This episode, recorded early so we could get this to you right after Labor Day weekend, focused on three big topics taking the AI news by storm this week: Enterprise-grade ChatGPT, Google’s AI advancements and updates, and copyright lawsuits that put OpenAI on the defensive. Plus, rapid-fire topics you won’t want to miss.
Listen or watch below—and see below for show notes and the transcript.
This episode is brought to you by Piloting AI for Marketers. Complete eight hours of AI education to build a foundation and a framework for your AI journey. Individual and group rates are available. A code to save $50 is mentioned in the podcast!
00:04:02 — ChatGPT Enterprise
00:13:28 — Google AI news
00:27:45 — OpenAI goes on the defensive
00:34:45 — The U.S. Copyright Office wants feedback
00:37:09 — Disney, New York TImes, and CNN block ChatGPT
00:40:39 — Google Search Generative Experience rolls out new features
00:42:57 — LSU is grappling with AI policy
00:46:55 — Grammarly’s new features for students
00:49:27 — AI and a golden era of cinema
00:52:20 — Schumer and the AI Forum
Introducing ChatGPT Enterprise
OpenAI announced they’re launching ChatGPT Enterprise. This is a version of ChatGPT with enterprise-grade security and privacy, unlimited higher-speed GPT-4 access, longer context windows, advanced data analysis capabilities, customization options, and more.
The move appears to be a response to enterprise demand for a safe, compliant version of ChatGPT, says OpenAI. “Since ChatGPT's launch just nine months ago, we’ve seen teams adopt it in over 80% of Fortune 500 companies. We've heard from business leaders that they’d like a simple and safe way of deploying it in their organization.” Now, it looks like they’re getting just that.
New Google AI Updates at Google Cloud Next 23
Google made some big AI announcements at Google Cloud Next ‘23. The event was headlined by Google’s announcement that Duet AI for Workspace, its generative AI tool in Gmail, Docs, Sheets, Slides, Chat, and Meet, is now generally available and has a no-cost trial.
As part of the event, Google also announced new models in Vertex AI, their suite of APIs for foundational models. You can now access Llama 2 and Code Llama from Meta using Vertex AI—and Claude 2 is coming soon. Also mentioned, there is a new digital watermarking functionality for Imagen, Google’s image generation technology. This is powered by Google DeepMind’s SynthID and could give us a preview of how we’ll be accurately identifying AI-generated images and text in the future.
OpenAI disputes authors’ claims that every ChatGPT response is a derivative work
OpenAI has finally broken its silence after being sued by a number of authors, all of whom allege that ChatGPT was illegally trained on their work without permission. OpenAI is looking to dismiss the lawsuits, saying: "the use of copyrighted materials by innovators in transformative ways does not violate copyright."
Unlike plagiarists who seek to directly profit off distributing copyrighted materials, OpenAI argued that its goal was "to teach its models to derive the rules underlying human language" to do things like help people "save time at work," "make daily life easier," or simply entertain themselves by typing prompts into ChatGPT. Citing a notable copyright case involving Google Books, OpenAI also reminded the court that "while an author may register a copyright in her book, the 'statistical information' pertaining to 'word frequencies, syntactic patterns, and thematic markers' in that book are beyond the scope of copyright protection."
Enjoy the episode! It was a busy week in the world of AI!
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: possibility here is that deep mind solves for this and licenses this capability or open sources it for the rest of the industry . So that's what they're trying to do and what I assume they're at least in talks with is you need universal standards for this. And they had a breakthrough in the technology that could allow other image generation companies to very quickly follow on on and have the same standard.
[00:00:23] 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.
[00:00:44] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.
[00:00:54] Paul Roetzer: Welcome to episode 62 of the Marketing AI Show. I am your host, Paul Roetzer, along with my probably exhausted co-host Mike Kaput. You did how many webinars this week, Mike?
[00:01:04] Mike Kaput: Three webinars this week, and we had one the previous week too, so, oh geez. We had what we were calling, Tentatively kinda internally industry week where we had the release of AI for Manufacturing Marketing Blueprint.
[00:01:19] Mike Kaput: So we're doing blueprint assets with select partners. So we did one for manufacturing, the AI for Financial Services Marketing Blueprint, and then AI for Hire Education. So we did three webinars back to back to back. Got a ton of. Amazing attendees and engagement, which is cool. But yeah, it's a bit of a, there's a lot of public speaking over the last few days.
[00:01:41] Mike Kaput: But the cool thing is they're also on demand for posterity here. So if you're interested in checking out one of those, definitely go to marketing ai institute.com, click on resources and under webinars, they're all there. So my hard work over this last week is now on demand.
[00:01:59] Paul Roetzer: I have not really heard from or seen Mike or Cathy for the last three days because they were, and then Mike was building decks before that, so, yeah.
[00:02:09] Paul Roetzer: Yeah. Good stuff. I mean, we heard a ton of amazing feedback and thanks to our sponsors, QuantPlus and Drift on those. Yeah, good stuff. So if you're in financial services, manufacturing or higher ed, go grab those free resources. All right. Today's episode is brought to you by AI Academy. So the piloting AI series in particular, which we've talked about before, I think we have over 600 people have taken this course now, the series of courses.
[00:02:37] Paul Roetzer: So we built this. In December of 2022, Mike and I put this together and launched it. So post ChatGPT as a step-by-step learning journey for marketers to guide them through adopting AI to advance their companies and careers. There's 17 on-demand courses, dozens of AI use cases and technologies, collection of templates, frameworks, and it's all available.
[00:02:59] Paul Roetzer: It's about eight hours of learning. You speed it up to 1.25, 1.5. You can get it done faster, but basically you can do it all in a day. This is about a decade of research that we put into building this series. So it's been extremely well received. It's kind of when we teach our intro to AI class, everyone always says like, what's next?
[00:03:16] Paul Roetzer: This is the series we built. I. So you can go to piloting ai.com to learn more and use AI Pod 50. We'll get you $50 off the registration for that. So again, piloting ai.com and you can learn more about that series. Alright, we are recording this on Friday, September 1st. We're doing it a little early due to the Labor Day holiday on Monday.
[00:03:37] Paul Roetzer: And so we were thinking we could either go Wednesday or let's just get it done on time. So this is coming out as usual on our Tuesday slot. But we are recording on Friday, so if something happens on Friday and it's not in here now, you know why we didn't include it? But we got plenty to cover without waiting until Monday as our usual recording date.
[00:03:57] Paul Roetzer: All right, Mike, let's get going on our main topics. All
[00:04:00] Mike Kaput: right, so first up is that Open AI just announced that they're launching ChatGPT Enterprise. Now this is a version of ChatGPT that has enterprise grade security and privacy. Unlimited higher speed, GPT four, access, longer context windows advanced.
[00:04:19] Mike Kaput: Data analysis capabilities, customization options, and much more. So this move appears to be a response to enterprise demand for a safe, compliant version of ChatGPT. OpenAI actually set in the announcement. Of ChatGPT Enterprise, they said quote, since ChatGPT's launched just nine months ago, we've seen teams adopt it in over 80% of Fortune 500 companies.
[00:04:45] Mike Kaput: And we've heard from business leaders that they'd like a simple and safe way of deploying it in their organization. So now it looks like they're getting. Just that. So Paul, I want to kick things off by just straight up asking about the elephant in the room. Did open AI just make a bunch of third party tools that are targeting the enterprise market specifically obsolete?
[00:05:08] Paul Roetzer: Yeah, that was my first. Thought, and you know, we've known this is coming, they talked about this a few months back, that there would be a business version. It's the obvious thing, related to, you know, Microsoft copilot on its way to come out. We're going to talk about duet AI from Google in a minute. So we knew these big players were going to build enterprise offerings.
[00:05:30] Paul Roetzer: The question just. Becomes moving forward. Do, do you need these other third party applications that are doing similar things? I don't know that we can answer that question yet. You know, we're talking with a lot of enterprises that are looking at all their options. At the moment, no one knows if you can make a single bet on one company because you have some of these third party software companies that are building more enterprise specific capabilities.
[00:05:57] Paul Roetzer: So all this stuff is great. The security, the context window, the code interpreter slash now known as advanced data analysis that keeps struggling to name this thing something that's relevant. I don't, the naming conventions are questionable, but. Are not. So you're going to have that code interpreter data analysis capability baked right in and they teased more advanced version of that is coming, which is going to be crazy.
[00:06:21] Paul Roetzer: But they're not building in yet, like template libraries and the ability to segment by teams and the ability to, you know, build in by persona. Although they did tease roles and functions are coming. So data analysts, marketers, customer support, style guides, things like that, that some of these other players are, it's not HIPAA compliant.
[00:06:41] Paul Roetzer: If you're looking at healthcare, they're not HIPAA compliant yet. So if you're in the healthcare space, you probably can't just rely on this one. So enterprise purchasing is. Complex. There's lots of decisions that go in or variables that go into making decisions. So I don't think we can just say, and I'm sure there's been headlines, like, oh, it's a third party app killer, whatever it, it's not yet.
[00:07:02] Paul Roetzer: So I think that what you're going to do is see a lot of trials, a lot of experimentation, trying to figure out what is it actually capable of, and does it eventually limit the need for multiple tools that do this? But I mean, we're giving guidance right now. That you just need to experiment. Like we can't confidently say, yep, you're, you don't need all those other tools.
[00:07:25] Paul Roetzer: Now. It's just Chad T Enterprise is going to solve everything. Yep. But it's becoming more cloudy. You have Chad, PT Enterprise, we're going to talk about Google Workspace. You have Grammarly go, you have Jasper, you have writer, you have the third, you have the direct with the foundation models. Go build cohere. Go build with like, it's getting really, really, Complicated.
[00:07:43] Paul Roetzer: And I don't know anyone that actually has the answers because I don't know that they're there yet. I do think the path currently is to, have strategies in your company to have teams who are experimenting with all of these advancements and have multiple, vendors that they're testing for different use cases that are most relevant to your company.
[00:08:06] Paul Roetzer: And then we're going to have to see over the next few months how this all plays out.
[00:08:11] Mike Kaput: Did any features specifically mentioned as part of ChatGPT Enterprise stand out to you as particularly important either for ChatGPT specifically or overall for enterprise
[00:08:24] Paul Roetzer: grade tools? Well, the enterprise side in particular is that they're not going to train on your business data or conversations and the models won't, won't learn from your usage.
[00:08:32] Paul Roetzer: That's a huge issue for enterprises to get this through procurement, to get it through secure, you know, cybersecurity and it, and so that, that's just an essential step. I've seen some people like, yeah, I don't really trust 'em. Like there's going to be that, like are they really not? But I think that was a big issue that they had to resolve.
[00:08:51] Paul Roetzer: It's faster. Unlimited GPT for use is a huge deal. If you're, if you're scaling this in an enterprise, you can't be hitting caps every, you know, six hours during the day. The context window, I think it means up to about 25,000 words in each prompt. So that's huge in terms of its capabilities. And then again, I think the code interpreter, formerly known as Code Interpreter, now known as Advanced Data Analysis.
[00:09:17] Paul Roetzer: That to me is probably one of the great unlocks that most companies aren't even really thinking about yet. I think it can be a very disruptive technology once companies learn how to use it properly and especially once the next version of it comes out. So that, that definitely jumped out at me.
[00:09:34] Paul Roetzer: And then they did tease a self-serve version for smaller teams because I have no idea if we even qualify, like they don't have pricing on the site about it. You can only go in and request access, which I did for Marketing Institute, but I've, I assume that we're not going to get it anytime soon. I assume they're prioritizing for like a thousand plus person company or something like that.
[00:09:57] Paul Roetzer: So I don't have no idea if we'll get it, or if we're going to have to wait around for the business. For smaller teams version, they're pretty, they're not sharing those details right now. So you
[00:10:10] mentioned
[00:10:10] Mike Kaput: a bit about. Some of the considerations enterprises might have adopting this type of tool.
[00:10:18] Mike Kaput: Do you have any predictions about where this is going? A little beyond the short term? I mean, it does seem like OpenAI has a huge advantage in the fact that simply everyone knows about and has used ChatGPT over other perhaps less accessible or well-known platforms. Like where do you see this going?
[00:10:39] Paul Roetzer: So if I was OpenAI would buy one of the third party applications and I would build the killer AI writing app. Like they're, they're going to, like right now, they're playing nice with these third party tools and they're letting 'em have access to their APIs. And OpenAI to date probably isn't built to commercialize products like they, you know, this is all new to them.
[00:10:58] Paul Roetzer: ChatGPT sort of threw them into this world of building enterprise software. Go buy somebody who knows how to build enterprise software, just build it. Like if they do that, it, now we're talking about like a total shift in the tech landscape here, because there's nothing stopping them from doing that.
[00:11:17] Paul Roetzer: And if you can just go buy a team that already has built an enterprise platform, enterprise user interface understands the enterprise, if they chose to, they could take over this market. I think, I mean they obviously have the technology for it. It would appear. They're just lacking that kind of go-to market strategy of building the software and the user interface and then having the team behind it.
[00:11:41] Paul Roetzer: I will say I was, there was like this moment yesterday, so I lost access like many people to my ChatGPT plus account yesterday, I went to log yesterday morning, so, so on Thursday, August 31st, and it was just gone chat. My chat history was gone. My account was gone and I was like, did, did me applying for enterprise, like reset my account somehow?
[00:12:03] Paul Roetzer: Like what is going on? So I put it on Twitter and I tagged, Logan, the guy, you know, we've talked about, OpenAI. I don't remember Logan's last name, but he is really active on, Twitter. Shares a lot of stuff at OpenAI. And so I tagged him. I was like, this is a known issue. And he responded right away.
[00:12:18] Paul Roetzer: He is like, yeah, I got the same issue right now. So it was like obviously widespread. And I had actually gone into their customer support, the question, in my account in ChatGPT and like interface with them in a chat. And it was like kind of a push button, like what's your issue? Billing account, whatever.
[00:12:34] Paul Roetzer: So it wasn't like a true interface with GPT four or anything. Despite the fact that I'm sure they were getting flooded, someone actually responded to me within like 20 minutes and I was like, well, this is interesting. They have a customer support like movement at OpenAI. So it just started like that thought in my mind of wow, if they wanted to build like a really impressive customer support engine, an onboarding engine, a partner program, and like acquire or build like an enterprise software play.
[00:13:04] Paul Roetzer: It could be trouble for the ecosystem, for the, you know, the other players in the market. Because I don't know that Google and Microsoft are going to do that. Like, that's they haven't shown that ability yet. I don't think so. If OpenAI wanted to be the go-to like SaaS play here, they could probably do it pretty quickly and they got the money to do it.
[00:13:28] Paul Roetzer: So speaking of
[00:13:29] Mike Kaput: Google, another big topic going on this week is that Google just made some pretty significant AI announcements as part of its Google Cloud next 23 event that took place over the last few days. The event was headlined by Google's announcements that duet AI for workspace. Its generative AI tool in Gmail, docs, sheets, slides, chat, and meet is now generally available and has a no-cost trial.
[00:13:56] Mike Kaput: So as part of the event, Google also announced. New models in Vertex ai. Vertex AI is their suite of APIs that basically allow you to access a bunch of different foundational models. So now using Vertex ai, you can now access LAMA two and Code Lama, both from Meta and Claude two for Anthropic. Access is coming soon.
[00:14:20] Mike Kaput: Now as also as part of these updates, they debuted some new digital watermarking functionality for imagen Google's image generation technology. Now, this is actually interestingly powered by Google DeepMind by something called Synth id, and it basically could give us a preview of how we'll be able potentially to accurately identify AI generated images.
[00:14:45] Mike Kaput: And text in the future. So some big updates here, but let's start with duet AI for workspace. What features of this stood out to you and what should marketers and business leaders be paying attention to
[00:14:58] Paul Roetzer: here? I think it was last week we talked about that. I had just gotten access to the duet AI on my personal Gmail and we didn't know when this was coming.
[00:15:06] Paul Roetzer: Yep. And we, I hadn't thought to look out at the Google Next Conference coming a few days later and assume it was going to happen. But the thing that, you know, again, we've talked about this many times, but you have it in Gmail, you have it in Docs, you have it in Sheets, you have it in slides. Meeting, like it's going to be everywhere.
[00:15:23] Paul Roetzer: I will say for the workspace admins out there, real quick, I went to add this because you could go in and request a free trial. When they announced it, it's like you can go in and get the trial. So I did that and I got an email saying, great, here you go. Like if you're an admin, just. Go add it in your account.
[00:15:39] Paul Roetzer: And so I was playing around this morning trying to figure out how the, how the heck do you actually do it? So this is a quick tip for the Google Workspace admins out there. What you need to do, because it is not obvious, is go to your admin account. Click on billing and we'll put this flow in the show notes.
[00:15:56] Paul Roetzer: Click on upgrade, and then click on Google Workspace add-ons. So that's how you actually then get to the duet ai. Once you get there, you pick start trial of duet ai and this, what it's going to do is once it's turned on, it will be on by default and every user will have it on and they won't be able to turn it off.
[00:16:17] Paul Roetzer: So it will be a, a like a. A constant feature for them in Gmail Docs, everything. So once you do that, the pricing is, there's a trial plan for 14 days. There's a flexible plan, which is $36 per user per month if you want to pay monthly, or there's an annual plan that is $30 per user per month. So that's, that's what we're looking at, which is what the pricing that we've heard about copilot from Microsoft was going to be $30 per user per month.
[00:16:48] Paul Roetzer: So I'm guessing it's all the same, which maybe alludes to what ChatGPT enterprise pricing would look like. I can't imagine they're going to have something totally disproportionate to this. So that gives us a sense of kind of where the pricing is falling. So for me the, I mean, we haven't seen yet how good it is in Gmail and how good it is in docs.
[00:17:09] Paul Roetzer: Those are the two most obvious uses right up front for everybody because every business sends email and creates documents. This, it does get us back to this issue of like, your people need to be trained on this though, so like we're going to just turn these features on. And now in our organization, it's a small team and everyone obviously is involved in AI and understands this stuff.
[00:17:31] Paul Roetzer: So we don't have too many concerns about it. But if you're an admin and you turn this on and all of a sudden like accounting and finance and legal and HR and everybody else has access to these tools and they have no idea what the heck they are. So I would advise like if you are the admin and you're going to go turn this on and you have a company of 20, 30, 50 people and they're all going to come to work on Tuesday after Labor Day, and like they're going to have this help me write button, I'm going to know what the heck to do with it.
[00:17:55] Paul Roetzer: So that, that's just kind of like more of like a admin guidance. So give you a quick rundown. So the way this is going to work is in Gmail there is a help me write button. The the functionality will be, it can write your draft for you. It can formalize a draft, so change the tone. It can elaborate on your draft so it can add further detail to your email to build it.
[00:18:18] Paul Roetzer: It can shorten your draft, so decrease the length of the email or there's an I'm feeling Lucky option, which is introduce fun variations on tone and style for content you've drafted. So as a Google Workspace user, you will have that in both your mobile and your desktop app. In the docs version, the help me write button will again write it for you.
[00:18:40] Paul Roetzer: You can do tone, you can do summarization, so you can generate a summary of the entire doc. You can bulletize, is that a word? That's what they use. Bulletize. It is now. It is. I kinda like that word actually. So as you would assume, it means use bullets to represent your text. You can elaborate, so you can add further detail to build on generated text.
[00:18:59] Paul Roetzer: You can shorten, you can retry, which is have it generate another draft. Or you can customize it with your own instructions. So again, think of all of those individual use cases, you have to teach your team what these are. So what I would be doing right now, and maybe we should be doing this too, is like assign someone who is going to be the lead person testing all this capability over the next two weeks or three weeks or whatever it is.
[00:19:24] Paul Roetzer: Have a plan of how to integrate this stuff into your workflows and to guide people. Just turning these tools on is not going to get you full value from them. You, you have to go through and experiment with what they're going to be capable of. And then again, going back to Mike's original question, does this obsolete, these third party AI writing tools we were using, we have no idea.
[00:19:45] Paul Roetzer: Like until you take, well, here's the five ways. We're using a third party writing tool right now, and now that we have workspace, does it. Solve for those same use cases. Is it complimentary or is it redundant? That's the question you're going to have to ask, but you're going to have to do it at a use case level.
[00:20:03] Mike Kaput: That's really interesting. Yeah, I think that idea of assigning someone to figure this out over the next few weeks is a really smart one. I mean, there's so many ways that. To your point about turning it on, there's so many ways it could go wrong as well. I mean, there's just, people just need a baseline of education, understanding what's even possible with some of
[00:20:22] Paul Roetzer: these features.
[00:20:23] Paul Roetzer: Well, and I think for you and I, Mike, like we have, we deal with AI savvy people all the time. So they're like geeked about this. Yeah. There's going to be a lot of companies who are going to turn this on. For people who know nothing about AI and don't want to know anything about ai, like it's, I retweeted something from Allie Miller yesterday.
[00:20:41] Paul Roetzer: Like the thing that no one's talking about is the change management required to push technology like this into a system that changes like a paradigm shift in talent and in workflows, processes. That's the thing that not enough companies are even talking about right now, more or less solving for.
[00:20:58] Paul Roetzer: And we've seen it, we've talked to a lot of big companies where they're starting to think about it, but it's often happening in the marketing department. Yeah, the change management isn't being envisioned in all the other functions within the business, and that, that to me is like the thing that needs to be a priority in the second half of 2023 because as we talk about this, tech's only going to get faster.
[00:21:19] Paul Roetzer: Microsoft's going to come out with their version. We're going to, we'll talk about Google Meet in a second, but like, we're, this tech is just going to be everywhere all of a sudden. And if you haven't talked about that with your team and put planning in place and training in place, then it's not going to go well.
[00:21:35] Paul Roetzer: Yeah, so you mentioned
[00:21:36] Mike Kaput: what, maybe walk me through a bit about Google meet here that, because I know you had called that out as pretty interesting as well as part of some of these updates.
[00:21:44] Paul Roetzer: Yeah, so there was a section in there launch that I put on LinkedIn. It said to help you better engage during meetings, we're removing the burden of notetaking and sending out recaps, which cool on its surface.
[00:21:56] Paul Roetzer: Duet AI can capture notes, action items, and video snippets. That's a new one I hadn't heard of, like, Capturing video snippets and doing it almost like video shorts that are automated. That's kind of cool, in real time with the new take notes for me feature. And it will send a summary to attendees after the meeting.
[00:22:14] Paul Roetzer: It can even help get late comers up to speed with summary so far, as, as in quotes, which gives us a quick snapshot of everything they've missed. What if you can't make the meeting? This was the one where I was like, I don't know how I feel about this. What if you can't make the meeting and have some input to share with quote, attend for me.
[00:22:33] Paul Roetzer: Duet AI will be able to join the meeting on your behalf, delivering your message and ensuring you get the recap. I'm sorry, Mike, if your attend for me shows up for a one-on-one, you and I have. Yeah. And it just tells me what you wanted to say. It's, we may have some issues, so, yeah, no kidding. This goes to the change management function.
[00:22:53] Paul Roetzer: Like what if executives just start throwing their attend for me duet AI into meetings with like employees and they think it's fun and cool and their employees are like insulted that. Their leaders aren't taking the time to talk to them. Like again, nobody. Is prepared for this from an organizational perspective perspective, from a culture perspective.
[00:23:14] Paul Roetzer: So the tech is going to be here, it's here. And like, what the hell do we do with this? So I, that was, to me, that jumped out. I was like, oh my God. Like, I don't even know about the video snippets and the attend for me thing. I still don't like when people's note taker shows up. Yeah. Like you're on Zoom and you see somebody's like note takers there.
[00:23:30] Paul Roetzer: It's like I didn't give permission to like, send your note taker, like, what are you doing? Yeah. And people just like by default, so I could totally see people just like, Throw in their attend for me in there, and it's like, what is going on? Mm. So I don't know. I think the dynamics of virtual meetings seem like they're just about to change, because I know teams is doing a lot of these same things because Microsoft showed a lot of these functions in their demo video a few months back, and there's no way Zoom doesn't build this stuff in.
[00:23:56] Paul Roetzer: So Right. Meetings are just going to change and nobody's talking about it.
[00:24:01] Mike Kaput: We're also good at regulating meetings as it is, right. So it's, we're going to get, we're going to get flooded with some of this stuff. Yeah. So let's talk about a couple more quick updates. They announced. First up, what is the significance of the updates around Vertex AI and this ability to access some more top foundational models using Vertex
[00:24:23] Paul Roetzer: ai?
[00:24:24] Paul Roetzer: So, honestly, when I saw Vertex AI in the announcement, I was like, what is that? I know I've heard of it. I feel like I should know it, but I don't actually know what the heck Vertex AI is. They're just like, now we have duet ai, we have Vertex ai, so I had to dig in for a second and it is there. They're system that provides APIs to leading foundational models.
[00:24:41] Paul Roetzer: So the the, so you can like kind of prototype and, build things on top of these models. So the, the equivalent that I assume we're looking at is kind of how we talked about a w s bedrock or Amazon bedrock, where you can go in and you can get their Titan model or you can get clawed or you can get.
[00:24:58] Paul Roetzer: Here, I think is in there. So basically like an everything store for models. So I think what Google's doing here is saying, Hey listen, you might not use our language models and that's okay. We just want you to pay us for the compute. Like we want your data in our Google Cloud. And so if you want to use Lama, Code Lama or Claude two or Falcon or like whatever language model you want to use, just keep the data in Google Cloud basically.
[00:25:23] Paul Roetzer: Yeah. So I I, it seems like that's what Vertex AI is. So this probably more of like the tech data conversation, that audience, but just know that everyone is basically understanding that people are going to pick and choose their models and a w s Microsoft and Google want your data in their clouds. because that's where they make all their money.
[00:25:42] Paul Roetzer: So what about this
[00:25:43] Mike Kaput: synth ID thing, this ability to potentially watermark AI generated images and text? Are we close to finally solving
[00:25:51] Paul Roetzer: this? I don't think so. Like DeepMind's talked about this. And again, just quick recap. So DeepMind is the main AI research lab within Google. It was originally an independent AI research company acquired by Google, Google Brain, and Google DeepMind.
[00:26:04] Paul Roetzer: Merged about what, three, four months ago to make Google DeepMind Demis Hassabis. So we've talked a bunch of times, is the c e o of DeepMind and leads AI at Google. So DeepMind has been working on STH id, which we've, image gen is. Was introduced by Google in like April of 2022. It does what DALL-E two does.
[00:26:25] Paul Roetzer: Mid Journey does like give it a prompt that generates text or generates an image. It was never publicly released. You can only get access to it through their a p i to build with it. And it's now going to be infused into like Google Slides, I think, where you can build images within slides with prompts.
[00:26:41] Paul Roetzer: What they've done is they've created a way to watermark outputs from image gen only. This does not work for DALL-E or Mid Journey or any other things, so it's only an image gen functionality and it, it stays there even if the images are cropped and edit it out. So it, it basically is a true watermark within image gen images, so they'll be able to use it to identify ones that were built by Google.
[00:27:04] Paul Roetzer: the, the possibility here is that deep mind. Solves for this and licenses this capability or open sources it for the rest of the industry to solve for image. So that's what they're trying to do and what I assume they're at least in talks with is you need universal standards for this. And they had a breakthrough in the technology that could allow other image generation companies to very quickly follow on on and have the same standard.
[00:27:30] Paul Roetzer: So that's, that's my assumption of why we should pay attention to this is not as image gen widely popular yet. Yet it's that this could be a prelude to the rest of the industry following on. Gotcha.
[00:27:43] Mike Kaput: So our third big topic today is that OpenAI has finally broken its silence after being sued by a number of different authors.
[00:27:51] Mike Kaput: All these authors allege that ChatGPT was illegally trained on their work without their permission. OpenAI is actually now looking to dismiss these lawsuits, and they say, quote, the use of copyrighted materials by innovators in transformative ways does not violate copyright. Unlike, say Plagiarists, who are seeking to directly profit off of distributing copyrighted materials.
[00:28:18] Mike Kaput: OpenAI is arguing that its goal is to teach its models, to derive the rules underlying human language. This is going to help people do things like save time at work, make daily life easier, or simply entertain themselves by typing prompts into chat G B T. Now, as part of this kind of public response to the lawsuits, OpenAI, I.
[00:28:41] Mike Kaput: Also cited a notable copyright case that involved Google Books, and they reminded the court that quote, while an author may register a copyright in her book, the statistical information pertaining to word frequencies and tactic patterns and thematic markers in that book are beyond the scope of copyright protection.
[00:29:01] Mike Kaput: So that's a lot of like a mouthful of a lot of legalese here, but really what OpenAI seems to be saying is, Hey, you're concerned. We've broken all these copyright rules by training our models on your books, but actually the model is not being used to, take revenue from you or profit off you in any way.
[00:29:20] Mike Kaput: It's essentially for the good of humanity and innovation. Now, Paul, can you kind of break down what's going on here? Do they have a case here at all?
[00:29:30] Paul Roetzer: I have no idea from a legal perspective. And like I've often said before, Mike and I are not attorneys. We, I took a business law class in college. That's about the extent I thought about being a lawyer for like, I don't know, a month of college.
[00:29:44] Paul Roetzer: But that's, that's literally it other than the fact that I deal with IP attorneys all the time for my businesses and for our own IP needs. And we are authors, so we have a kind of a stake in this, right? from that perspective. Other than that we are observers of this. And I will say that, like I posted this on LinkedIn and in, in, in, in the way we always do of just like conversation starter.
[00:30:07] Paul Roetzer: We're, we're not taking a stand. I have no idea what the, what they'll eventually say in the courts. I do think this goes to the Supreme Court eventually, not maybe this case, but I think. Solving for this, does eventually go to the highest courts. But I, what often happens is we get, like really smart attorneys who come up in the comments and post stuff.
[00:30:26] Paul Roetzer: So there was actually a lot of, there was a pretty active comment section. I had 42 comments on this one on LinkedIn, and a number of them were attorneys. So I would say, read their comments, read their perspective, but it seems that generally, we just don't know. I think the key to this was that up till now we made some assumptions that the Google Books case would certainly be a part of their argument.
[00:30:49] Paul Roetzer: But this is the first time we're actually seeing them put in print. What their arguments are, and that in essence they're saying like, you know, it doesn't, it's not plagiarism. We're not stealing anything. It's the statistics behind how words are formed and hmm This has been done before and there's nothing we're doing that's different and the laws basically need to catch up to innovation in essence.
[00:31:11] Paul Roetzer: So, Certainly some people are going to say, yeah, that makes a lot of sense. It's fair use, like case closed. And other people are going to say, no, that's not, that's not how this works. And I, it's going to have to be litigated. Like, I just don't know how this gets solved. And they were trying to throw out all but one of the charges, I think.
[00:31:27] Paul Roetzer: Yeah. Like there, the fact that it was actually was a violation of copyright there, trying to get all the other stuff out of it. So I, it is just going to be really intriguing to follow along. But I just thought it was, it was fascinating to hear them. State clearly why they thought what they were doing was okay.
[00:31:45] Paul Roetzer: And I will say the other thing that keeps coming to mind here is so many times, Major breakthroughs in technology are borderline illegal when they're done. So you think about how Uber basically like, took over the taxi industry, like they were basically illegal when they were doing it. What Tesla has done was self-driving in many cases could probably be argued as like illegal, like right, right.
[00:32:08] Paul Roetzer: From how they advertise full self-driving, to actually using that technology on the roads like it. These, the technologists have to push the frontiers. To get to where they're trying to go. And sometimes it's questionable whether it's legal or ethical to do it. And that's certainly what we're seeing here.
[00:32:29] Paul Roetzer: Like you could listen to arguments on both sides. And at the end of the day, the courts are going to decide whether what they're doing or have done was legal or not. and then we'll see. Either they pay fines or they have to license things or they have to destroy the models. I, I've said before, I don't see it coming to that, like I, yeah, I don't think the innovation stops, but I think there's going to be some bumps in the road for sure of some, some sort from a legal perspective.
[00:32:54] Mike Kaput: So obviously we, you've mentioned we need to, these will need to be resolved in court, but I'm curious about what you think this means for people on the other end of this equation. I mean, obviously several authors, publishers, et cetera, are pretty unhappy. I have to imagine seeing a response like this makes 'em even happier.
[00:33:15] Paul Roetzer: I don't know. It didn't change my feelings on it, but I think we're too close to it to be really objective. Yeah. In that we understand both sides and we're living this every day and we have a deep understanding of how the AI works. I think for the average author who maybe doesn't follow AI or is afraid of ai, it just reinforces their hatred for it and that they're convicted that, that people are stealing it.
[00:33:37] Paul Roetzer: So I sympathize with, with that side for sure. Like, I don't have a strong, like what is right and wrong here. I think that it probably was a violation of historical copyright law in some way. And I think there's a chance that what copyright law looks like moving forward evolves and it's no longer illegal.
[00:33:59] Paul Roetzer: I don't know. I mean, I think that's the most likely outcome, but I'm not saying. I think it should be one or the other. I'm saying I'm not educated enough in law to be a person to tell you what I think is right and wrong here. I Right. I just think that's what the judges are for and that's what the courts are for, is to solve for this.
[00:34:16] Paul Roetzer: I think that the companies doing it are making ethical decisions right now about if what they're doing is in the gray area, do they really believe it's for the greater good and they're going to push forward even if it ends up being that they break some laws along the way. Seems to be what's happening right now.
[00:34:35] Paul Roetzer: Let's
[00:34:35] Mike Kaput: dive into some rapid fire topics and kind of relate it on the heels of these open AI lawsuits and responses. Beginning this past week on August 30th, the US Copyright Office has actually opened up Public com, a public comment period around AI and copyright issues. So the copyright office is soliciting people's perspectives on basically three big.
[00:34:59] Mike Kaput: Questions, how AI models should be used, should use copyrighted data when they're being trained. Whether AI generated material can be copyrighted when a human isn't involved, and how copyright liability would or should work. Ai. Now the office is soliciting these written comments that are due by October 18th, and all replies must be submitted by November 15th.
[00:35:26] Mike Kaput: So Paul, when you're looking at this, what did you make of the fact that there opening up this public comment period around copyright and ai, can we anticipate more action in the US at least on this topic?
[00:35:40] Paul Roetzer: On March 16th, when the Copyright office, updated their guidance on Journal of ai, they announced that they were going to do a number of listening sessions and they were going to be open to the public.
[00:35:49] Paul Roetzer: So I don't think there's any surprise here. They have, at least from a PR perspective, taken the right moves to hear people out allowing the public to participate in what the future of copyright law looks like. So, Again, I mean, just from an outside observer perspective, I think they're doing the right stuff.
[00:36:08] Paul Roetzer: They held these sessions, they're hearing people out, they're allowing people to comment. You know, I don't know where it goes from here, but I think it's the right move to involve people in the conversation. and maybe you learn something along the way that, that evolves how you, how you look at this from a legal perspective.
[00:36:23] Paul Roetzer: So, yeah, it's good. I mean, I haven't actually thought about like publicly, you know, creating comments or submitting anything to them. But I'll be really intrigued to find out. So yeah, I mean, I think we have a lot to look forward to this fall. It does seem like going to 2024, there's probably going to be movement one way or the other on this.
[00:36:42] Paul Roetzer: And we'll start learning how the government looks at this. So
[00:36:47] Mike Kaput: as this is all happening with copyright and lawsuits, some companies are taking things into their own hands. So we talked about on a few episodes ago, the fact that OpenAI has recently given people guidance on how to block your website from being crawled by GPT bot, which is the crawler it uses to scan people's websites and use that content for training models.
[00:37:09] Mike Kaput: So we just found out that some major media companies have now. Blocked open AI from crawling their sites. And these include some pretty notable names here. So among a list of companies that have done this include the New York Times, C N N, Disney Reuters, Bloomberg, the Washington Post, E S P N. And a bunch of others.
[00:37:32] Mike Kaput: Now, as we're looking at this, what do you think of these publishers doing this? What does this mean for the quality of the training data that Open AI's models are able to
[00:37:42] Paul Roetzer: access moving forward? I assume it's primarily leverage to negotiate licensing deals. Like, I don't, I don't think that long-term they block this out.
[00:37:53] Paul Roetzer: And we'll talk a little bit about, Google's, you know, search experience, generative search experience in a minute. The future of search and how that all plays out. But these sites depend on traffic from the internet. So I don't know, like I think that it's largely that. But then there's other weird thought came to mind, which is like, we talked about Books three and the 170,000 pirated books, and they needed those books because, You know, if you, if you just train these models on the internet, there's a lot of crappy content, a lot of bad writing.
[00:38:24] Paul Roetzer: So what you really need is examples is really good writing, that's factual and, you know, follow specific styles. And that's what these media outlets have is good writing from humans. So that's great training data that can be heavier weighted in these models. So the question I started pondering, and I put this in one of the comments was like, In some twisted way, is it possible that the language model and tech companies don't save media by bringing their deep pockets to fund journalism?
[00:38:55] Paul Roetzer: So if these language model companies need great writing to train their models, I. Wouldn't it be weird if they realize, well, let's just fund the building of great media companies and let's fund journalism because we need great human writers to train our models continuously. Not like one and done build gpt five, and then we don't need writers anymore.
[00:39:17] Paul Roetzer: What if, because the thing that the journalism industry, the media industry, what they struggle with is they're ad supported. It's all driven by ad dollars. Yep. And so everything becomes clickbait. because you gotta get, you know, the advertisers, you gotta get the impressions up. And like, what if we could just reimagine what that industry is?
[00:39:35] Paul Roetzer: If it didn't have to do that, what if the money actually was there because the output was the ability to train better models. So these tech companies, like for them to throw a billion at great writing to train great models doesn't seem like a big deal with like, how much advertising do you have to sell to do that?
[00:39:56] Paul Roetzer: And so I just, I started thinking about like, I wonder if there's like some play here where these big tech companies start building their own real, like media outlets and journalists like funding journalism schools and like, what if. I don't know, because journalism schools are struggling. You, you know, I came outta journalism school.
[00:40:15] Paul Roetzer: You're a writer by trade. Like they're, it's not in a good place because you don't know what the career path looks like. But if you could actually fund it, like, I wonder if that would make a play. Like, I don't know. I mean, I was just kind of like thinking out loud here, but, yeah, it could be really bad or it could end up being really good.
[00:40:32] Paul Roetzer: I'm not sure which, which direction this goes.
[00:40:36] Mike Kaput: So in another piece of news this week, Google's AI powered search results, which are called Search Generative Experience, or SS g e is now. These are now rolling out links to website within websites within these results. So if you get an AI powered search result, these are like conversational results to queries.
[00:40:57] Mike Kaput: That you give Google. So similar to kind of how ChatGPT would respond if you asked for information, paragraphs of natural language text. Now the links that will now be in these results are going to start appearing as down arrows within the results that you can click to reveal the website, where the information came from, and then navigate it to it if you choose.
[00:41:19] Mike Kaput: Does this feel like a sustainable way to you to kind of offer the conversational search results while still preserving all the things we've talked about with a potential decline in organic web traffic or Google's ad revenue? Does this seem like a way forward? For AI powered search,
[00:41:37] Paul Roetzer: it seems like something worth experimenting with.
[00:41:41] Paul Roetzer: You know, in the article they talked about the different ways Google's tested this to see how people interact with it, what they click on, what they like about it. So kind of like focus group style, and actually like just data-driven style. My impression is we're going to find out pretty quickly, like as this rolls out, I, you know, within two weeks, three weeks, four weeks, I think we're going to start to see whether or not.
[00:42:04] Paul Roetzer: Organic traffic starts dropping. So, you know, in your Google Analytics, mark August 30th is a milestone date, and watch what happens to the organic traffic afterwards. So I think on future episodes we'll be able to do some updates based on industry data of what's going on. It does seem like they made a very, aggressive effort to have the best of both worlds.
[00:42:29] Paul Roetzer: Yeah. The generative response and the links within, and the links to the right. So I don't think Google is trying to like, you know, throw away the traffic they're sending to all these brands and publishers. It seems like they're trying to solve it and I think they're going to keep testing it and trying to figure out the best way to do it.
[00:42:50] Paul Roetzer: But you know, it'd be worth watching your data moving forward. Now, since this move's rolling out,
[00:42:57] Mike Kaput: So on episode 60 of the podcast, we actually talked about AI's potential impact on schools. And it turns out that Louisiana State, U University L S U, is one of the many schools that appear to be struggling to adapt to ai.
[00:43:12] Mike Kaput: So according to a report in the Louisiana illuminator, professors and administrators are trying to deal with AI's impact on the college and. Specifically L SS U saw a 500% increase in reports of plagiarism after popular plagiarism software that they used released a function to detect if AI wrote an assignment.
[00:43:36] Mike Kaput: So Paul, I think we've kind of predicted. Tensions and issues like this would happen, as schools try to navigate artificial intelligence specifically around detecting plagiarism and using AI detectors to see if AI actually wrote assignments. So let's set the record straight here. Can we rely on AI tools to consistently and reliably detect plagiarism?
[00:44:02] Paul Roetzer: No. I mean the, the, well one, I mean, it's not going to be plagiarism per se because it's. Like a plagiarism checker can be pretty accurate. An AI writing detection tool is different and is not necessarily accurate. So as of now, there's no scientific evidence that these things are usable or relevant and they're not going to solve for it in the future because the AI will just keep finding ways to hide itself in essence.
[00:44:30] Paul Roetzer: So to me, it by, by by relying on these tools, you are just. Challenging the cheaters to cheat better. Like I really think that the only path forward for schools, and we're going to keep seeing, you know, case after case that supports this idea is, integration and education like, Infuse the tools into your teaching.
[00:44:54] Paul Roetzer: Teach the teachers how to use them, teach the students how to use them in a responsible manner and just be open about it. So if you're going to use it, show your prompts, but you also have to explain the critical thinking that went into doing it. You know, like you there. There's just no other way forward in education.
[00:45:11] Paul Roetzer: But to accept that these tools exist and to find ways to do it. And it's going to start with teaching the teachers. What they are, how they work, and how best to use them in the classroom. It, it's the only way forward, in my opinion. This is one of those things, like, I actually have a very strong opinion about this and like, yeah.
[00:45:26] Paul Roetzer: I, I've been very consistent with that opinion and having talked with a lot of educators, a lot of, administrators, I'll, I'll tell them the same thing every time. So perhaps
[00:45:38] Mike Kaput: because there's so much confusion or uncertainty around AI and education, open AI has actually released a teaching with AI guide, and this is pretty short at the moment, but it offers some guidance from teachers on things like suggested prompts for AI tools, explanations of how the AI is actually working, and how some of these teachers are actually using it.
[00:46:01] Mike Kaput: In the classroom. So when you saw these guidelines, Paul, did any jump out at you as particularly helpful or
[00:46:07] Paul Roetzer: noteworthy? I. Yeah. I dunno what you, but I kept, like, this was tweeted by Sam Altman. Greg Brockman, yeah. Their c t o, I forget her name at the moment. Like this was tweeted as though, like, this was like sa saving education.
[00:46:21] Paul Roetzer: Yes. And I went to the page, I was like, where's the guide? Like it Yeah. It's like 500 words. Yeah. Like, so I I was expecting like, some awesome thing. So maybe that's just the start of it, but like, I think it's the start of something helpful. I don't, I mean, if you're a teacher and you see this and you go there, and you're underwhelmed, join the club.
[00:46:45] Paul Roetzer: But I like that they're at least thinking about it and starting to do something about it. Yeah.
[00:46:51] Mike Kaput: Another, education focused update here. The popular AI editing and writing tool, Grammarly is actually releasing new AI powered features specifically for students. So this seems to be a suite of generative AI features that they're specifically targeting at students, including the ability to.
[00:47:12] Mike Kaput: Use the tool to brainstorm ideas, offer feedback on drafts of essays, provide revisions, and interestingly generate citations from Grammarly's prompts. What do you think of Grammarly's move here to target
[00:47:26] Paul Roetzer: students with this? When I first saw it, I was like, okay, so. Schools think they're going to keep this out of the classrooms.
[00:47:34] Paul Roetzer: And then you have Grammarly with like, here you go, students like, yeah, it's all these, it just like, that was the first thing I thought is it further validates like, you can't ignore this and just do a policy against it, that all AI is cheating. Like Grammarly and other players are just going to push this at students and so they're going to force a change in behavior, a change in learning, and we just have to solve for it.
[00:47:57] Paul Roetzer: The other one I saw is, U dot com. U Y O u.com. They came out with a student version for like 6 99 a month or something like that. So u.com is also a, like a generative AI tool and search engine. I think, and I haven't played around with it much, but, so everyone's going to come for students and they're going to position their tools as helping students and they're going to position the future of education and schools are going to have no choice but to like, Accept this and kind of come along and solve for it and help these students prepare for the real world.
[00:48:29] Paul Roetzer: And that was when my first thought when I saw grammars is like, well, good for them for like going for it. But it just put, it does put schools in a really tough spot to have to move very quickly and hopefully Grammarly steps up and helps 'em do it. Like, that's the thing is I think open AI and Grammarly and Microsoft can Google, like they have to put.
[00:48:46] Paul Roetzer: Billions into the educational system. Like they have to have commitments. Like we're seeing all this talk about responsible ai. Great, put your money, put your open ai, 20% of your resources into solving for super alignment. But how about we also put a couple billion into solving for the future of education.
[00:49:01] Paul Roetzer: and that's where I would love to see this go, is like, not only are we building products for these students, but let's go help the teachers and administrators solve for this. And not in some like dis dispersed way. Like let's have. Collective efforts that help move this forward this school year. Don't, don't wait for the school year to be a train wreck and then solve next year.
[00:49:19] Paul Roetzer: Like, do this now. You have the money, you have the clout to do it. Like go do it.
[00:49:27] Mike Kaput: So in an interview with the media outlet, Semaphore, we saw the C e O of Runway, an AI company we talk about quite a bit. They make AI powered video generation and creative tools. Their c e O says that AI could usher in a new golden era of cinema.
[00:49:45] Mike Kaput: So the C e o Cristobal Valenzuela said that quote, new types of movies and a volume of content and stories that we might have never thought of before will start to appear. Now this was interesting because also in this interview with all this optimism, he acknowledged, look, this, the disruption that AI might cause in the film industry, especially as we've got these ongoing writer and actor strikes that have.
[00:50:09] Mike Kaput: Ground Hollywood to a halt. You know, some of their concerns there are actually specifically around ai. So Paul, we had both kind of seen this on our radar and kind of found it interesting. How are you balancing or viewing this balance between all these brand new creative possibilities that AI opens up, but also the vast concerns that some of these human creatives and people who do creativity for a living, so to speak, have about the
[00:50:37] Paul Roetzer: technology.
[00:50:38] Paul Roetzer: Runway's a company we obviously have talked about quite a bit on this show. He, their c e o Crisal doesn't I haven't seen him as someone who hypes stuff. Like runway when they hype something they deliver. Yeah. Like their products. Gen one, gen two amazing. Their suite of AI magic tools are amazing.
[00:50:57] Paul Roetzer: He does see around the corner on what this is going to do to the creative world, especially in, in video. And so when he's saying stuff like this, You have to assume that he has a pretty good idea of what's about to happen. I think it just to me keeps coming back to education. Like we can't stop what's happening.
[00:51:19] Paul Roetzer: the tech like runway's, gen two went from four seconds of video from text to video to 18 seconds in like four months. So like we're going to see this exponential curve in terms of the capabilities of this technology. And there is no signs of it stopping in text, audio, video code, anything images.
[00:51:39] Paul Roetzer: So we have to figure out what it means and how it's going to disrupt different career paths, different industries, and we have to start solving for it. the solution is not to stop it from happening. It's not going to stop. So, I don't know. I mean, when I see stuff like this, I just, I always just feel this greater sense of urgency to do more to educate people about what's happening.
[00:51:58] Paul Roetzer: Yeah. And to try and find the opportunities within it.
[00:52:02] Mike Kaput: Yeah, that's interesting. It's, I think one big thing that really resonates with the work that we do is, you know, we say it often. It's not like we have the answer to everything in every industry. It's more giving you the tools to figure out answers on your own in your particular role or domain.
[00:52:18] Mike Kaput: Yep. So last but not least, on September 13th, Senate Majority Leader, Chuck Schumer is hosting an AI insight forand this includes some really heavy hitters in ai, including Elon Musk, Sam Altman of Open ai, mark Zuckerberg, Google, c e o, Sundar, phai, Nvidia, C E O, Jensen Wang, bill Gates, and some other big AI names.
[00:52:43] Mike Kaput: Now, This is actually a closed door bipartisan forand Schumer has previously hinted that it would quote, lay down a new foundation for AI policy. Paul, how significant is this? What's going on here? I.
[00:52:58] Paul Roetzer: So first I would say, I mean, th this room would be insane to be in and to like fly people together.
[00:53:05] Paul Roetzer: Yeah. Send your meet,
[00:53:06] Mike Kaput: send Google meet there to your meeting assistant to listen in.
[00:53:10] Paul Roetzer: Yeah. I wanted to go in person on this one. But second, I think we, we have responsibility to say the diversity of this list is horrendous. Like, yeah. and they got blasted for the initial list, I think. And then they went out and added a little bit of diversity to it.
[00:53:24] Paul Roetzer: I know, Tim Br I saw who was fired from Google, their ethics team back in 2021. I think I saw on Twitter, she said, yeah, I got a kind of a weird last minute invite to this. I assume it's just to balance the room. No. Like no thanks. I'm not there as like a token balance your list of people because you're getting blasted.
[00:53:44] Paul Roetzer: So I've seen some calls that this needs to get shut down before it happens. because of the lack of diversity and the lack of, Like what, how that looks in the industry and then the reality of how it plays out. Like are these really the right people in the room If you're trying to talk about ethics and responsibility, to have a bunch of CEOs whose stake in this is like, they don't want to slow the innovation down at all and it's going to be very one-sided.
[00:54:08] Paul Roetzer: So there are some people they've added to the list recently that, I think we're meant to both balance the diversity of it, but also the perspectives within it. I don't know how you do it. I mean, unfortunately the reality is a lot of these AI companies are. Are run by, specific people and like that's who's going to be in the room, but they have to do more to, to balance it.
[00:54:33] Paul Roetzer: So I think it's good that it's happening, that the idea of the event is happening. I think it's fascinating to get all of these people into a room. I think it'll be fascinating of the efforts by foreign governments to also get in that room through certain channels. Yeah, so it, it again, just like there's going to be so many things like this where there's these, like what may end up being historic meetings, when we look back on things five years from now and like what happened at that meeting that maybe we learn about later on.
[00:55:04] Paul Roetzer: But I'm sure with this group stuff's going to get talked about and leaked to different people and we'll hear some stuff, but, Interesting. I hope that they either reschedule it and expand the diversity or find some ways to get better perspectives, more diverse perspectives in the room, for this first go around.
[00:55:22] Paul Roetzer: All
[00:55:22] Mike Kaput: right, Paul, that's another major. Act week in ai. Thanks for demystifying and decoding it for us. I appreciate it. I know the audience does. Thank you
[00:55:33] Paul Roetzer: once again. Yeah, well go rest your voice, man. Three webinars and a two podcasts in one week. Yeah. Yeah, it's a lot. And I know you had a bunch of other stuff too, so thanks for all you're doing as well.
[00:55:42] Paul Roetzer: All right. Well talk to everyone, next week.
[00:55:46] 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. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.
[00:56:08] Paul Roetzer: Until next time, stay curious and explore AI.