% of that job. What we talk about all the time is. [00:30:00] efficiency gains within that role to where you may just not need as many of those people.

[00:30:05] Paul Roetzer: So, you know, top of mind for me right now is tax returns. So if you think about humans doing tax returns, it’s a very manual process. Well, if The AI technology gets really, really good at assisting CPAs in doing these things. And to the point where maybe they can do five to 10 times the number of returns, that they used to do, but in the same amount of time, the question becomes, do you need as many?

[00:30:30] Paul Roetzer: the same can be applied to any form of knowledge work. It’s just a question of if the AI does 10, 20, 50 percent of that job of what that person does 150 hours a month or 180 hours a month. If it’s doing that percentage of it, do we need as many people doing the work anymore? And so that’s, that’s kind of, I think you and I are very similar playing here, at least, you know, based on the conversations we’ve had.

[00:30:54] Paul Roetzer: It’s, it’s not that we think AI is going to come in and just fully automate job, people out of [00:31:00] jobs, but it seems very apparent it’s going to drive massive efficiency gains in a lot of knowledge work roles. and it’s going to happen probably pretty quickly. And so will industries and companies have the time to adapt to where they can find other responsibilities for those people?

[00:31:19] Paul Roetzer: Or are they just going to need fewer people to run these companies? And so I think that’s, that’s why studies like this are important because we don’t have the answers. Like the best economists in the world have research out about this and they have theories, but no one actually knows what’s going to happen.

[00:31:36] Paul Roetzer: And so we do surface this conversation a lot when we see reports that we think are worth paying attention to, not because we’re pointing at it saying, see, we told you like this, but we’re just saying it’s another data point. It’s another perspective on a very important topic that doesn’t get enough conversation.

[00:31:53] Paul Roetzer: I wish at least like in the United States that it was a much bigger part of the upcoming election, but I haven’t heard anything about [00:32:00] this. In fact, it’s not even a topic, which is crazy to me.

[00:32:04] Mike Kaput: Yeah, it really seemed like the only mainstream political conversation we had around this was when Andrew Yang, for instance, was like running when he had the whole universal basic income, but that is really falling far into the background over the years.

[00:32:17] Paul Roetzer: Yeah, it doesn’t, when you look at the kind of the rhetoric of the current election cycle, I just, I don’t see this even getting surfaced as a topic, which is, worrying to me because that’s, you know, 10 months from now or the next, you know, whoever the president’s going to be in the United States will take office about 12 months from now.

[00:32:37] Paul Roetzer: And that means we’re now a year further out from doing anything about a technology that is racing through, industries.

[00:32:46] Mike Kaput: Especially with everything we just talked about with open source and the previous topic too, regardless of how slow or fast AI adoption is. At the enterprise level, people have this stuff running on their laptops right now.

[00:32:59] Paul Roetzer: Yeah, and [00:33:00] on your phones. I mean, there’s

[00:33:01] Mike Kaput: technology on your phones. So this study takes a really big picture kind of macroeconomic approach to the topic and totally understandable because it’s the IMF doing this. But I’d love to talk for a minute about the micro, the on the ground impact, like If I’m your average business owner or business leader, small, medi large business, like how is this changing my perspective on hiring and human resources?

[00:33:29] Paul Roetzer: Yeah, the guidance we give organizations is that you need to start future proofing your hires. So if you’re looking out ahead and you have job descriptions of the people you’re going to hire, You should have someone who has a reasonable understanding of AI technology and where it’s going to go in the next 12 to 18 months and can look at that job description and say, if that’s 180 hours now per month.

[00:33:56] Paul Roetzer: It’s probably 110 to 90 hours, [00:34:00] 12 months from now. Do you make that hire? And so what you could do is literally go through, I know you and I both teach this when we run workshops. Take the job description and look for the things that can be intelligently automated. Look at the bullet points of what people do.

[00:34:12] Paul Roetzer: So you can do that with your current open roles and look at those jobs and say, Okay, we were going to hire five people for this role. Based on an assessment of that role, looking 12 months out, maybe we only need two, let’s, let’s hire two instead of five. So we don’t have to reduce. So again, we’re not saying don’t make hires or that you’re just going to run with a ton fewer people.

[00:34:35] Paul Roetzer: Maybe it’s just, you don’t grow as quickly. We’re seeing this consolidation in the tech industry already. Like just three weeks into January, I’ve seen tens of thousands of layoffs in Silicon Valley. Now they’re not saying it’s all because of AI, but they’re looking at efficiency gains within companies saying we can run a much leaner organization now.

[00:34:51] Paul Roetzer: So that’s one thing you can do is. Don’t make new hires without assessing how AI is going to impact those roles. The other thing you can do is take your [00:35:00] existing teams and roles and do an assessment and say, okay, 12 to 18 months from now, how much is AI going to be doing this person’s job and start planning now.

[00:35:10] Paul Roetzer: And that doesn’t mean start cutting jobs. Now that means start finding ways to reallocate resources, to reskill and upskill people. So I think the best organizations are just going to take a very proactive approach to this. And if it ends up that it doesn’t gain 10, 20, 30 percent efficiencies for different roles, I mean, what’s the worst case that you were prepared that it did?

[00:35:32] Paul Roetzer: Like that’s, that’s my feeling on this is because we don’t know the smartest economists in the world don’t actually know what happens. Isn’t it wise to prepare for both outcomes? One is nothing changes. Which like business as usual, the alternative is we gain a ton of efficiency and maybe we don’t need as many people doing the jobs that are currently there today.

[00:35:53] Paul Roetzer: and then the third thing, I guess, to look out ahead is what are the jobs that might be created over the [00:36:00] next 6 to 12 months that we’re not even thinking about? And I think that’s an exciting thing. And that, again, like You may not have the people in your company that can envision that, but go find the people.

[00:36:09] Paul Roetzer: There are consultants out there. There are people who can come in and kind of help you think this through, but really start to think in a more innovative way of what does a future marketing organization look like? We had Dan Slagg and the CMO of tomorrow. io spoke at our Macon conference last year, and that was what he did.

[00:36:25] Paul Roetzer: He kind of like re imagined what a marketing org chart looks like and kind of started restructuring. I think you’re going to see a lot of really forward thinking companies doing that.

Altman says ChatGPT will have to evolve in “uncomfortable” ways

[00:36:35] Mike Kaput: All right, let’s dive into our rapid fire topics this week. So first up, OpenAI CEO Sam Altman says that the company’s next model quote, will be able to do a lot, lot more than today’s models.

[00:36:49] Mike Kaput: This comes from an interview Altman gave to Axios at Davos this past week. He said that the development of the company’s future AI products will need to [00:37:00] allow, quote, quite a lot of individual customization and that this would make, quote, a lot of people uncomfortable. Now that’s because he foresees future AI tools giving different answers for different users based on their values.

[00:37:16] Mike Kaput: Altman also said the company’s top priority is launching its new model, which many think is likely to be called GPT 5. And there was a juicy bit of gossip revealed during the interview when Altman said that he quote, isn’t sure on the exact status of Ilya Sutskever’s employment at the moment. Sutskever is one of the OpenAI board members and employees who led last year’s coup against Altman.

[00:37:42] Mike Kaput: So Paul, what jumped out to you about this interview? I’d be curious about your thoughts both generally on Altman’s perspective on where AI is going and also would love to hear about what’s going on with him and Ilya Sutskever.

[00:37:55] Paul Roetzer: Yeah, so I’ll first say this, the uncomfortable [00:38:00] thing you led with, we’ve, we’ve known this is coming.

[00:38:02] Paul Roetzer: I mean, they started talking about this within probably the first 30 days of GPT 4 coming out in March 2023. There was a lot of pushback, early on that it was too liberal, that it was too far to the left, from a U. S. political perspective. And so they had to go in and kind of make a lot of adjustments to get it closer to the center so people stop complaining and then the people on the left complained and, you know, it was just as politics does.

[00:38:29] Paul Roetzer: Everybody has to complain about something. And so they said then that the future versions would let you determine how you want your model to talk to you basically. So they pretty much implied there was going to be like a political filter and that this is going to be done by country, it’s going to be done by individual.

[00:38:47] Paul Roetzer: where it’s kind of like you get to choose what media outlets you read. you, you choose what kind of information bubble you live in. They’re basically going to let your language model be a bubble too, if you want it to be. [00:39:00] And so they see that as the only path forward, I guess. Inflection has taken a bit of a different perspective.

[00:39:07] Paul Roetzer: So again, Mustafa Salomon, InflectionPi has basically said, we’re going to build it with what we think are the right human values. If you don’t like it, don’t use it. Anthropic Claude has basically built on some fundamental human values that are generally agreed upon and that’s kind of what you’re going to get with their platform.

[00:39:24] Paul Roetzer: So we’re going to start to see this sort of fragmenting of how these models work based on what they’ve been trained to do and how they’ve been kind of red teamed or what guardrails have been put in place. So, I do think it’s going to cause a lot of problems. I mean, you’re going to see a lot of negative mainstream media headlines when these things come out.

[00:39:44] Paul Roetzer: They say uncomfortable things more commonly, but that’s what you’re going to get with the open source models. We talked about Llama 3, like you’re not going to have those guardrails in Llama 3. So, yeah, I mean, I think that was my first take is. It wasn’t terribly surprising, the fact that they’re [00:40:00] working on GPT 5, not surprising at all.

[00:40:02] Paul Roetzer: I think people need to prepare themselves for a very, I don’t know if disruptive is the right word, like, it’s going to get weird. We’re going to get LLAMA 3 at some point this year, it sounds like in the first half of the year. We’re probably going to get GPT 5 in the first half of the year. We’re most Definitely, I assume, getting Gemini Ultra from Google in the first half of the year, maybe the first quarter.

[00:40:27] Paul Roetzer: Anthropic Clause, like, whatever he says about the reasoning ability, and its ability to understand its ability to do all these personalization things, they’re all going in that direction. So, I don’t think we’re going to have, like, this moment where models just look completely different and, like, completely foreign to us, but I think you’re going to have a massive step up in what they’re capable of doing, as we start to progress.

[00:40:53] Paul Roetzer: And actually the one thing I put this on, I think this was on Twitter, but the one day last week I had this random thought, like I had [00:41:00] listened to this podcast from Nathan Labenz, I think is, is his name. and he had been part of the red team for GPT 4. So he had access to it. before they put the guardrails in place.

[00:41:13] Paul Roetzer: It’s raw ability. And it got me wondering, like, when I was listening to these interviews with Sam, and Sam’s a pretty persuasive guy, and he’s talked about how these things are going to be like superhuman at persuasion, and it got me thinking, like, do you think that Sam and Greg and Ilya before he, you know, kind of ousted or wherever Ilya is, in limbo right now, do you think they use a raw version of GPT 4, like an unedited, unrestricted version of GPT 4, and then carry that forward.

[00:41:45] Paul Roetzer: Once they’re done training GPT 5, do they keep like the raw power of the original model to themselves? I don’t, I have no idea what the answer is to that, but I find myself starting to wonder. Because when you listen to Nathan explain [00:42:00] what GPT 4 was capable of before they took all these restrictions and put them in there, it’s like, wow, like how much of an advantage would a company have if they still had the raw power of those models before they were.

[00:42:14] Paul Roetzer: kind of neutered. I don’t know. I think

[00:42:16] Mike Kaput: this really does tie back to our previous podcast discussions about effective, accelerationism. This idea, the idea that some people, some leaders in Silicon Valley believe that technology, technological progression at all costs is the greatest good. Given that lens, I would be shocked if they weren’t.

[00:42:39] Mike Kaput: Keeping this at least a couple of the people in this mix have sympathy for those types of beliefs I would be just flabbergasted if they had I don’t know how much they’re using it. Maybe by Destroying it or getting rid of it. I that would shock me. Yes. We’ll never know that Conspiracy there. [00:43:00] 

[00:43:00] Paul Roetzer: Yeah, but I was just curious because it’s like how do you if he knows?

[00:43:04] Paul Roetzer: I mean super persuasion that makes you think like he’s seen it like that. He’s Has access to persuasive capabilities beyond what we’re seeing because he generally tweets things out when he knows them to be true. so yeah, I’m not saying he actually is, that’s how he’s doing it, but he’s got a pretty good pulse on it.

[00:43:22] Paul Roetzer: So yeah, I, and the Iliad thing was fascinating when they just said point blank, like what’s his role? And he’s like, I don’t, I don’t know. It’s like, you’re the CEO. You, you don’t know. If he’s in the company still or not, that was weird. There’s,

[00:43:35] Mike Kaput: I’m sure there is much more to that story that will come out.

How OpenAI, Meta, Google are planning for 2024 elections

[00:43:40] Mike Kaput: So in some other news, OpenAI also recently announced that it’s ramping up its efforts during the 2024 election season worldwide. Quote, To prevent abuse, provide transparency on AI generated content, and improve access to accurate voting information. Notably, the company said in a statement, quote, We’re still working to [00:44:00] understand how effective our tools might be for personalized persuasion.

[00:44:03] Mike Kaput: So until we know more, we don’t allow people to build applications for political campaigning and lobbying. Meta and Google also announced plans in late 2023 that were centered around elections. Meta says it is continuing what it sees, and it claims, are robust efforts to prevent misinformation and influence campaigns that are from, say, foreign governments or bad actors designed to sway elections.

[00:44:30] Mike Kaput: It’s also added some new AI Focus measures to the mix. In certain cases, advertisers will have to disclose. Now if they have used AI to create a political ad. On Google’s end, the search giant plans to restrict certain election related questions. In its generative AI search responses and in barred advertisers are now required to disclose when they contain in their ads realistic synthetic content.

[00:44:57] Mike Kaput: That’s been digitally altered or generated [00:45:00] and YouTube creators will soon need to acknowledge this as well. So, Paul, what problems are you most worried about during, you know, obviously elections in the US are the most relevant ones to us, but just in general worldwide elections, in this kind of age of AI and do you find that these measures go far enough to

[00:45:22] Paul Roetzer: combat them?

[00:45:24] Paul Roetzer: I’m glad that they’re focused on it and working on it. I mean, I’m sure they’re putting a lot of resources behind it. That is good. As we’ve talked about the show before, I don’t. I don’t think it’s going to matter that much. Like just this morning, I saw, there was a deep fake robo call from Joe Biden calling Democrats in New Hampshire, telling them not to go out and vote on Tuesday because they’re only going to help Trump get into office if they use their vote on Tuesday, implying to people who maybe don’t understand you get to a vote in the both elections that it’s not just the, Tuesday, but.

[00:45:57] Paul Roetzer: So I [00:46:00] just think that the amount of synthetic content that’s going to be created and weaponized is just going to be too much to contain. And I think that education to citizens about what AI is capable of and the need to like, I don’t know if it’s a fool’s errand to pursue, but like, not believe what you see online or what you hear.

[00:46:21] Paul Roetzer: I have this conversation with my kids all the time, like verified sources. Like if you see or hear anything online, it has to be a verified source. If it’s an influencer who you believe, make sure it’s coming from their channel. That not like. Reshared or coming from somewhere else where it could be deepfaked.

[00:46:38] Paul Roetzer: And so I have this conversation daily with my 10 year old, my 12 year old about how to trust, how to find trusted and verified sources. I don’t think we can that quickly educate society. I’m not sure that a lot of society wants verified and trusted sources. I think they, they just want to share what they see and the crazier it is, the.

[00:46:58] Paul Roetzer: More likely they are to share it. [00:47:00] So, I don’t know. I really want to be, like, optimistic about, about this in terms of the election, and I really struggle to find a positive angle to any of this, to be honest with you. I hear ya. I laugh, because I, so otherwise I’ll just cry if I think about this.

OpenAI says it’s “impossible” to create useful AI models w/o copyrighted material

[00:47:22] Mike Kaput: So in some other news, OpenAI has said, quote, It would be impossible to train AI models without using copyrighted material. Now this comes from a statement the company submitted to the United Kingdom’s House of Lords as part of an inquiry. Being undertaken by that body and in it OpenAI comes out and says Quote because copyright today covers virtually every sort of human expression including blog posts, photographs, forum posts scraps of software code, and government documents, it would be impossible to train today’s leading AI models without using copyrighted [00:48:00] materials.

[00:48:00] Mike Kaput: The company also said that it can’t just train AI models on public domain books and drawings. They’re just too old and limited to quote, meet the needs of today’s citizens. Paul, what did you make of this argument? It seems like OpenAI is kind of increasingly relying on this idea that public material is both necessary and quote, fair use.

[00:48:23] Mike Kaput: to include in AI training.

[00:48:26] Paul Roetzer: Yeah, I know we talked a little bit about this on the previous episode, you know, the New York Times lawsuit of OpenAI. I don’t know. I mean, I feel like this is at some point just going to snowball out of control to where no matter what they do from a legal perspective, again, like the open source models are out there trained on all the same stuff.

[00:48:47] Paul Roetzer: Like, How do you shut them all down? Like maybe, I don’t know, maybe there are some legal maneuvers I’m not aware of or some technical maneuvers I’m not aware of that once this stuff is permeated through [00:49:00] online and throughout society, like how do you take it all back and stop it? I get that it might stop OpenAI from like legally training future models, but I also feel like that’d be part of their legal argument.

[00:49:09] Paul Roetzer: It’s like, Stop us, like it’s not going to stop all of them. Like there’s models are everywhere. So I don’t know. I mean, it’s again, a topic we’re going to kind of keep pursuing and we may do some more deep dives. for the AI for Writers Summit, we’ve got actually an entire session, with a lawyer, IP lawyer on this topic, like copyrights and everything.

[00:49:27] Paul Roetzer: So it’s definitely an area I’m really intrigued by. but OpenAI obviously has a lot on the line to convince the governments of this.

The AI phones are coming

[00:49:40] Mike Kaput: So, Samsung has signaled in the strongest way possible that it is all in on AI powered smartphones. At its Samsung Unpacked event last week, they unveiled a new, their new Galaxy phones, and all of these phones are infused with AI. Now, this is largely thanks to [00:50:00] Galaxy AI, which is Samsung’s on device and cloud based AI models.

[00:50:05] Mike Kaput: Galaxy AI is embedded in every one of its new Galaxy S24 phones. And it gives the users a host of generative AI capabilities that are actually powered by Google’s Gemini models. So some of the initial AI features they’ve demoed on these phones include something called like circle to search capabilities, where you circle parts of text, videos, and photos to get instant search results.

[00:50:30] Mike Kaput: You can use AI to give you live translations when you’re on a phone call, which is pretty cool. There are voice transcription and summarization capabilities. Generative AI image and photo editing, and generative AI chat features that help you write messages, back to people when you’re communicating.

[00:50:50] Mike Kaput: So Paul, this kind of ran under a headline in The Verge called the AI phones are coming. What do we mean when we say AI [00:51:00] phones? Like how are these Fundamentally different from the smartphones we know and love.

[00:51:06] Paul Roetzer: I mean, I think they probably should just put the generative AI phones are coming . I dunno. I mean, we’ve had AI in our phones for a decade.

[00:51:12] Paul Roetzer: Yeah. Or more. So it’s not like the iPhone doesn’t have, or the Pixel doesn’t have AI in it, but it just seems like they’re just deeply infusing all aspects of generative AI into the phone, which they’re all going to do. But I knew, so I had actually flagged this one at some point last week and then. Sunday night, I get a text from my dad saying, Did you see that, Galaxy AI ad for the Samsung phones?

[00:51:37] Paul Roetzer: And I happened to miss it. I was, we were watching the football game and that’s, I guess it was on the Bill’s, Chiefs game. And so it happened twice. So I missed it both times and he’s like, You gotta look it up. So I have not seen it yet, dad, but I, I, Mike and I both look, we were trying to find the ad online.

[00:51:51] Paul Roetzer: I couldn’t find it. no, I think this is. And this is part of my challenge with this whole wearables category, which we’ll dive back into another time. The [00:52:00] phones are going to become so smart and it’s the thing we’re already used to that that’s why I just don’t think I’m going to need a rabbit device and a pin on my shirt and a pendant from my neck and glad like.

[00:52:13] Paul Roetzer: The phones are going to be the thing. They’re already the thing that hundreds of millions or a billion plus people use every day. So this is absolutely going to be a 2024 thing. Your phones are going to become infinitely smarter. I think it’s going to be really cool application of it.

Elon Musk gives Tesla ultimatum: Another 12% of shares or no AI, robotics

[00:52:30] Mike Kaput: So, last but not least, Elon Musk is back in the news, or did he ever leave?

[00:52:36] Mike Kaput: Musk posted on X that he feels uncomfortable growing Tesla into an AI and robotics leader without being awarded 12 percent more of the company, which would give him about 25 percent ownership in Tesla. He said if this doesn’t happen, he’d prefer to build AI products outside of Tesla. Now a portion of his statement that he put on X gives us a [00:53:00] little context into why he claims this is important.

[00:53:02] Mike Kaput: He says, quote, I am uncomfortable growing Tesla to be a leader in AI and robotics without having around 25 percent voting control. Enough to be influential, but not so much that can’t be overturned. Unless that is the case, I would prefer to build products outside of Tesla. You don’t seem to understand that Tesla is not one startup, but a dozen.

[00:53:24] Mike Kaput: Simply look at the delta between what Tesla does and GM. So, Paul, you’re a long time watcher of Elon Musk and his companies. Like, what’s going on here? Why is it so important to control Tesla more as the company develops powerful

[00:53:39] Paul Roetzer: AI? Yeah, I love how Elon Musk uses Twitter because that tweet was in a reply to some dude with 11, 000 followers calling him out as like, he already owns 411 million shares, like what else does he need?

[00:53:53] Paul Roetzer: And that’s when he, you know, so he responds to this random dude. with this [00:54:00] explanation of like, I want another quarter of the company basically. yeah, II I mean, I get what he’s saying, like he’s looking out ahead and saying, okay, if, if I put Optimus into this business and all these other AI things we’re building, and I let that live underneath the Tesla domain and all the shareholders get pieces of that and they become as powerful as I think they’re going to be, I don’t want to be at the mercy of.

[00:54:26] Paul Roetzer: Shareholders to dictate, or at least not as much as I am currently with roughly 13 percent ownership of the company that he has. so, it’s a significant jump though. I mean, to go from, 13 percent ownership to 25 percent ownership is a massive leap. I understand why he said, I don’t know what’s motivating it at the moment.

[00:54:49] Paul Roetzer: I don’t know like why all of a sudden this is like a topic. Usually there’s some other thing brewing in the background where he tweets things like this. I don’t know. It’d be really interesting. That’s a, [00:55:00] it would be a crazy thing to see, but he does get into like, there’s some compensation plan case going on in Delaware that.

[00:55:10] Paul Roetzer: went to trial, was held in 2022. There was no verdict yet or something. So it’s like, this isn’t something that’s going to come to a head in the next couple of weeks, but it is interesting that someone’s of his stature is basically saying online, like, give me another, however many billions of dollars worth of stock, or I’m not, I’m going to take my companies out of here and the value of them.

[00:55:31] Paul Roetzer: I sold some Tesla shares that day, I’ll tell you that. There’s not really something you want to see as a shareholder of Tesla that like the CEO is threatening to extract massive future value from the company. That’s a little unsettling. As much as I love my Tesla and you know, the company itself makes great products.

[00:55:50] Paul Roetzer: that’s weird to see. I can’t think of another company where that would be okay for the CEO to do something like that.

[00:55:58] Mike Kaput: All right, well, that’s a [00:56:00] wrap for this week. I would encourage everyone to also check out Marketing AI Institute’s newsletter. Our newsletter called This Week in AI covers a ton of different topics that we didn’t get to cover this week.

[00:56:14] Mike Kaput: cover on the podcast today. You’ll get an in depth breakdown of some of our main topics as well as all the other news that we just couldn’t fit into this episode. So it’s one of the best ways out there to quickly stay up to date on super relevant, super actionable AI news. And you can get that at marketingaiinstitute.

[00:56:33] Mike Kaput: com. I also, in honor of Paul traveling, want to mention that, if you are looking for a dynamic, speaker on issues in AI and business, both Paul and myself do dozens of talks every single year, to help marketers, executives, and entrepreneurs understand the opportunities and obstacles that AI presents to their businesses.

[00:56:57] Mike Kaput: So, if you’re interested in bringing one of us in to [00:57:00] educate your organization at your next conference, corporate event team meeting, just go to MarketingAIInstitute. com, click about and click speaking to learn a little more about the kinds of talks we give and get in touch. Paul, thanks again for breaking down and demystifying the world of AI this week.

[00:57:19] Paul Roetzer: We really appreciate it. Yeah. Appreciate you putting everything together and I will hopefully see you in the office on Wednesday. Sounds good. All right. Talk to you soon. Thanks everyone.

[00:57:30] Paul Roetzer: Thanks for listening to the Marketing AI Show. If you like what you heard, you can subscribe on your favorite podcast app, and if you’re ready to continue your learning, head over to www.marketingaiinstitute.com.

[00:57:44] Paul Roetzer: Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.

[00:57:52] Paul Roetzer: Until next time, stay curious and explore AI.[00:58:00]