Generative AI has so many benefits for marketers. But with the rapid pace of adoption—often with little to no oversight–-issues are quickly arising.
From educational concerns to legal ramifications, Paul and Mike discuss this on this week’s Marketing AI Show podcast. Also, stick around for an update on the developments between OpenAI and Microsoft.
First up, higher education has been rocked by ChatGPT. Professors and educators need to rethink assignments when it comes to essay writing, take-home tests might become a thing of the past, oral exams will be on the rise, and identifying plagiarism has a new twist. Schools are banning the use of ChatGPT in assignments, but how can they truly know how the assignment was completed? And should educators teach students how AI can augment their learning? One student has created a tool to identify content generated by ChatGPT, but is that really the solution?
Next, generative AI is having some legal troubles. Stability AI has received formal notification of impending litigation. Intellectual property, derivative works, and copyright violations are all discussion points as these technologies advance and learn. As the line isn’t clear, and precedent has not been established, the court cases will continue to mount. GitHub and Midjourney are in the thick of this as well. Once fast to market, Reuters reported that Sam Altman says OpenAI’s GPT-4 will launch only when they can do it safely and responsibly. “In general we are going to release technology much more slowly than people would like. We’re going to sit on it for much longer…” That’s a good thing, but is it too late?
Finally, we discuss a new development regarding Microsoft and OpenAI. On Microsoft Azure’s website, they announce, “Today, we are excited to announce the general availability of Azure OpenAI Service as part of Microsoft’s continued commitment to democratizing AI, and ongoing partnership with OpenAI.” As part of this, DALL-E 2 and ChatGPT can be integrated into their clients’ cloud apps. Developers are currently required to apply for access, and have to describe their intended use cases and applications before they can get access.
Catch up on the latest news and think about these developments and what it means to you and your business. Listen to the podcast below or in your favorite podcast player.
00:02:53 ChatGPT and education
00:17:10 Generative AI and the law
00:29:37 An update on Microsoft and OpenAI
Links referenced in the show
- Piloting AI for Marketers Course Series
- Marketing AI Conference (MAICON)
- Alarmed by A.I. Chatbots, Universities Start Revamping How They Teach - The New York Times
- Paul's post on AI and education
- Will ChatGPT Kill the Student Essay? - The Atlantic
- ChatGPT Will End High-School English - The Atlantic
- This 22-year-old is trying to save us from ChatGPT before it changes writing forever : Planet Money : NPR
- NYC bans access to ChatGPT on school computers, networks - Chalkbeat New York
- Getty Images is suing the creators of AI art tool Stable Diffusion for scraping its content - The Verge
- Stable Diffusion AI art lawsuit, plus caution from OpenAI, DeepMind | The AI Beat | VentureBeat
- Measuring Fair Use: The Four Factors - Copyright Overview by Rich Stim
- This Copyright Lawsuit Could Shape the Future of Generative AI
- General availability of Azure OpenAI Service expands access to large, advanced AI models with added enterprise benefits
- Microsoft launches Azure OpenAI service with ChatGPT coming soon - The Verge
Watch the Video
Read the Interview Transcription
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: I think by the end of 2023, you're not hiring someone unless they're AI savvy, like you're, you either have your internal training program to make them ai. Or you are actively searching for practitioners and leaders who not only are good at their discipline domain, but understand artificial intelligence and can help move your organization into the future very quickly.
[00:00:20] 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.
[00:00:40] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host.
[00:00:48] Paul Roetzer: Welcome to episode number 30 of the Marketing AI Show. I'm your host, Paul Roetzer, along with my co-host Mike Kaput, chief content Officer at Marketing AI Institute, and co-author of our book, marketing Artificial Intelligence, AI Marketing in the Future of Business. What is up, Mike? How's it going, Paul?
[00:01:04] Paul Roetzer: Well, considering we pick three topics each week, and I think as of this morning we were locked in and I feel like we already changed one or two of 'em. It's just like we say every. It is insane how quick things are moving. And I actually thought today, this had been a slow week. .
[00:01:22] Mike Kaput: I was, uh, doing research and realized one of the sources I was using had published their article like 20 minutes before I was doing it.
[00:01:30] Mike Kaput: So yeah, we're on the cutting edge this
[00:01:32] Paul Roetzer: week. Wow. Yeah, it's, it's wild as always. So we'll jump into that in a second, this episode. Brought to you by the piloting AI for Marketers Online Course series. We've been talking about this the last couple episodes. If you've been listening, uh, Mike and I actually created this series.
[00:01:48] Paul Roetzer: It launched in December, 2022. It's meant to be a step-by-step learning journey for marketers to guide them through adopting ai. To advance their companies and careers. We're hearing from a lot of corporations that are actually looking at this as an internal training program for teams. Uh, if you're a, a marketing leader, it's perfect for that.
[00:02:06] Paul Roetzer: We can actually talk to you about guided learning options where Mike and I can come in and, uh, even, you know, do ask me anything and kind of help build an internal training program. So the series is 17 on demand courses, there's dozens of use cases and technologies, templates, frameworks, pretty much everything you need to really develop a comprehension, a strong comprehension within your organization of AI and figure out how to, uh, pilot and scale it in your organization.
[00:02:31] Paul Roetzer: So visit piloting ai.com to learn more, and you can use AI Pod 54 $50 off registration. So again, that's piloting ai.com and I'm gonna turn it over to Mike for. Three hot off the presses topics that again, we just like shuffled around this morning.
[00:02:51] Mike Kaput: All right, thanks Paul. So first up we're talking chat, G P t disrupting education.
[00:02:58] Mike Kaput: So we've seen a ton of reports over the last past of the last week that school administrators, college professors, administrators, educators, and parents are all scrambling to deal with the fallout from chat G B T. And by fallout we mean there's. Absolute explosion of students at all levels of the education system using chat G P T to write essays, answer exam questions, complete in-depth assignments.
[00:03:27] Mike Kaput: I mean, they're probably using it for a hundred other things too. And honestly, some of these popular academic use cases are going viral on sites like TikTok and Reddit, encouraging even millions of more students. To experiment with the tool and in response, there's been a variety of somewhat chaotic and uneven ways that schools are trying to adapt to this new reality.
[00:03:51] Mike Kaput: So some public school systems like New York City and Seattle have just outright banned the tool. Other educators almost overnight have actually completely redesigned their coursework in response to chat G P T. So they're doing things already, like prioritizing oral exams and handwritten assignments to get around what can happen when chat G P T is actually doing your homework for you.
[00:04:19] Mike Kaput: and even other professors and educators have started to experiment with and adopt new and untested AI detection tools and an attempt to crack down on chat G P T generated content. So the reason we're talking about this is not only is it a huge issue in society, but it raises all these bigger questions about how we educate students at every level in a world where AI can write and respond well enough to fool educators and.
[00:04:47] Mike Kaput: Downstream of that, what does that mean for the next generation of talent coming into industries like marketing or industries across different types of business functions? So I wanted to start off by asking you, Paul, is banning or regulating Chachi PT in schools and universities, is it the right approach or why or
[00:05:10] Paul Roetzer: why not?
[00:05:12] Paul Roetzer: Listen, we're we're not administrators, educators. It's not what I do for a living. Um, so I am not gonna pretend to know all the intricacies and challenges that these schools from middle school all the way up to, you know, higher education have to deal with to drive change. At the same time, I've spent a lot of time with friends, deans of schools, presidents of schools, professors, principals, um, school presidents.
[00:05:42] Paul Roetzer: Grade school and high school level. Uh, enough time to understand that there are lots of complexities to this so we can sit out on the sidelines and question schools for not moving fast enough and, and say what seems like the obvious thing to us. So I just wanna put in the context that we're not pretending to know the real difficulties that schools have to drive change.
[00:06:08] Paul Roetzer: Um, I have probably spent the most time at the higher education level, um, you know, with universities talking about the challenges, and this is even before AI of. The traditional way things are done and taught how the different systems work, um, how tenure works. Like there's a lot of things that schools have to deal with.
[00:06:29] Paul Roetzer: And so it's not as easy as saying, oh, you should be thinking about AI and infusing it into your classroom. You don't just flip a switch and do stuff like that at universities. So about a month or so ago, this was, I think right after Chad G P t, I put, uh, something up on LinkedIn where I sort of posed some questions.
[00:06:46] Paul Roetzer: It was like one, I was thinking about it as a. So I think about my kids are in, uh, fifth grade and fourth grade, and I do wonder what should they be taught about ai? I wondered sometimes myself, how much I should teach them about it when they have homework assignments at night. Should I be showing them how to use AI to do the assignment or not?
[00:07:04] Paul Roetzer: So I have these questions as a parent, which then leads me to, well, what should my kids'. Point of view, be like, what should they be doing? Are they even thinking about this? Now, luckily my kids go to an amazing school and I'm friends with the president of the school and we have these conversations, so I know that they're forward thinking and looking at this.
[00:07:22] Paul Roetzer: But even then I would challenge them like, we get your peers involved. Like we, you need to be. Like this conversation needs to be wider spread. You're not gonna solve this on a one school level. This is a, a institutional system level. So the, the, the post I put on LinkedIn was one, as a parent saying like, I don't know the answers to this, but I, I, I think we need to infuse this into the kids' learning cuz it's going to be a part of their life.
[00:07:43] Paul Roetzer: The second I was kind of acknowledging that whether you're at grade school, high school, or college, You're going to be faced with some very difficult questions, and I think those are just getting amplified. So I'll just run through those real quick and then we can kind of hit on those a little bit. So if I'm an administrator at college, a dean, president, whatever, And I'm looking at saying, okay, we have to move fast.
[00:08:04] Paul Roetzer: We have to think about the experiences and the curriculum and how we're gonna do this and, and the impact it's gonna have on the, the value of a degree from this university one year, three year, four years from now. So these are the questions I pose. Does anyone in the school understand AI and how it's changing education?
[00:08:19] Paul Roetzer: That on its own is like a pretty loaded question. Cuz the answer is probably no. Like there are, I'm, I mean, I'm not going to generalize, o overly generalize here, but most AI knowledge in universities is centered within the computer science department. It is. It is treated as a technical discipline. There are very few professors and educators I have come across who are treating AI as a business discipline, and that to me has been the fundamental flaw for the last five.
[00:08:46] Paul Roetzer: And I have talked to a lot of universities about this, that my belief years ago was there needed to be an AI 1 0 1 course at every university, and it was not just a business class. It was every single major should take an AI 1 0 1 course, that they should understand the fundamentals of this technology and how it was going to disrupt their career path.
[00:09:07] Paul Roetzer: I don't care if you're an artist, a. A salesperson, a, a chemical engineer, AI's gonna be in all of it. And this has been obvious that this was coming to us, at least to us for years. So we've been having these conversations. So now if you're a university, you have to flip the switch and figure this out. Who are you going to, like, who in the university can do that?
[00:09:26] Paul Roetzer: So then, okay, let's say what is the school's point of view on ai? What's its policies on its use in and out of the classroom? So that, that's the exact question you posed. Like I don't know the answer to that, but it seems. Very close-minded to me to say, let's just not allow it. Next is if we want to teach ai, not the technical stuff, but the business society stuff, who even knows the topic well enough to lead a class on it?
[00:09:50] Paul Roetzer: Hmm. Where are you going to get professors that can do anything other than just show the stuff and at a cursory level explain it? Who, who actually understands the technology and the fundamentals and its impact on business? How do we keep up with the rate of change? Universities are not structured to move quickly on this stuff.
[00:10:08] Paul Roetzer: Hmm. Uh, what are we doing today? Curriculum experiences, testing that will be obsolete in the next three to five years. How do we adapt our take home? Essays dead. The answer to that one is yes. I put How does using AI affect development of critical thinking? What skills do we need to be teaching? What remains uniquely human in the age of ai?
[00:10:25] Paul Roetzer: These were just like off the top of my head, three minutes writing a LinkedIn post. Questions like if you sat back and thought. The future of education and the role AI is gonna play, you have to start projecting what do career paths look like in five years? Like if you have freshmen right now in any discipline, what in the world does their job look like when they get outta school in four or five years?
[00:10:49] Paul Roetzer: Like, it's, it's really hard for the best AI researchers to project out that way. So it's, it's a very challenging time. So again, I, I don't wanna. Pile on educators and administrators like it is, they have an insanely difficult task ahead of them to figure this out. And it's something that I care deeply about.
[00:11:09] Paul Roetzer: Like I've stayed very well connected to Ohio University, which is where I graduated from. I've, I've spent a lot of time with the business school, the communication school, the journalism school. Um, the president's a friend of mine. He is a wonderful president, you know, very visionary president. And they've been thinking a lot about this stuff and, and looking at it for years.
[00:11:27] Paul Roetzer: Um, Harvard, I know our friend Dave Edelman, who is actually one of the first people to review our book. He's building a course on this stuff at Harvard Business School. So it's like there are universities moving on this. We used, uh, university of Florida as a case study in our book. We talked about them in like chapter 15 or 16 about is this the model for the future of education?
[00:11:46] Paul Roetzer: Where they had deal with Nvidia, where they had supercomputer on campus that was actually, you know, accessible to any, you know, discipline. So I just, I have a lot of thoughts obviously on this, but they're not all formulated. Like, I don't sit around all day trying to solve the future of education, but I've had these amazing conversations over the last five, six years with people who were seeing this coming, but in some cases couldn't within the confines of.
[00:12:13] Paul Roetzer: Education works, move as quickly as they wanted to, maybe to to drive change. And I think chat, G p t just made it where you have no choice now, like mm-hmm. , it's here and we have to go now. We can't just talk about it anymore.
[00:12:29] Mike Kaput: So I realize we're talking about kind of first draft thinking here, and we don't have this all figured out and it's really hard to figure out.
[00:12:36] Mike Kaput: But let's take a stab at this. I'm curious kind of if I'm an administrator, let's say at a university specifically, let's say specifically in the marketing or business program or school. What do I do? What should I even be thinking about? Because chat, g p t, like we've talked about, is just the tip of the iceberg.
[00:12:55] Mike Kaput: You could solve for that tomorrow, and you still have not solved for overall AI education and curriculum and training to prepare students for the
[00:13:06] Paul Roetzer: workforce. So, day one to me is an AI 1 0 1 course. I mean, I'll go back to what I said years ago. I, I think you have to have one at universities and if, if you don't have one, that's my number one priority is to find someone who can teach.
[00:13:18] Paul Roetzer: Now, I would hope it's gonna be the most popular course on campus by next year. So you may have to have multiple of them . Um, but I think you have to start with the fundamentals and you have to treat this as a, a cross-disciplinary, uh, area of study. It is, this is not a pure business play. It is not a pure technical play.
[00:13:37] Paul Roetzer: This is, every industry needs to understand this stuff. Every graduating, um, person needs to understand artificial intelligence and its impact. So I start. The other thing I proposed to a a few different people was I would audit your alumni group and I would find anyone working in artificial intelligence and I would invite them back to campus for a symposium.
[00:13:58] Paul Roetzer: So I would, I would involve alumni who are working in the space, whether they're. Practitioners on the technical side, or if they're business people or researchers or whatever it may be. So hopefully if you're, you know, a lot of larger universities are obviously gonna have smaller universities may struggle to find, you know, grads who are working in ai.
[00:14:16] Paul Roetzer: But it's a simple LinkedIn search. Just all people who graduated from university X and uh, have AI in their title or ML in their title. and I would get them on campus in like now, and I would just have think tanks. I would talk about where is this going, what should we be doing? And I'd be looking at curriculum and experiences.
[00:14:36] Paul Roetzer: Um, Hugh Sherman, the president at Ohio University, to his credit in 2018, brought me down to campus, like talk about visionary from a university perspective. And we had a half day workshop with all business school faculty and administrators on what is ai, what's its impact on education? And then we did hours of group brainstorming and sharing.
[00:14:59] Paul Roetzer: Around how to infuse AI into the curriculum and experiences for students. So I think that was a wonderful starting point for Ohio University. I think other universities should be doing the exact same thing. Um, In their case, we did it the day before school started. So everyone was back on campus. Uh, the, all the professors and administrators, but they didn't have any obligations.
[00:15:19] Paul Roetzer: So it worked out perfect. So I would be looking at ways to be running those kind of style workshops or summits for the faculty and staff to involve them in one that they understand it, cuz don't assume they do. Most don't. So one, give that education, and we were, I mean, Finance, hr, sales, marketing. We had all these people across all these different areas of, of business in the room.
[00:15:44] Paul Roetzer: And so like once you teach the fundamentals, their minds light up with all the ways it's gonna impact the, the major that they're teaching. You know, accounting, finance, legal, uh, business law, you know, that kinda stuff. And that's what you wanted was this room of people we're like, oh, well, in business law it's probably gonna do X, Y, and Z.
[00:16:01] Paul Roetzer: We should be thinking about this. It's like, yes, perfect. And we just whiteboarded everything and then synthesized all those findings. So I would start conversations and I would involve alumni, and I would involve faculty. But the core to all of it is they have to understand it first. This is the problem.
[00:16:18] Paul Roetzer: We're having a chat G P T right now. So all these people racing ahead using it, pretending like they know what they're talking about with AI because they use chat G P T for a week and they have no understanding of the fundamentals of large language models or AI as a larger discipline. and we can't have that at higher education.
[00:16:35] Paul Roetzer: The, the professors have to understand AI at a deeper level. They need AI won't of course, themselves. Yeah. Then they can teach this in, in the application of their discipline. So I do think that there are very practical steps that can be taken, like if you're in journalism or communication school, immediately figure out how to infuse chat.
[00:16:52] Paul Roetzer: Chippy t in writing tools into the curriculum now, like this semester. Don't, don't wait around. But at a higher level, I would be looking at these fundamental steps of involving people in the future of education, because that's what we're talking about here. That's what's at stake.
[00:17:07] Mike Kaput: That makes a ton of sense.
[00:17:08] Mike Kaput: And you know, our second topic here shows just one example of why. This is so important to understand at a cross-disciplinary level, and not just for your engineering students or your engineering professionals or your STEM people, because our second topic today is about kind of the real world consequences that people are figuring out related to some of these AI tools that businesses are using.
[00:17:35] Mike Kaput: So a handful of major generative AI companies, companies that generate text, image, video, et cetera, using AI models. Are getting sued for their use of copyrighted or protected material in training those AI models. So this is a development that has pretty significant implications for anyone who is using tools that use AI to generate texture images.
[00:18:01] Mike Kaput: So one big impending lawsuit comes from the stock photo company, Getty Images. And they are threatening legal action against stability. AI Getty claims stability. AI used millions of copyrighted stock voters and images to train its stable diffusion image generation model stability. AI is unfortunately also being sued by three artists who are lobbying similar copyright infringement claims.
[00:18:30] Mike Kaput: And they're also, as part of that lawsuit, suing Mid Journey, another AI image generation company. And on top of all of this GitHub, which is a code repository, it's very popular among programmers and the company that owns it, Microsoft are being sued over a tool called Co-Pilot. Now, co-pilot is a very popular AI tool that generates code for you.
[00:18:54] Mike Kaput: It's also a generative AI tool. And the suit alleges that co-pilot was actually trained on GitHub's public code repositories. And some of those, uh, code repositories are actually protected by open source licenses. And you know, as an aside, open AI actually is also named in this lawsuit because their models.
[00:19:16] Mike Kaput: Power co-pilot. Now, all of these are pending and just kind of getting started as legal actions, but they're really kicking off this aggressive public debate about the fair use of content that is being used to train AI models. And these models, these handful of really seminal important models are being used by hundreds, if not thousands of AI startup.
[00:19:41] Mike Kaput: Companies and projects. So they're all relying on these models, which are now being challenged in court about their use of copyright material. So I think it's still too early to say which way the wind will blow on this, but regardless. Of outcome, the cases are going to set important precedents and have pretty far-reaching consequences for anyone who uses ai.
[00:20:05] Mike Kaput: And they're probably, I would think you would agree just the beginning of what, it will probably be a wave of legal actions. So what are your thoughts on the overall copyright concerns being raised here?
[00:20:21] Paul Roetzer: Not, I'm not a lawyer. We'll start off there, nor do I pretend to be one. Um, I did take business Law a really long time ago.
[00:20:31] Paul Roetzer: It was actually a fascinating class. Uh, I thought about being a lawyer for about three weeks before I went into journalism . Um, anyway. So I, I see two major issues here. Uh, well probably more than that, but we'll start with the two. So the first is the training data. This is, this is a known issue that the language models are trained on data that they probably didn't get permission to train on.
[00:20:51] Paul Roetzer: And the image generation models also are trained. The video models will be the same any, anywhere where this occurs, there's going to be this question of where did they get the training data and did they have permission? So you're seeing some backtracking now where like, stability, ai, CEO is actually not common in specific on this case that I'm aware.
[00:21:07] Paul Roetzer: Or the suit, but at least in general, uh, said, Hey, we're gonna find ways to let people opt out of our next model. Mm-hmm. where it, you know, it's basically like a do not train on AI with my content sort of thing. You're seeing stuff like, um, Shutterstock I know has a deal with open AI where they're. I believe trying to compensate the original artists or the training, I don't pretend to understand how that would work or how it would ever, ever be an equitable exchange for training.
[00:21:39] Paul Roetzer: Um, so on the training side, the arguments I have heard, uh, is let's say I go read 10 articles about generative ai and then I synthesize that learning and I write an article about generative. You know, here's 10 things to know about generative ai. Now I'm combining the 10 articles I wrote into my knowledge base to produce that article.
[00:22:04] Paul Roetzer: Hmm. The argument you'll hear from Open AI and others is what's the difference if the AI goes and reads 10 articles and synthesizes its findings and it writes the original content based on those 10 things? What does it need to cite them for? It's not, it's not citing specific information, it's synthesizing learning, which is what we do as humans.
[00:22:26] Paul Roetzer: Hmm. So the Supreme Court case that seems inevitable to me, could come down to this very argument of like, what is the difference between how the AI learns and how the human learns? Um, the AI is not copying and pasting, it's not stealing anything. It's not scraping anything. It's learning. If I wanted to paint like Picasso and I had any artistic ability whatsoever, which I don't, I could just go study the entire library of Picasso works, and then I could create an original image in the style of Picasso.
[00:22:56] Paul Roetzer: Am I, am I plagiarizing? Am I stealing, am I infringing on anyone's copyright? I, I don't, I don't believe I am in that instance. So again, what is the difference? I'm not saying I'm in that camp. I'm not saying I agree with. I'm just saying this is the argument you're going to hear and why lawyers exist to fight these things, uh, and why the Supreme Court exists to decide these things in the United States.
[00:23:20] Paul Roetzer: The other issue I see here though is, The fair use of content, um, where you are potentially limiting someone's ability to make money. Mm-hmm. , if, like right now, the way Google gets around this is they provide links. You could question whether Google snippets are actually infringing on people's ability to, to, to generate value from their creation.
[00:23:41] Paul Roetzer: Um, because they just present the answer. But in, in the case of a language model generating a response, so like the example I've heard a number of podcasts is like a restaurant review or. Um, you know, what's the best Irish pub in Cleveland, um, to get a burger and a Guinness kind of thing. And if Chad g p t just replied and said, go to PJ McIntyre's in West Park Station, um, and get this burger with no citation whatsoever, but it actually got it from Yelp.
[00:24:08] Paul Roetzer: Like that's a, that's a difference. Did it get it from Yelp or did it not because it got it from Yelp and didn't give me a link to. Then I, it just infringed on its ability to monetize its data and its, um, commercial assets. And I just went to PJ McIntyre's without having to click on any links. So what the argument is now is like, well, you need the links, the citations in there.
[00:24:31] Paul Roetzer: Well, that's assuming that the output of the language model has a citation to be had. It didn't, it didn't get the information from Yelp. It actually curated probably, 10 or 20 sources about best pubs in Cleveland. There was a Cleveland do co cleveland.com article and wherever else people get their food information from and out of that, it's synthesized that you should go to PJ McIntire.
[00:24:53] Paul Roetzer: So the, uh, the, the case I keep hearing is, well, we'll just infuse citations into it. And I've seen like startups popping up that are doing citations. Um, I've seen, you know, Sparrow, we talked about last week, deep Mind is working on citations and I was like, citations of what? That's not how the model works.
[00:25:09] Paul Roetzer: So, It's really confusing to me where this goes. Yeah. From a to to make it tangible to people. As a brand, I would be very cautious about how you're using these tools because if you're using a third party tool that is leveraging these APIs like access to say, open AI to use this, and all of a sudden open AI gets in trouble for where their language model came from or how it was trained, you may be building a business on top of something.
[00:25:40] Paul Roetzer: You might not have access to in the near future for, for whatever legal reason may exist. If I'm an investor and I'm investing in these, these different platforms that are building on the APIs of these companies that might be in legal jeopardy in the near future, I just, I have no idea what the legal precedent is going to be here or when it's going to occur.
[00:25:58] Paul Roetzer: Mm-hmm. , but it sure seems inevitable that lawyers are going to be clamoring for class action lawsuits on this stuff. Like basic knowledge of how the legal industry works. We're looking at billions or trillions of dollars in market value to be created, they're gonna want a piece of that because there are questions to be asked, rightfully so, and there are legal cases to be brought.
[00:26:19] Paul Roetzer: And I would just exercise caution as people that use these tools as investors in the companies that build these tools. And as the companies themselves, like I think we talked about it maybe on the last. I of like, is Sam Altman gonna have a Zuckerberg moment? You can just see Sam Altman sitting in front of Congress getting grilled on how language models work and you know, which side of the aisle chooses to take on language models first.
[00:26:45] Paul Roetzer: Is basically, I think, where we're at right now. So I, I just, I gotta guess 2023, we're gonna see some crazy legal cases at least moving forward, where these questions are gonna get asked a lot more. Um, and I just would encourage people to pay attention. Because it's gonna affect you as an end user, as an investor, as a company that's building these things.
[00:27:07] Paul Roetzer: Um, this is, it's not a clean story of how this stuff got built and where it's gonna go from here.
[00:27:14] Mike Kaput: And so just to clarify here for maybe some people that are newer to this topic, when we're saying be cautious or be careful, and again, none of this is legal advice. We're more advising that you may want to.
[00:27:27] Mike Kaput: Aware of how much of your strategy or your efficiencies or your business planning are relying on certain models or tools that could run into trouble down the line. As far as we can tell, though you are not. Breaking the law by using a dolly tool or using a generative AI text generator. So businesses don't necessarily need to worry about, oh my God, should I be using this tool at all?
[00:27:55] Mike Kaput: It's more about, you are saying, taking a longer view of paying attention to these cases, the precedents and how you're using the technology. Is that
[00:28:04] Paul Roetzer: correct? Yeah, I, I mean, I'll give you a practical example. If you use AI to create an icon for your logo, my understanding today is you cannot copyright an AI generated logo.
[00:28:15] Paul Roetzer: So if you wanted to get a trademark or you not trade copyright, you can't get a trademark on it. Trademark on, okay. But in a copyright, so like we actually submitted some materials for copyright to the US Patent and Trademark office for copyright, and there was AI generated image in the decks, and we can't, those can't be, uh, part of a copyright.
[00:28:33] Paul Roetzer: Hmm. So again, I'm not giving legal advice, but I'm saying is if you are generating. Content through ai, can you hold the copyright to the content? Um, and if you're generating images or video with ai, can you protect those in the same way you would traditionally protect things you would create for your organization.
[00:28:52] Paul Roetzer: So the intellectual property, talk to your attorneys. Get a, get an IP attorney. Get somebody who knows is studying this space cuz it's so new. You're gonna go find experts in generative ai, but find IP attorneys who can advise you on this stuff, who can keep track of the current case. and give you the best guidance they can is I guess my recommendation here.
[00:29:12] Paul Roetzer: That's what we've been doing the last couple months with a lot of the AI stuff we've been working on is what can we even protect. Yeah, there's
[00:29:21] Mike Kaput: probably a golden opportunity outside of doing the actual, uh, legal work for some enterprising attorney out there to release some type of guidance or product on, on this topic.
[00:29:33] Mike Kaput: Absolutely . Yeah. And all right, so I think this third topic really kind of brings home some of the things we've been talking about because it shows how fast. This technology right or wrong, as these legal cases are being figured out, as we're figuring out what the impact of the technology is, how fast it is being baked into commercial applications.
[00:29:59] Mike Kaput: So, you know, we have talked at length about Microsoft on the last few podcasts because they are just on fire. A number of AI advancements and a number of really savvy moves in the AI space. So we talked in the past couple weeks about reports that Microsoft might invest as much as 10 billion into open ai, and now they've actually announced formally that companies can now access open AI's most advanced AI models and tools Through this.
[00:30:35] Mike Kaput: Microsoft Azure Open AI service, so as part of Azure's open AI service. You can actually access and build on top of solutions like chat, G P T 3.5, which is the model that's powering chat, G P T
[00:30:51] Paul Roetzer: G, PT 3.5, not chat, G P T 3.5. Just to clarify, uh, sorry,
[00:30:55] Mike Kaput: that was my mistake. G P T 3.5 is the model that is powering chat, G P T, and then you can also access things like Codex, which is another.
[00:31:05] Mike Kaput: An AI model that generates code, uh, as well as Dolly two, which generates images. So Microsoft says, as part of this Azure Open AI service, you'll be able to access those types of models, build on top of them, integrate them into um, your business. And also they have hinted that soon they did not give a date.
[00:31:25] Mike Kaput: You were actually able be able to access chat g p t as well. So that is coming down the line and. This obviously opens up huge commercial opportunities for Microsoft, but also opens. Incredible opportunities for customers who are using Azure OpenAI service because you can now start using these types of powerful tools in your own business.
[00:31:50] Mike Kaput: Now, one interesting aside here that's related to some of the IR regulatory and legal concerns we've discussed is that Microsoft has also, uh, Publicly with this announcement, uh, taken the time to talk about their guardrails that regulate how this all works. So currently, and this could change, developers are required to apply for access to Azure opening eye service, and as part of that, they have to describe their intended use cases and applications before they can get access and do anything with the tools.
[00:32:25] Mike Kaput: So Microsoft has been pretty public about. Being responsible with rolling out this technology commercially. So initially, Paul, I wanted to get your thoughts. What do you think of this latest move by Microsoft, especially given what we've talked about on the previous couple
[00:32:40] Paul Roetzer: of podcasts? So the, the guidance I'll give to people on this is, um, when we started studying ai, you know, I started back in 2011.
[00:32:51] Paul Roetzer: I had no idea what I was looking at. Like, I, when, when news would come out, I had no idea if it was significant. And so over time, over the last 11 years, what I've done is curated a, a very, um, specific group of people, uh, journalists, analysts, entrepreneurs, investors, and when something happens that I think is significant, and I've gotten better, obviously over time at gauging for myself, whether something is significant.
[00:33:18] Paul Roetzer: I saw the article last night and I'll go look and say, okay, what are they saying? Like, this seems like a really big deal to me, which is what happened yesterday. I saw this immediately and I was like, well, that's a big deal. So I go to Twitter, start looking at my list and sure enough, you're starting to see right away the, the tweets.
[00:33:35] Paul Roetzer: So you had, um, Eric Hortz, who is the Chief Scientific officer of Microsoft, tweeted API X is now available. GBT 3.5, codex D two. API access to chat G B T on deck, Satya, the c e O of Microsoft tweeted Greg Brockman, the co-founder and c t O of Open AI tweeted. So everyone that was, anyone at the leadership teams of these companies was tweeting the story, which is when you know it's a big deal.
[00:34:02] Paul Roetzer: Now. I think it's ironic, like they just released this sort of blog post, like I was kind of laughing as a marketer last night to like this big announcement and like they just put out a blog post and do some tweets. It's like Elon Musk style marketing. Just tweet the. Um, so it's a big deal because the people who know think it's a big deal.
[00:34:21] Paul Roetzer: And that's often like a, just a, a litmus test for me is like, are the people who are on the inside thinking this is a bigger deal as I think it is. Um, so the answer to this one was yes. The second piece is there's been so much rumor for the last few weeks about OpenAI, Microsoft, like you said, we talked about on the show at length.
[00:34:39] Paul Roetzer: And it was the first moment where I looked at someone was like, oh, this is Microsoft officially saying this, like they're actually acknowledging something. And there's been very little that I've seen come out directly from Microsoft in recent weeks about any of this stuff. It's just been all hearsay.
[00:34:53] Paul Roetzer: And sourced, you know, media articles. So I think it's, it's big because it's Microsoft stepping on and saying, here, here is what we're doing. It's starting to show how they're going to infuse this, and I assume, we'll, probably in the near future hear the final details about whatever the investment ends up being that they're supposed to be making.
[00:35:11] Paul Roetzer: But I also immediately went and looked and said, when is Google's next earnings call? Because, I think what we have to be constantly looking at this year, and we talked about this on the previous episode, is the domino effect of the big tech players and the plays they're going to make into ai. Mm-hmm. . So Microsoft, my, my one former PR person read into this is they try to get out ahead of something else, like, I feel like something else is about to, to.
[00:35:41] Paul Roetzer: Because the news of this just sort of got out real fast. Mm-hmm. , um, through these channels. And sometimes you'll do that if you're trying to get out ahead of something else, you know, is coming. So I feel like they got this out and the answer to when is Google's earnings call alphabet. And technically the parent company is February 2nd.
[00:36:00] Paul Roetzer: There is no way in my mind that they. Tell the story of what they're doing with AI before that, they're gonna get grilled by analysts on that call. Hmm. So if you go into the earnings call on February 2nd is Alphabet, and you don't have answers to what you're gonna do to combat Microsoft's play with open ai, I, I would think your stock's gonna get crushed.
[00:36:22] Paul Roetzer: So we've been saying this for years of like, look at the analyst reports. What are they saying about AI as like a way to project out where these companies are going and which ones are actually the best investments based on what they're doing with ai. I, I, I will be fascinated to hear that call or to read the transcripts from that call.
[00:36:39] Paul Roetzer: So I, I just think that it's a big deal because it integrates open ai. It's a big deal because Microsoft is officially saying something and it might be a bigger deal because of what comes next. Cuz it's gonna trigger Google and, well, it may even to be the final triggering moment, but everything that's going on is going to cause Google and I assume meta.
[00:37:01] Paul Roetzer: And probably some other players to come out with some something of significance around their play with ai. And I've heard, you know, like you and I are both big fans of the All In podcast. Yeah. Um, I, I was hearing like them, the one theory was that they open source Lambda, that they just like, they're like, fine, just let it all out.
[00:37:20] Paul Roetzer: Like we're gonna play this game. Like, we'll just let it all go, let Palm go and lay him to go like, And I, I mean, is that a realistic play? I don't know. But lamb, lamb is the large language model. The main, the main model that you hear about at Google, uh, if you're not familiar. So part of it they're saying is like, Google may have better tech and just like, let's let it all out and then just like take control back on the market.
[00:37:42] Paul Roetzer: And then I still think that if Meta's smart, like Facebook, they, they recapitalize the AI function. They have a massive AI research lab led by La Jan Lacoon doing insanely cool. And I don't, I don't know that the average business person even knows it's happening. Like I don't, I mean, we know, and I, but I think we take for granted that we follow the space so closely that we know generally like kind of what they're working on there.
[00:38:05] Paul Roetzer: Um, but I don't know that the average business person even knows that Fair exists, that, but and who, who Jan Lacoon is like, that's crazy to me that there are, there are business people today who don't know who Jan Lacoon and Demi Saabas and Jeff Hinton and all these players are, but that's actually probably the majority that have no idea.
[00:38:23] Paul Roetzer: That's why we always tell people, read Genius Makers, then you'll understand . Yep.
[00:38:28] Mike Kaput: Genius makers by Cade Metz, who is New York Times AI reporter and also spoke at our, uh, Macon conference one year. Um, definitely one of the seminal books in, uh, figuring this stuff out. I think we both would agree. Yeah. Yeah.
[00:38:43] Mike Kaput: So that relates to kind of the final question I have for you here, and as we wrap up today's episode. Um, you mentioned you think that maybe CEOs and executives probably don't know who a lot of these major players are, and that, that seems very obvious and we've talked about chat. G p T has obviously changed things pretty significantly in building broader awareness about what we've been talking about for almost five, five years now.
[00:39:11] Mike Kaput: With the Microsoft plus OpenAI slash Chat, two BT moves going on. If I'm a C E O or an executive who hasn't been paying attention, there's probably a good chance I'm starting to, and there's probably a good chance at a bigger enterprise, especially that I know for a fact our organization is using Microsoft Azure or considering it or something similar.
[00:39:36] Mike Kaput: If I'm looking at this as a. What, what should I even be thinking about if I'm getting kind of blindsided and being like, wait a second. The thing being used in my kids' school to like cheat on tests is now being available , potentially in the technology we're using every day. How do I take advantage of this?
[00:39:55] Mike Kaput: How do I catch up, get up to speed, start thinking about it, what
[00:39:59] Paul Roetzer: do I do? Yeah, I mean, that's a challenging question. I, I think. You know, if I was in those shoes, I would probably be looking around my organization saying, who needs to be on the committee to figure this out? Like, you ha if you have a chief, a officer, great.
[00:40:14] Paul Roetzer: Like obviously that's your first call and, and maybe the only call you need to make. But a lot of middle market and enterprise companies, certainly small businesses don't have ahead of ai. Um, so I think you're trying to figure out who in the organization can help figure all this out very quickly. . And then how can you build a roadmap to become AI emergent as an organization to infuse AI into marketing, sales, service, ops, product, hr, finance, all these areas over time.
[00:40:41] Paul Roetzer: Obviously if you have Microsoft Azure, your first call is probably to whoever owns that relationship and saying, mm-hmm. , what should we be doing here? You and I did a consulting gig for a, a healthcare company, uh, a few years. Where this is what actually what happened, like we were in there, we're analyzing the site and the recommendation engine that didn't exist and, but we basically were like, Hey, you can build a recommendation engine based on this and this can drive revenue here and kind of laid out this roadmap.
[00:41:05] Paul Roetzer: But you need these capabilities and, uh, The tech person was like, oh, we have Microsoft Azure and they're willing to give us credits to, to use the machine learning capabilities. Mm-hmm. , it's like, well, there you, there you go. Like, and so all it took was like building the business case and having the right person in the room who knew that Microsoft was incentivizing the use of their AI tools.
[00:41:24] Paul Roetzer: Uh, for this type of organization and o often running, they went. So yeah, I mean, I would look at, whether it's Google Cloud or a w s, they, they all have AI capabilities, so I would start with that. Who is your core cloud provider? Again, for context, a w s dominates this market. I don't know what the current market share is, but I would say somewhere between 38 and 43% or so of the market is owned by aws.
[00:41:48] Paul Roetzer: Then Microsoft is in, I think, the low to mid twenties, and then Google Cloud is somewhere around I single digits, low teens in terms of market share for cloud. So this is a major play by Microsoft, also from a cloud perspective. Mm-hmm. , you know, can, can they? And they catch up to aw w s which just seems like the runaway winner in this space, um, with a play like this.
[00:42:07] Paul Roetzer: But yeah, I mean, as a C E O I would, I would figure out who do I have on my team that can, can figure this stuff out? And if you don't have anybody internally, then, then get somebody, like either get an outside consulting firm or, um, you know, go bring somebody in. But, uh, good luck finding that person. I mean, it's.
[00:42:23] Paul Roetzer: The business savvy AI people aren't growing on trees right now. Like that. That's a, that's like one of the, kind of the trends or bets we're making in 2023 is AI savvy business people and practitioners are not in high demand right now. Like the people who actually understand AI and can comply to content marketing are advertising or social.
[00:42:41] Paul Roetzer: I think by the end of 2023, you're not hiring someone unless they're AI savvy, like you're, you either have your internal training program to make them ai. Or you are actively searching for practitioners and leaders who not only are good at their discipline domain, but understand artificial intelligence and can help move your organization into the future very quickly.
[00:43:03] Mike Kaput: That's the, uh, that's the money quote right there that we end the podcast on . Paul, thank you as always for all the time and insight and sharing your insights with the audience. I think people are gonna get a lot of value out of this episode.
[00:43:17] Paul Roetzer: Yeah, it's good stuff. I mean, it's fun to talk about. I'm glad we've gone to this format where we, otherwise, again, like you and I just like zoom, message each other all day and we never, like, sometimes we'll think to put something on LinkedIn and otherwise we just don't
[00:43:29] Paul Roetzer: But I, I will say, For the listeners who are engaging, um, outside of the episode, like keep doing it. Like we're, yeah. I'm starting to get inspired by the things I'm seeing people putting on LinkedIn in our Slack community. Um, people are taking knowledge and just running with it and Yep. You know, really expanding it into their own research areas and disciplines and so it's really encouraging to us to.
[00:43:52] Paul Roetzer: All the work that's being done and, and the podcast itself. What we looked at the numbers recently, Mike, like since we went to this format, um, the average episode is up. Was it like 180% or something like that? That downloads per episode? Yep. So just we know that there's more listeners out there. It again, it just sit here and record this thing and put it out into the web and never know if anyone's on the other end, but the data would tell us that there's a lot, lot more people on the other end listening.
[00:44:18] Paul Roetzer: So thank you for listen. And yeah, we hope you just stay curious and keep pushing, uh, forward with what's possible with AI in your area and, and keep engaged with us. We'd love to hear from you and learn more about what you're doing. Uh, last, uh, thing I'll say is we just open a call for speakers for our marketing AI conference, Macon.
[00:44:36] Paul Roetzer: It's gonna be July 26th, the 28th in Cleveland. So if you're doing cutting edge stuff and generative ai and. On the legal side, you have case studies, um, C M O Insights. We're really interested in, uh, brand stories in particular. We love to hear from people who are doing cool things on the brand side. So if you've got a talk in mind, uh, just go to macon.ai.
[00:44:58] Paul Roetzer: It's just m a i c o n. Ai, right? Is that what it is? Yep. Okay. Yep. Um, and that'll take you to the event site. And then on the speaker page there's a form submission. So again, we, we'd love to, um, hear your ideas and, and maybe see you in Cleveland in July. So yeah, thanks again, Mike, and we'll talk with everybody next week.
[00:45:17] 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 marketing ai institute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.
[00:45:39] Paul Roetzer: Until next time, stay curious and explore ai.
Cathy McPhillips is the Chief Growth Officer at Marketing AI Institute.