Marketing AI Institute | Blog

[The Marketing AI Show Episode 69]: AI Agents Will Soon Roam the Internet, 2023 State of AI Report, and Execs Say Half of Skills Obsolete in 2 Years Thanks to AI

Written by Claire Prudhomme | Oct 24, 2023 1:47:42 PM

From lawsuits to the expansion of AI Agents, we are back with even more AI news this week on Episode 69 of The Marketing AI Show. Paul and Mike explore the growing role of AI Agents on the internet, discuss Air Street Capital's 2023 State of AI Report, and delve into a recent study highlighting executives’ worries about today's workforce skills becoming outdated within two years because of AI.

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

This episode is brought to you by our sponsors:

Use BrandOps data to drive unique AI content based on what works in your industry. Many marketers use ChatGPT to create marketing content, but that's just the beginning. BrandOps offers complete views of brand marketing performance across channels. Now you can bring BrandOps data into ChatGPT to answer your toughest marketing questions.

The AI for Agencies Summit is a virtual half-day summit happening on November 2. The AI for Agencies Summit is designed for marketing agency practitioners and leaders who are ready to reinvent what’s possible in their business and embrace smarter technologies to accelerate transformation and value creation. To register, go to AIforAgencies.com and use the code AIPOD50 to get $50 off your ticket.

Listen Now

 

 

Watch the Video

 

Timestamps

00:05:00 — The rise of AI Agents on the Internet

00:16:06 — Air Street Capital just released its sixth annual State of AI report

00:26:00 — Bombshell survey from edX reveals execs' belief that half of all skills in today’s workforce will not be relevant two years from now, thanks to AI.

00:35:08 — Marc Andreessen Techno-Optimist Manifesto

00:42:00 — Upcoming U.S. election indicative of a misinformation disaster due to AI-generated fake content

00:45:42 — Google announces they will defend users of its generative AI products

00:47:22 — Anthropic being sued by Universal Music Group

00:49:07 — DALL-E 3 available to all users in ChatGPT Plus and Enterprise

00:51:04 — OpenAI’s new tool detects DALL-E 3 images

00:52:52 — Descript teases new AI updates

Summary

AI agents could soon roam the internet, working and acting autonomously.

AI agents could soon autonomously do work for us on the internet. These AI agents are autonomous systems that can do far more than chat. Through the use of text prompts, you can instruct AI agents not just to give you a response, but actually go perform tasks for you online by using websites, apps, and tools like spreadsheets to achieve goals.

According to a new report in The New York Times, many researchers think that these agents could replace the vast majority of white-collar work.

We are already witnessing early iterations of this technology. Nvidia has developed an agent that can play Minecraft. Meanwhile, OpenAI has been enhancing ChatGPT with agent-like functionalities, enabling it to interact online via plugins and process data. Additionally, our colleagues at HyperWrite have recently launched a comprehensive AI agent named 'Personal Assistant'.

Air Street Capital releases its sixth annual State of AI Report.

Air Street Capital creates an annual technical deep dive into the top trends driving AI and the market for AI technology. The report typically runs into hundreds of slides and offers an unparalleled look at the state of AI today.

The full report is well worth reading, but here are some highlights that jumped out to us:

GPT-4 crushes other large language models—and humans. Says Air Street Capital: “GPT-4 is the best model across the board. It solves some tasks that GPT-3.5 was unable to, like the Uniform Bar Exam where GPT-4 scored 90% compared to 10% for GPT-3.5. On most tasks, the added vision component had only a minor impact, but it helped tremendously on others.”

Efforts are growing to clone or beat proprietary models like GPT-4 through open-source projects like Meta’s LLaMA.

Although all these efforts could run into roadblocks, they note that it is unclear how long human-generated data can sustain AI scaling trends. (They even say that some estimate that data will be exhausted by LLMs by 2025.)

It was also noted that it is unclear what the effects will be of adding synthetic data into model training. Multimodality is also one of the biggest trends driving growth and excitement in the market.

GenAI is also having a breakout year, with $18 billion of VC and corporate investment.

They also note that “AI is forecast to affect a series of sensitive areas, including elections and employment, but we’re yet to see a significant effect.”

Last but not least, the rate of change and innovation is continuing at a breakneck pace. They note that many high-performing models are easy to “jailbreak” or, get them to do things they are not supposed to do. It is also becoming increasingly hard to evaluate state-of-the-art models consistently as capabilities so rapidly advance.

Executives Claim AI Will Render Half of Skills Outdated in Two Years

According to a new bombshell survey from online education platform edX, executives believe that nearly half of all skills in today’s workforce won’t be relevant just two years from now thanks to AI.

The survey polled 1,600 full-time US-based employees, including 800 C-Suite executives and 800 knowledge workers. Among the C-Suite executives surveyed, they estimate that 49% of all skills that exist in the workforce today won’t be relevant in 2025. And they feel that 47% of their workforce is unprepared for the future of work.

87% of the C-Suite say they’re struggling to find talent with AI skills. 79% of the C-Suite executives fear that if they don’t learn AI, they won’t be prepared for the future of work.

Interestingly, there’s a big disconnect between the C-Suite and employees. 56% of the C-Suite said they believe executive roles at their company could be “completely” or “partially” replaced by AI.

Whereas only 20% of employees think “most” or “all” of their roles could be automated by AI. Additionally, 58% of employees surveyed say they’re taking steps to adapt to AI, but only 24% are using their company’s learning and development programs to do so.

There are more exciting technology and AI updates in the Rapid Fire section of the podcast, including an examination of AI’s impact on the 2024 presidential election, Descript’s new AI updates, and Anthropic’s lawsuit from Universal Music Group. Listen, subscribe, and we’d love your review!

Links Referenced in the Show

Read the Transcription

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

Paul Roetzer: [00:00:00] it's my number one concern right now is disinformation, misinformation, synthetic content being spread.

Paul Roetzer: like it's just going to be a train wreck.

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.

Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.

Paul Roetzer: Welcome to episode 69 of the Marketing AI Show. I'm your host, Paul Roetzer, back again with my co host, Mike Kaput. Good morning, Mike. Morning, Paul. I feel like we start every week talking about travel. I am not traveling this week and I'm so grateful every day on the way to school with my kids. We do a what are you grateful for today?

Paul Roetzer: and mine today was that I do not have to [00:01:00] travel. . For the next like three weeks. So I am home and loving life, being home. That's fantastic. You're home too, right? Like you don. . Okay. I got, I just have two, two one day trips in November. So I am, I'm very much looking forward to time at home and catching up on all the stuff that wasn't getting done while I was traveling all these weeks.

Paul Roetzer: we have an, a jam packed show for you today. There's some really interesting topics to talk about some, new research that's come out. And we have a whole slew of stuff that didn't even make the cut today that we would love to get to, but we would need a two hour episode to do it. . we'll talk at the end, but make sure you subscribe to the marketing Institute newsletter because there's 8 to 10 additional topics each week that we're not getting to that we're including in that newsletter.

Paul Roetzer: So you can find that at MarketingInstitute.Com. Check out the again, check out the newsletter. Mike does a great job with that. It goes out every Tuesday morning along with the podcast. [00:02:00] So we are recording on Monday, October 23rd. It is about 10 a. m. If you, again, context, none of any big news breaks, it's not in here.

Paul Roetzer: That's why, so this episode is brought out to us by BrandOps. Many marketers use ChatGPT to create marketing content, but that's just the beginning. When we sat down with 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.

Paul Roetzer: 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 and apps and action. Thanks to BrandOps for sponsoring again. And it's also brought to us by the marketing AI Institute, AI for agency summit.

Paul Roetzer: This is coming up fast. Mike and I were looking at the calendar Oh my gosh, that is next. So Thursday, Mike, is that what we [00:03:00] decided? Thursday, November 2nd. . So the half day virtual AI for agency summit is happening. It's about five hours of content. There's, I'm doing, AI Emergent Agency.

Paul Roetzer: Mike's got one I'm building around partner programs. There's one on intellectual property. There's one I'm building advisory services around AI, a whole, collection of inside the agency examples. So if you work in an agency or lead an agency, absolutely. Check that out. It's just AIforagencies.Com. You can use AIPOD50 for 50 off. So again, that is coming up November 2nd. It's coming up fast. It should be great. It was fun for me to, to build that keynote. It really kind of put me back in my days of running an agency. So I think most of our listeners know I ran a marketing agency for 16 years.

Paul Roetzer: Mike and I worked together for, I think, nine of those years, roughly. And, we were HubSpot's first partner program back in [00:04:00] 2007. And so I wrote a book, the marketing agency blueprints. I spent a good part of my life, professional life in the agency world. And I've been kind of out of it for the last couple of years.

Paul Roetzer: And so it was, it was sort of fun for me to dive back in and, just the way I approached it, Mike, I think I was telling you was. What would I do today if I was still running an agency? that's how I built my presentation was kind of knowing what I know about AI, knowing what I know about running agencies.

Paul Roetzer: If I was back leading an agency today, what would I do? And maybe we'll share that because it's valuable insight I think for any business today, but certainly for agencies, but maybe in a future episode we'll dive into some of that stuff. But definitely check that out again, AIforAgencies.com

Paul Roetzer: but I want to get to our main topics because we've got some stuff to talk about today, Mike. So I'll turn it over to you. Alright.

nies.

Paul Roetzer: Enable people to interact and to take actions with what we create. Mike Kaput: Sounds good, Paul. First up, we are seeing the rise of AI agents that could soon autonomously do work for us on the internet. [00:05:00] So AI agents, when we use that term, we're talking about autonomous systems that can do far more than just chat.

Mike Kaput: So using text prompts, you can actually instruct AI agents, not just to give you a response, but actually to go perform tasks for you online by using websites, apps, and tools spreadsheets in order to achieve goals. We have a new report this past week in the New York times where many researchers are quoted as saying, they think that these types of agents could eventually replace the vast majority of white collar work.

Mike Kaput: We're already seeing early versions of this technology out in the wild. The New York times profiled NVIDIA, who created an agent that plays the game. Minecraft OpenAI is increasingly adding agent like features to ChatGPT. It can now do certain things like take certain actions online through plugins and do things with data.

Mike Kaput: And our [00:06:00] friends at HyperWrite have actually released a full AI agent called Personal Assistant that is still in the early days, but is able to actually perform tasks on the internet. So Paul, first off, why are AI agents such a big deal?

Paul Roetzer: Well, I think their impact on knowledge work is. Just so unknown, not only knowledge work, but just consumer behavior, online search, like everything, like the way we do what we do today, they, they seem to hold the potential to, to reshape that.

Paul Roetzer: And as we've talked about on this show many times, there's so much uncertainty around just the impact of language models, like on their own, we just look at GPT for, whatever comes next. When we try and project out the impact on work and the impact on consumer behavior, that's hard enough.

Paul Roetzer: But when you throw in the possibility of AI being able to take actions on our behalf to book the trips for us to fill out the forms to string together a [00:07:00] series of actions, that's that's truly transformative to everything we do with devices, and that's really hard to wrap your mind around, and I have yet to really see any research that Does a reasonable job of trying to estimate that.

Paul Roetzer: And in part, because the tech is so early, no one's really seen it applied in a way where you can project out at scale, what it's going to be able to do. It's almost like if you rewound back to like 2017, 2018, when we first got our glimpses of what these language models could do, and in 2018, you're trying to say.

Paul Roetzer: Well, by 2022, 2023, here's what they are going to be able to do. That's kind of the stage we're at here. We're, we're in these very formative days. There's a, a company we've talked about before called Adept.Ai is the, is the URL. They just announced a 350 million round in March of this year, but they have a, a, a [00:08:00] model called act one that is sort of designed to be a foundational model for this, these AI agents.

Paul Roetzer: Just like we have all these foundational models now in language. And so it seems like we're just very much on like the cusp of this stuff working. You mentioned hyperwrite. I know they are early, they positioned it as an alpha product. They know it doesn't work all the time. It breaks a bunch, but it seems like we're just now starting to get a glimpse of what this can mean to how we as consumers interact with websites and information the how we as compa

Mike Kaput: So it seems like the promise is there for these AI agents. It seems to me though, am I right in saying we're pretty early?

Paul Roetzer: Oh, . I mean, that's like I was saying, I think we're just at the foundational levels like hyper rights.

Paul Roetzer: One of the first ones you can test adept. You can just join their wait list like you can't even go and [00:09:00] get access to it and they'll show you a bunch of use cases. But, I think the Minecraft example you mentioned is really interesting because that's being able to see this applied. I think it helps to start thinking about like some practical examples, at least for me, that's always a good way to think about it.

Paul Roetzer: So. Like at HyperWrite and Adept, you can think about any example here or any company, but for example, let's say I want to go find job candidates on LinkedIn. How would I do that right now? I would probably figure out what those candidates look what companies they might be working for. And I'm going to go around LinkedIn.

Paul Roetzer: I'm going to click on a bunch of profiles. I'm going to collect information. so I'm going to put together this, okay, I'm going to take these seven steps. I'm going to conduct this search. I'm going to look for this information. I'm going to copy and paste this over, like whatever I'm going to do.

Paul Roetzer: Imagine you have to do any of that stuff. Imagine a series of these agents that they, they are able to go on your behalf and do that. Go find me candidates on LinkedIn that meet these criteria. And then the agent goes out and it looks through LinkedIn [00:10:00] and it curates all this stuff. And then it comes back to you with a brief.

Paul Roetzer: Here are the 25 candidates I found that met your criteria. You can do the same thing with sales prospects. I was thinking about an example where each year for Macon, I go hunting for speakers. And so you're trying to find people who have spoken at other conferences because AI speakers historically have been kind of hard to find people who have actually done the stuff and are able to tell these stories.

Paul Roetzer: So I go through a list of, okay, go to other event sites, review their agenda, find anyone that did an AI talk, then go to LinkedIn and pull their profiles. it's this very manual process that up until now, there's no other way to do it. The only more efficient way is to get someone else to do it for me.

Paul Roetzer: So I may like have someone on our team or like in big companies, do you start thinking about these really manual things that you're offshoring for more cost efficiency, but I think it could be a massive disruption to that process. Like maybe it's not a lot of [00:11:00] offshoring in the future. Maybe you're just using AI agents to do this stuff.

Paul Roetzer: The example I gave. Back in February, when we talked about this on the podcast episode, when I wrote that world of bits article about how OpenAI was working on this very thing. And the example in that one I gave was in HubSpot to send a single marketing email is a minimum of 21 clicks. And so if you think about the applications to marketing into business in the future, it may just be a prompt, send an email to anyone who's, attended the webinar last week.

Paul Roetzer: And if they are already an existing customer, send them this email. If they are not, like you're just, you're giving a prompt to the machine of what you want. And then the AI agents can go and build the email. They can go take action. So I think that's the biggest distinction here is right now we have models that can output things for us.

Paul Roetzer: They can give us briefs. They can ideate. They can outline things. They can transcribe things like they can do that. But they can't take any action on our behalf. [00:12:00] This is, they are able to go do that. And it's, it is again, really hard to imagine the impact this can have if these work. And I think 2024, we're going to start to get a lot of options to start playing around with this technology.

Paul Roetzer: And it's not going to be very good. It's probably going to be like GPT-1 level, like where the language models were when we first started. But it seems like it's going to iterate pretty quickly. So I don't think a bunch of marketers and business professionals, like next year, you're just going to have AI agents doing your work for you.

Paul Roetzer: But it does seem like we're on that kind of one to three year horizon where these things really start to take shape and impact how we do our jobs.

Mike Kaput: So let's take a stab at maybe talking through some of the implications here with the full caveat that , we're still some years away from these being really, really good, but it does seem like if AI agents reach their potential, we have a lot less need for people doing certain digital work online.[00:13:00] 

Mike Kaput: In the way we've described and that's a lot of people. what do you think as the implications here?

Paul Roetzer: honestly, like it's just so hard project, but when we talk about we commonly say like 80 percent of what knowledge workers do will be AI assisted to some degree in the next one to two years.

Paul Roetzer: And Most of the instance we're thinking about there is actually just generative AI and like machine learning, making predictions. We're not really factoring in AI agents actually doing the work for them per se. And so that's what I'm saying. Like so much of the projections right now about the impact on knowledge work are really just based on the tech we know today that already works.

Paul Roetzer: If you start thinking about this, like it's. I don't know. it's going to change everything. Like I had a conversation with a product team earlier this year, like a SAS company that builds marketing and sales software, [00:14:00] and they were starting to ask questions around well, what do we even need a navigation on our site?

Paul Roetzer: how does this change the way we design user experiences and apps and websites? So I think it's truly like this. Almost like first principles thing where you have to go back to the very basics and say, well, what is the future of? Interactions with websites, what is the future of seeking information?

Paul Roetzer: What is the future of search? Like it's so potentially transformational to everything we know to be true about business that it becomes hard to, to accurately project anything. So I don't know. I don't, I really don't, I haven't played around with hyper rights personal assistant yet. I was waiting until it, it was going to be, viable until you could really get that experience.

Paul Roetzer: And they just announced, like I saw Matt Schumer, who's the CEO just tweeted last week. It was like last Friday that they just released an updated version of personal assistant. And, it [00:15:00] operates on your browser to actually complete tasks at site sources. You can trust what it says. And it kind of goes on.

Paul Roetzer: So we'll put the link to Matt's tweet in the post, but, they have content on their website that talks about this new personal assistant that they say is the most powerful one available today. And so I think like you can go and start to experience for yourself the impact that it might have knowing it's still, again, kind of an alpha product, but it's probably the best way to interact with this technology and start, making some assumptions for yourself.

Paul Roetzer: But I think. Sometime in mid to late 2024, we're probably going to have to start planning for AI agents the way we were planning. In the end of 2023 for generative AI moving into the next year. So right now, if you're in a business and you're not thinking about the impact of generative AI on your workers, on your production processes, on really every aspect of your business across all business functions, you're, you're probably going to be behind next [00:16:00] year.

Paul Roetzer: And I think, come this time next year, we may be in a similar place with AI agents.

Mike Kaput: All right. So next up air street capital just released its sixth annual state of AI report, and this is well worth paying attention to because each year this firm does a technical deep dive into the top trends that are driving AI and the market for AI technology.

Mike Kaput: Now the reports quite a bit to go through, it's these typically run into hundreds of slides and they go pretty deep and are very technical often, but they do offer a really unparalleled look at the state of AI today. So if you want to dive into the full report, it's well worth it. But I did want to extract some highlights that jumped out as we were reviewing it.

Mike Kaput: First up, they point to GPT 4 as this year's kind of clear winner when it comes to an AI model. they are basically showing how it crushes other large language models and, honestly, [00:17:00] humans at certain tasks. They said, quote, GPT 4 is the best model across the board. It solves some tasks that GPT 3.

Mike Kaput: 5 was unable to, like the uniform bar exam, where it scores 90 percent compared to 10 percent for GPT 3. 5. On most tasks, the added vision component so far had only a minor impact, but it helped tremendously on others now. Also, at the same time, they've highlighted the trend of people trying to clone or beat proprietary models like GPT-4, mostly through open source projects like metas, llama, Open source models, but efforts to develop large language models and other foundational models could soon run into roadblocks.

Mike Kaput: They note that it's unclear how long human generated data can sustain AI scaling trends. They even say that some estimate the data that LLMs can use to actually train the human generated data will actually be exhausted. [00:18:00] By 2025, and it's unclear right now what the effects will be. Once we start adding synthetic data into model training now also, air street and notes that multimodality is one of the biggest trends driving growth and excitement in the market, going beyond language in these models and their ability to use different mediums.

Mike Kaput: to create outputs. They also highlight that generative AI as a category is having a breakout year. There's been 18 billion of VC and corporate investment in that space. And they also note that quote, AI is forecast to affect a series of sensitive areas, including elections and employment. But we're yet to see a significant effect now, last but not least, the rate of change and innovation they are profiling in this market is just continuing at a breakneck pace.

Mike Kaput: They note that many of the high performing models right now, despite their guardrails are pretty easy [00:19:00] to quote unquote jailbreak or get them to do stuff they are not supposed to do. And it's also becoming increasingly hard to evaluate. the state of the art models because their capabilities are so rapidly advancing.

Mike Kaput: Now, there's a lot to unpack here, but Paul, it's sometimes hard to remember with everything going on that GPT 4 came out only in March of this year and just quickly Change the game. Can you maybe just contextualize and talk for a minute? Just how fast the AI space is moving

Paul Roetzer: in GPT for it came out in March, but realistically, it was like seven months old when it was released.

Paul Roetzer: So it's well over a year old technology. they probably finished it in summer of 2022 and they had to do some like red teaming and safety stuff with it. But , we're still right now like the most powerful model we have. And granted, it's had some improvements like vision and language or, the audio and things like that have been added to it.

Paul Roetzer: [00:20:00] But at its core, the model that's still leading the world is well over 12 months old. So we've obviously seen just rapid change, though, not only in the models, but in the awareness from businesses, the urgency generative AI strategies and policies. So it is kind of hard to look back and say, wow, that was only March.

Paul Roetzer: I actually recently, when I was building a presentation, I was scanning like our podcast episodes, and it's so fascinating cause you just, even if you just look at the three main topics each week, You get this snapshot and sort of where we've been this year. And it is quite remarkable how much has happened just since, middle of March when GPT-4 came out.

Mike Kaput: So what also jumped out at me in this research is the importance of. multimodality and the sheer amount of money that's pouring into this space now [00:21:00] for people that aren't following AI as closely, some people are tempted to think that AI is just this crazy hype cycle. And there's no question.

Mike Kaput: There's lots of hype around it, but really it is real technology for real use cases that's backed by very real investment dollars. is that your read on it as well? It seems like despite the hype, we have quite a solid market of actual technology being deployed. For commercial applications.

Paul Roetzer: I think that like the VC funding is certainly not at previous levels overall, but it's very hard as we've said before in the show to get funding right now for a software company.

Paul Roetzer: If you don't have a clear AI road map and a strong AI team. I think the assumption right now is. All software is going to be AI software in the not too distant future. And there's a lot of organizations that are struggling to make that change. A lot of these software companies are kind of challenged to evolve.

Paul Roetzer: And I think that it's hard right now to embed on the winners. [00:22:00] And the reality is like 90 percent plus of these startups aren't coming around in two years. so as a business, it's. There's so much innovation, so much opportunity to build AI into your company and to use these software companies that are getting funding.

Paul Roetzer: But it's very hard to project out whether or not those companies will still be around in a year or two. And some of these companies are getting a lot of funding and I have no idea if they are going to be around in a year or two. So it's a, it's a challenging phase, from that perspective, but, we certainly have seen a lot of big money going into some of these big companies.

Paul Roetzer: So of those, I don't remember, you said like 17 billion or whatever, if you start thinking about open, I got 10 and tropics got a couple bill, like realistically, it's probably an 80, 20, where like 80 percent of that funding is going to a very small percentage of, a very small number of companies.

Paul Roetzer: So. It can be almost a little bit misleading. I think it's still very hard to raise money right now. But the [00:23:00] opposite side of that is it also doesn't cost as much to build a company. I think Matt Schumer, who I just mentioned, he tweeted a couple of days ago, like a lot of these companies are just lighting money on fire.

Paul Roetzer: Like there's, you can build AI companies without having to raise a ton of money. If you're not trying to train your own foundational model, like realistically, the only thing that costs a bunch of money is to train these models. Or if you staff up really fast and try and go really hard at specific markets, but the smart play for a lot of these companies is probably to play the long game and don't raise a ton of money and just build and like companies like replit and some of these others we've talked about so many times are making it possible for people with ideas to build things and not need a bunch of money.

Mike Kaput: So kind of based on what we're seeing in this research and what we've discussed, like if I'm a business leader, how should I be thinking about the pace of AI innovation? how should I be thinking about my company's response

Paul Roetzer: to it? I honestly feel like [00:24:00] this can just be overkill. So I wouldn't I wouldn't recommend this research report to a lot of people.

Paul Roetzer: I think it has a lot of interesting findings. I think it, touches on a lot of the topics and research that we cover throughout the year, but it is a curation of a bunch of third party research. And it's a lot of very dense slides, a lot of kind of advanced knowledge that doesn't have a ton of practical takeaways for you as a business person.

Paul Roetzer: And it goes into research, it talks about industry, it covers politics, safety, it makes some kind of very far reaching predictions about the next 12 months that aren't super applicable to you. So I would say if you're more like moderate advanced in your knowledge of AI and you really want to dive in deep to the topic.

Paul Roetzer: Give this report a read, if you're more kind of a beginner level and you're just trying to get surface level and you want like Mike's asking, what's the actionable thing to do? I wouldn't probably spend a bunch of time on this report. It's a lot [00:25:00] to process. I think that what you should do is, pick a couple pilot projects, pick a couple of problem statements and go solve something like The quickest way to get value out of AI is to actually start doing it, experimenting with it.

Paul Roetzer: And again, like this report isn't necessarily going to give that to you. It's not like it, synthesizes all this and says, here's the five key takeaways and here's what you need to do. It's just a dump of research into a couple hundred slides. So again, it's good reference. They've been doing it for, I think you said like six years.

Paul Roetzer: I look at it every year. I rarely do anything with it. it's not, it's not like we're going to talk in the next study about some new research. It's not that, like it's not going to give you a lot of that information. So good, good reference, definitely covers a lot of the trends, but it's a very broad, in terms of its application.

Mike Kaput: So in our third topic you referenced, we have some new research that's kind of a new bombshell survey from [00:26:00] online education platform edX. In it they find that executives that they pulled believe that nearly half of all the skills in today's workforce will not be relevant just two years from now, thanks to AI.

Mike Kaput: Now, this survey pulled 1, 600 Full time U. S. based employees, and it included 800 C suite executives and 800 knowledge workers. And among the C suite executives surveyed, they estimated that 49 percent of all skills that exist in the workforce today won't be relevant by 2025. And they feel that 40%, 47 percent of their workforce is unprepared for the future of work.

Mike Kaput: Now, Some other interesting things that jumped out here. 87 percent of the C suite said that they were struggling to find AI talent or talent with AI skills. 79 percent fear that if they don't learn AI, they [00:27:00] won't be prepared for the future of work. And interestingly, there's. It's somewhat of a disconnect between some of the answers from the C suite and some from employees in this survey.

Mike Kaput: For instance, 56 percent of the C suite said they believe executive roles at their company could be quote completely or partially replaced by AI, whereas only 20 percent of employees think that most or all of their role could be automated by AI. Additionally, 58 percent of employees surveyed say that they are taking steps to adapt to AI, but not that many are using their company's learning and development programs to do so.

Mike Kaput: Now it's unclear if they are not using their learning and development programs and have them available, or if these don't exist in the first place.

Paul Roetzer: They don't exist. I think you can probably… Our research would tell us. . I'm fairly confident that they don't really exist. Yes.

Mike Kaput: So. Let's start off here, Paul, by talking about the big finding.[00:28:00] 

Mike Kaput: Obviously, it's only a single survey. You have to take all these things with a grain of salt. But it is pretty shocking to me to read that, 800 C suite leaders generally are thinking that, half the skills in their workforce will not be relevant in 24 months. What did you make of this finding?

Paul Roetzer: I agree with that. you, the first thing I do when I get to reports like this is I go to the methodology. Who are the people answering these questions? Because there, as we have said before, there's bias in every single research report you're going to read. And so the first thing I think about is, okay, so they've got 800, C suite, 500 of them are CEOs.

Paul Roetzer: Largely based on our experience, most CEOs still have a very Baseline at best understanding of AI. So to ask CEOs to project this sort of stuff out is flawed at best. Like they are just not going to know. However, it is what they perceive to be true. So I think [00:29:00] that there's, do I put a ton of weight in, in the number you said about.

Paul Roetzer: the percentage of skills more than half in the current workforce won't be relevant in 24 months. No, like the number means nothing to me, because it's not accurate. That being said, the fact that these CEOs think that is kind of crazy. I don't know what they would base that on, but like more than half of skills in the current workforce wouldn't be relevant within two years.

Paul Roetzer: I don't, I don't agree with that for one. But I think it's a very fascinating perspective that they do. And I would, I would love to have like a focus group and like dig deeper into it. ,

Mike Kaput: for sure. And that's a really good point. I think what's interesting about some of these findings is just whether they are.

Mike Kaput: Correct or not, directionally showing where people's heads are at, because there is a lot of [00:30:00] interesting perspectives on this stuff that are motivating real behavior, whether it's the correct actions or not. So I found that pretty interesting. Do you see this kind of disconnect between some of the C suite answers?

Mike Kaput: And the employee answers, it seems again, rightly or wrongly, it seems like leaders are thinking way more aggressively about AI's potential to augment or automate work or roles, much more so than their employees are in this survey.

Paul Roetzer: Again, this is like the uncomfortable conversation that we have to keep coming back to.

Paul Roetzer: CEOs, especially if these CEOs work for publicly traded companies or private equity backed companies, they have a responsibility to keep costs down. And so right now, a lot of work gets offshored to keep costs down. If they have a way to use AI to achieve the same outcome, they are going to pursue it.

Paul Roetzer: And so I think [00:31:00] that there's always going to be a disconnect because the objectives are different. The employees aren't thinking about necessarily the same outcomes that the CEOs are charged with thinking about. the fact that there is a massive discrepancy isn't surprising to me at all because they are just coming at it from a different perspective.

Mike Kaput: let's kind of wrap up here talking about that education component. what jumped out to me is that only 58 percent of employees surveyed said they were even taking steps to adapt to AI, which I guess seemed low to me. And we talked about, 24 percent are using company education to make changes.

Mike Kaput: Can we maybe for listeners contextualize just how little education formally exists out there at these organizations?

Paul Roetzer: And keep in mind like this, this research is sponsored by edX. there, there is a natural desire to create [00:32:00] awareness around the gap, but it is a real gap, regardless of who sponsored the research.

Paul Roetzer: To us, it's the biggest gap. it's always, anytime I give a talk, I always start with what do you do next? Education and training. It is the fundamental thing and it's because it can't be centralized with a few people that there has to be wide scale understanding of AI. And I think that overall, I actually thought that was a pretty good research report.

Paul Roetzer: Like I thought a lot of the findings were very valid and I think that. This is one that matters, like it, it's the only path forward, like we can't slow down the technology, we can't, on an individual basis really affect whether it's used for good or bad, like there's, we can go in ourselves pursuing the use of this for good, but really the only thing we have control of at the moment is preparing people, it's AI literacy, it's experimenting with AI, encouraging that culture of experimentation, , [00:33:00] I thought that was important.

Paul Roetzer: There was the couple that like there was one I found is absolutely shocking and it's at 79 percent of executives predict that entry level knowledge work. Jobs will no longer exist as AI creates an entirely new suite of roles for employees entering the workforce. I have no idea what the heck they are even saying there, like 79 percent said entry level knowledge workers won't exist.

Paul Roetzer: So again, you have to when you, if you do read this one in full and I don't, it's a pretty quick read, you can read it in five minutes. Just know that, you can't take some of these things as fact, or even close to facts. They are just opinions, but some of these opinions are kind of crazy.

Paul Roetzer: But it did end up with 79 percent of executives fear if they don't learn how to use AI, they'll be unprepared for the future of work. That is 100 percent true. And then while most executives feel optimistic about how AI will impact them, some feel [00:34:00] overwhelmed by the pace of change. I think that is true, and I think The biggest takeaway for me here is just how CEOs are thinking about this.

Paul Roetzer: And we don't have a ton of research, out there about that. And so I think this is a valuable report, just providing that perspective. And I think that last line there, they feel overwhelmed by the pace of change. It is very true. we hear that all the time. and I know a lot of our listeners probably feel that way.

Paul Roetzer: I've said before, I feel that way sometimes, like there are just days and weeks where It's just too much. And, so I think this is helpful perspective from the CEO.

Mike Kaput: , for sure. And I think a big takeaway for me is related to that here, which is you're really trying to, if you're trying to keep tabs on this stuff and really adapt in a practical way, your career or your role to take advantage of AI and protect yourself and build a competitive advantage, thinking about the differing incentives, CEO versus an [00:35:00] employee is really, really helpful.

Mike Kaput: For you to get ahead of some of these changes and predict where things are going. All right, let's dive into our rapid fire topics for this week. So first up, famed entrepreneur and venture capitalist Mark Andreessen just dropped a controversial article called the techno optimist manifesto. And in it, He lays out the argument that the purpose of humanity is to advance technology, and that by doing so, we have the opportunity to create nearly limitless abundance.

Mike Kaput: Now, some people have seen and praised this manifesto as a really powerful call to arms for us all to embrace technology and build an ultimately positive future for our species, but others, like Kara Swisher, posting on X, and Steven Levy in Wired, have criticized this as kind of the work of And out of touch or maybe even cult like billionaire kind of either wearing rose colored glasses or even actively just talking [00:36:00] his own book.

Mike Kaput: I think the manifesto is worth reading just because of who it comes from and the hype and talk it's generating online. But the part on AI was pretty interesting because they called this out specifically. And it's a little controversial as well, because he says, we believe AI can save lives if we let it.

Mike Kaput: Medicine among other fields is in the stone age compared to what we can achieve with joined human and machine intelligence, working on new cures. There are scores of common causes of death that could be fixed with AI from car crashes to pandemics to wartime friendly fire. We believe any deceleration of AI will cost.

Mike Kaput: lives. Deaths that were preventable by the AI that was prevented from existing is a form of murder. We believe in augmented intelligence just as much as we believe in artificial intelligence. Intelligent machines augment intelligent humans, driving a geometric expansion of what humans can do. So Paul, I think [00:37:00] you had some kind of thoughts on this.

Mike Kaput: It's a very Stark article. He's, I think, intentionally trying to be pretty, controversial in the article. What did you make of this manifesto?

Paul Roetzer: it's very one sided, obviously, like they try and paint this as it's binary. it's one or zero. it's this or that. And it's just not the way it is.

Paul Roetzer: Like it kind of kind of lost me right away because it starts off with lies and truths. And in the lies, it says, We are told that technology takes our jobs, reduces our wages, increases inequality, threatens our health, ruins the environment, degrades our society, corrupts our children, blah, blah, blah, blah, blah.

Paul Roetzer: And then it's, we're told to be pessimistic. And it's well, you should be slightly pessimistic. Like to me, there's just, AI is very nuanced. All of those things that he's saying are lies are in part true. Now, does the good outweigh them? [00:38:00] Probably their view is yes, the good outweighs them, but they are basically saying it's a one or a zero, like it is, or it isn't.

Paul Roetzer: And so I feel like it can be very misleading for people who don't have all of the context about AI and maybe this is like, you've never listened to the show before and you don't have kind of the way we think about things and how we try and present things and more of a neutral environment that lets you process information and make decisions for yourself how you're going to think about these things.

Paul Roetzer: It is a manifesto so it is meant to very much take a stand, but I just, I really struggle. With stuff like this, because I feel like there's just too many unknowns to be so clear and confident in your position right now, and especially when we're looking out to the future, of course, it's going to be wonderful for scientific discovery and medicine, and there's going to be all these amazing use cases, but to just.

Paul Roetzer: belittle or like [00:39:00] toss to the side people's concerns and fears because you don't share them or you're so overly optimistic that those concerns and fears are misguided. It's, I think it's just disingenuous, or it's completely unempathetic. so when I talk to people and they are, they are afraid that's.

Paul Roetzer: That's how they feel like they are afraid for a reason, either because their own experience or what they know to be true. And I just feel like as a society, we just don't listen enough. this is my problem with politics. Like we have these extremes, you have far right, far left, and like but what happened to just listening to each other about the middle?

Paul Roetzer: And I feel like that's what's happening in AI to some ways is we're getting these extreme opinions and I just don't feel like something like this helps that much. Like it doesn't show empathy to the other perspectives and I don't know. I get that they generally they, they hate mainstream media and they are trying to like [00:40:00] combat like the negative headlines about AI and they are trying to influence regulations and I don't know.

Paul Roetzer: It, I had a hard time reading, honestly. Like I got part of the way through the lies and then I jumped ahead to the truce and I was okay. And then I jumped ahead to some other stuff and it just, it did feel very call like to me, like it was just, and I know it's part of this like effective alteration, alteration ism.

Paul Roetzer: Is that how you say it? The E-ACC. So it's just not for me. I don't know, like I. I read it because I have to, because I have to process this opinion. And I think the key here, the key takeaway for me was kind of like the previous research report, where you have to understand the perspective of the CEOs making the decisions. In this case, it's important to understand the perspectives of the people who are building the AI.

Paul Roetzer: And are going to have a disproportionate impact on the future of business and humanity, and he [00:41:00] would fit into that category. So whether I like it or not, this is what he and others believe who are building the technology that is affecting us all. And so it's kind of like politics where sometimes I have to dive in and like figure out what in the world is going on and what are these people doing just so I understand the perspective and then I got to get out and like go back to my regular life because I can't stay in there too long.

Paul Roetzer: that's kind of how I feel sometimes about this. It's just it's just drama and I don't know, so I suggest reading it. I think it's important to understand the perspective. I don't think anyone should. see it as a call to arms and like jump in and be , that's me.

Paul Roetzer: That's how I feel. I'm kind of with Kara that it's a little, it's just a little bit one sided.

Mike Kaput: So next up, a new report in the information [00:42:00] warns that the upcoming us. This election cycle is going to be, quote, a misinformation disaster thanks to AI generated fake content, writes, Sam Lesson in the information, quote, How bad will it be?

Mike Kaput: This cycle is going to make 2016's issues with Russian bot farms look downright quaint. Unlike the techno fever dream of the risk AI. Is a clear and present danger in its ability to destroy trust in the system and create chaos and the author then argues that this problem could cause a misguided regulatory response where once the stuff starts happening at scale,

Mike Kaput: the U.S. Government. will be forced to act aggressively to hold social media companies responsible for all the content that appears on their platforms, which will, in this author's view, drive political discourse to private messaging channels. Paul, this is a topic we've sounded the [00:43:00] alarm on several times. Kind of interesting to see it really logically.

Mike Kaput: gamed out in this article. What did you make of this idea of AI being a clear and present danger to the U. S. presidential election?

Paul Roetzer: It 100 percent is. I don't know if Marc Andreessen agrees or not, or if it's lies, but this is real. Like Again, it's my number one concern right now is disinformation, misinformation, synthetic content being spread.

Paul Roetzer: You could mix in the AI agent stuff that we didn't even, that we talked about earlier today, like you can string that stuff together into this to do, calls ads, like all this stuff, like it's just going to be a train wreck. Like I don't. I, like I generally am optimistic about the outcomes of AI over, long term impact on humanity.

Paul Roetzer: I am not optimistic about the impact on the next election cycle at all. I feel like we're just completely unprepared. [00:44:00] As a society and we're unprepared from a government perspective. I agree they are probably going to take some, some near term actions that a lot of the technologists, the techno, what did, they call themselves?

Paul Roetzer: The techno optimists. Techno optimists won't agree with but I think they are going to have to, I just don't see it being enough. Like I don't. I don't know. I do worry about it. I wish I had an answer to this one. It's like how we avoid this but it's it I think all we can do is educate our friends and family Not to believe what they see on social media like really that is the problem as social media Enables it to spread so quickly and it's going to get so human like at such a rapid scale That education the knowing to trust specific resources, like have your go to resources that actually verify information and just [00:45:00] assume what you're going to see online is not true.

Paul Roetzer: again, we're seeing this already with wars going on right now. Anytime I see someone reshare images of buildings blowing up, it's I have no idea if that's a video game or something from 2018 or someone just made it with Unreal Engine and NVIDIA. I don't know. And so I just don't. Even spend time on it.

Paul Roetzer: And so until I get it from a verified source that it is true, then I just kind of assume everything I'm seeing isn't true when it comes to that. And I think we need that culture, in the coming months, somehow to get people to a point where they, they have skepticism about the validity of what they see until they get it from a verified source.

Mike Kaput: So next up, Google has announced that it will defend users of its generative AI products if they are accused of intellectual property violations. We talked about Microsoft doing something similar, making a similar announcement about a month ago. Google says it will defend customers on two fronts. First, it's going to [00:46:00] defend them from lawsuits related to using models that themselves have been trained on copyrighted material.

Mike Kaput: Second, it will protect customers accused of generating outputs. That violate copyrighted work. So Paul, when we talked about this happening with Microsoft, it was a interesting, but B we were a little skeptical that this provides a ton of confidence to a customer because you're still getting sued at the end of the day.

Mike Kaput: Is that how you see this?

Paul Roetzer: It's interesting. Like they are all I think Adobe's maybe done something similar to, , again, I think our perspective was there. they are obviously getting pushback, probably, especially the enterprise level about concerns around this. And I just wonder if it's going to make a difference.

Paul Roetzer: Like I, if now all of a sudden the legal departments, the big companies are going to say, okay, great. Google is going to cover our costs. I don't know that it's going to make a material difference on adoption rates. It's probably a nice talking point in the sales process, but I don't know, this is probably one we [00:47:00] need to maybe go have a conversation with some internal, inside counsel at these enterprises, maybe, IP attorneys.

Paul Roetzer: Get a better perspective on this one, because it's just not something I personally have any clue how it's going to play out, from a legal perspective, but it's certainly fascinating that all the big companies are going the same route, so they must have something we don't know.

Mike Kaput: So an AI company we talk about quite a bit, Anthropic, the maker of Claude, a chatbot, is being sued.

Mike Kaput: by Universal Music Group. The music publisher claims that Claude, can be used to reproduce copyrighted song lyrics from well known artists when you prompt it to. The music publisher also contends this is a problem because there's, plenty of existing websites that publish song lyrics, but they are saying that The major players that do that are actually paying to license the lyrics.

Mike Kaput: Paul, what did you make of this lawsuit?

Paul Roetzer: I assume the next one we'll hear is Anthropic's going to cover your [00:48:00] claims. If you generate music with their system, it's just the next in the long line of copyright. Lawsuits that we're going to be hearing about. I don't know why they would only pick Anthropic, like I assume the same thing can happen in ChatGPT and others.

Paul Roetzer: Like I haven't tried to replicate it myself, but usually when we see this, it means, okay, , there's probably others coming. Other people are going to join the lawsuit and, get it against other tech platforms. when we talk about these companies, Anthropics, the one we just mentioned, a few weeks back, raised up to 4 million from Amazon and maybe a couple other, or 4 billion, sorry, and a couple other billion potentially coming from Google and others.

Paul Roetzer: And certainly, those billions are going to go to training these models, but you got to assume a billion or so is being set aside at each of these companies to cover all the lawsuits. And I say that Half jokingly, but probably not, I would assume somewhere in their business plan is defensive lawsuits around how the models were trained and, I would guess [00:49:00] they've, they are building up a steady in, in, in house, legal team at all these different AI companies.

Mike Kaput: The next step, DALL-E 3, the latest AI image generation tool from OpenAI, which we've talked about a couple of times, it is now finally available to all users in ChatGPT plus and enterprise. So you can use DALL-E 3 if you haven't been able to before right within ChatGPT plus to produce high quality images in seconds just by typing text prompts.

Mike Kaput: Now, interestingly, DALL-E 3 is built to decline requests. that asks for an image in the style of a living artist and OpenAI also now offers the option for creators to opt their images out from training future image generation models. Paul, we knew this was coming, but just to kind of reiterate, can you walk us through kind of your experience with DALL-E 3.

Mike Kaput: it seems like it's been quite valuable already for you.

Paul Roetzer: I think if it was an episode or two ago, this, we just did this [00:50:00] like episode 68 or 67, you can go back in the archives if you haven't listened to that one and hear kind of the more in detail, but my experience was. Pretty impressive.

Paul Roetzer: I think the simplicity of having it right in ChatGPT, the fact that DALL-E 3 rewrites your prompts for you. So you can, you don't have to be great at prompting. So someone like me who is not a designer, not an artist, I can now all of a sudden get more value out of an image generation tool because it doesn't rely on my prompting ability for the output.

Paul Roetzer: So , again, if you're not paying the 20 bucks a month yet for ChatGPT plus. Pay the 20 bucks for a month and experiment with GPT 4, it connects to Bing, and now you have DALL-E built right in too. You have the vision capability. So , again, if you're looking for pilot projects and you're trying to figure out where to start, just pay the 20 bucks in, the month of November and spend 30 days playing around with ChatGPT Plus and seeing everything it can do for you.

Mike Kaput: Now, OpenAI also says that it's building a tool [00:51:00] that can detect images created by DALI 3 with a high degree of accuracy. So this is according to Meera Murati, OpenAI's chief technology officer, speaking to Bloomberg. Murati said that OpenAI's new tool is quote, 99 percent reliable and is currently being tested internally and that they plan to release it.

Mike Kaput: Publicly at some point. However, there's not currently a timeline for public release. Now, Paul, did this surprise you that we're ostensibly close to technology? It can start creating or identifying DALI 3 created images?

Paul Roetzer: No, because we had a couple episodes ago, we talked about DeepMinds. What is it?

Paul Roetzer: DeepSith, I think, is... What they called it, Adobe has something similar. So it seems like this is a very solvable problem right now in AI is, image recognition, being able to know if AI created it now, I'd love to see more collaboration amongst these firms. If everyone's developing their own solution, it's like Google said that that [00:52:00] DeepMind can determine DeepMind generated images, OpenAI saying we can determine OpenAI generated images.

Paul Roetzer: It'd be nice if. There was a single tool where you could just identify any image, so like a universal standard around how they do this. I, hopefully, maybe that's something that's being talked about behind closed doors and sometime next year we have a tool like that. Don't assume though that this also means we can recognize AI generated text.

Paul Roetzer: Text is way harder than images and I assume video is probably going to be the same way. Video will be easier like images text is not solved for and right now there is no Reliable way for any AI detection system for writing yet People are still are out there claiming that it's possible, especially in schools that turn selling products on it.

Mike Kaput: All right. Our last topic for today is about one of our favorite AI tools, Descript. Descript just teased the first of a slew of big new [00:53:00] AI updates that are coming for its popular AI powered audio and video editor. Now to start, they said they've completely overhauled their AI voice tool. So now you have really powerful AI voice capabilities right in Descript.

Mike Kaput: Notably, You can actually generate AI voices to quote overdub parts of recordings. You were able to do this before, but they have overhauled the AI voice model to be even better. So basically you can just type in, new, new parts of a script over. the human speaker and you don't need them to re record anything, it will create an AI voice clone of them.

Mike Kaput: There's also some new text to speech capabilities. We're seeing this happen across a lot of different tools and trends where we're now able to create very realistic, essentially, voice AI clones of real people. So you'll be able to do stuff like that in Descript as well. Now, this is just the beginning.

Mike Kaput: The company says in the next eight Or 10 weeks, we're going to drop a [00:54:00] whole series of new AI features, all built right into Descripts workflows, all designed to help you find your creative flow as easily in audio and video editing as you can when you're writing. So we'll definitely cover those more, those additional updates once they come out.

Mike Kaput: But Paul, Seems like it's a pretty good sign that Descript continues to innovate. we found so much value in the  tool already

Paul Roetzer:. They’ve, I felt like they were shipping AI features all the time already. And their announcement was , okay, we're finally going to catch up on generative AI now.

Paul Roetzer: It's well, Oh, okay. You haven't been already. So I'm really anxious to see what gets released over the next couple of months. I will say again, like If you're starting from scratch and just trying to figure out what to do with AI, get the 20 bucks a month for ChatGPT plus you have an audio tool. You have image generation, image recognition.

Paul Roetzer: You have language generation, AI assistant. Like it's all there for 20 bucks a month. Pay the 20, 30 bucks a month for descript. If you're doing podcasts and webinars and transcriptions, like awesome, you're going to have a tool [00:55:00] there and then get yourself like runway to experiment with videos. You can build a piloting tech stack of AI tools for under a hundred bucks a month.

Paul Roetzer: Like there's no excuse at any size company not to be playing around with this stuff. So we're big fans of Descript always have been, definitely worth checking out. So that was a lot, Mike, but there is a lot more I mentioned at the beginning, if you're not subscribed to the weekly newsletter, so it's just marketinginstitute.com/newsletter. Or you can just go there and click on resources and it's under there. I'm just going to quick hit. These are things that didn't make the cut. So we have, someone discovered the system prompt and OpenAI that steers DALL-E 3 fascinating read. Clear view AI is built facial recognition recognition system that can be used to find anyone online, find any stranger, find anyone based on their image.

Paul Roetzer: It's creepy as hell, but fascinating. We had an inside story behind how OpenAI dropped work on a new AI model. That was a good read. That was about a model they [00:56:00] were building around the time GPT 4 came out that they had to can. Microsoft CEO released his annual letter. It's all about AI. That was a good read from Satya Nadella.

Paul Roetzer: Meta's new AI technology can read your mind. That one was creepy. But it actually like tells what you're thinking about. it's wild research. Meta recruits striking actors to train AI avatars. . They paid people like a daily fee to basically train, all their emotions and facial, how they manipulate their face and things they say it was wild.

Paul Roetzer: So if you're interested in the Hollywood stuff, Stanford researchers score 10 AI models on how to, how transparent they are a little bit more like kind of advanced reading. New AI app automatically dubs videos in 28 different languages. There's tech you can now do that with lip syncing, Mike mentioned, called LipDub.

Paul Roetzer: And then Inflections AI Assistant Pi is now connected to the internet. So it was not, it now is. As is ChatGVT, as is BARD. So again, all those links and additional reading [00:57:00] in the Tuesday newsletter. So check that out, marketinginstitute. com slash newsletter. Mike does an awesome job each week. And this is kind of like we curate throughout the week, all of our favorite things we're reading and then the best things make the podcast, but each week where the cutting board, there's like five, 10 links that we just can't get to.

Paul Roetzer: , awesome stuff. Mike, as always, thanks for, leading the charge on this and leading the conversation, forces us to have this conversation each week because. Man, like before this, before we did this episode, episode each week, I wasn't, I just wasn't staying as up to date as I could.

Paul Roetzer: So this conversation forces me each week to kind of catch up on what's going on. So appreciate you. Appreciate all of our listeners. We will be back again next week. It'll be, I'll be going into November. It'll be Halloween. Isn't it? Isn't next? , I think so. Okay. Wow. I can't believe we're heading toward November already.

Paul Roetzer: All right. Thanks everyone for listening. We'll talk to you next week.

Paul Roetzer: [00:58:00] 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. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.

Paul Roetzer: Until next time, stay curious and explore AI.