This week’s episode features Paul Roetzer giving a year-in-review of thew world of AI. It’s been a busy six months, and Paul references some of the top stories and developments of 2023. See the timestamps and summary for more details.
Listen or watch below, or find us on your favorite podcast player.
This episode is brought to you by BrandOps, built to optimize your marketing strategy, delivering the most complete view of marketing performance, allowing you to compare results to competitors and benchmarks.
Watch the Video
Timestamps + Links
*NOTE: All of the stories listed below were covered in previous podcast episodes. See our YouTube link where Paul and Mike originally covered that story.
00:05:08 — Jan. 16: ChatGPT Disrupts Education; Alarmed by A.I. Chatbots, Universities Start Revamping How They Teach - The New York Times (Paul and Mike discuss on YouTube)
00:08:05 — Jan. 28: Marketing AI Institute’s Responsible AI Manifesto; The Responsible AI Manifesto for Marketing and Business (Paul and Mike discuss on YouTube)
00:11:51 — Feb. 1: UBS Announces That ChatGPT Hit 100M Users in January; ChatGPT sets record for fastest-growing user base - analyst note | Reuters
00:12:33 — Feb. 6: Google Bard Is Announced; Google AI updates: Bard and new AI features in Search (Paul and Mike discuss on YouTube)
00:14:13 — Feb. 7: Microsoft Announces AI-Powered Bing and Edge; Reinventing search with a new AI-powered Microsoft Bing and Edge, your copilot for the web (Paul and Mike discuss on YouTube)
00:14:53 — Feb. 24: Meta Publicly Releases LLaMA Under a Noncommercial License Focused on Research Use Cases; Introducing LLaMA: A foundational, 65-billion-parameter large language model
00:20:24 — Mar. 6: HubSpot Releases ChatSpot in Public Alpha (and Content Assistant in Private Beta); Working Smarter, Not Harder: HubSpot CRM Introduces New AI-Powered Tools to Boost Productivity and Save Time (Paul and Mike discuss on YouTube)
00:22:19 — Mar. 7: Salesforce Announces Einstein GPT in Closed Pilot; Salesforce Announces Einstein GPT, the World’s First Generative AI for CRM (Paul and Mike discuss on YouTube)
00:25:02 — Mar. 14: Google Announces PaLM and Expanded AI Features in Google Workspaces; The next generation of AI for developers and Google Workspace (Paul and Mike discuss on YouTube)
00:27:04 — Mar. 14: Anthropic Announces Claude; Anthropic | Introducing Claude
00:29:03 — Mar. 16: U.S. Copyright Office Releases Generative AI Guidance; Federal Register :: Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence (Paul and Mike discuss on YouTube)
00:30:48 — Mar. 16: Microsoft Introduces Microsoft 365 Copilot; Introducing Microsoft 365 Copilot – your copilot for work
00:32:55 — Mar. 20: Runway Announces Gen-2; Text-to-video AI inches closer as startup Runway announces new model - The Verge
00:34:34 — Mar. 21: Adobe Announces Firefly Generative AI Capabilities; Adobe Unveils Firefly, a Family of New Creative Generative AI
00:36:25 — Mar. 22: AI Experts Release Open Letter to Pause AI Development; Pause Giant AI Experiments: An Open Letter - Future of Life Institute (Paul and Mike discuss on YouTube)
00:40:05 — Mar. 30: Release of AutoGPT; Developers Are Connecting Multiple AI Agents to Make More ‘Autonomous’ AI (Paul and Mike discuss on YouTube)
00:42:24 — Mar 31: Italy Bans ChatGPT; Italian regulators order ChatGPT ban over alleged violation of data privacy laws - The Verge
00:43:28 — Apr. 13: Amazon Announces Generative AI in AWS; Announcing New Tools for Building with Generative AI on AWS | AWS Machine Learning Blog (Paul and Mike discuss on YouTube)
00:44:55 — Apr. 19: Marketing AI Institute Statement on Knowledge Work; How AI Could Impact Millions of Knowledge Workers in the Next 1-2 Years (Paul and Mike discuss on YouTube)
00:47:02 — May 1: Geoff Hinton Announces He Has Quit Job at Google Over Safety Concerns; ‘The Godfather of AI’ Quits Google and Warns of Danger Ahead - The New York Times (Paul and Mike discuss on YouTube)
00:49:35 — May 10: Google Announces Generative Search Experience; Supercharging Search with Generative AI
00:50:35 — May 16: Sam Altman Testifies Before Congress; Mr. ChatGPT goes to Washington: OpenAI CEO Sam Altman testifies before Congress on AI risks | CNN Business (Paul and Mike discuss on YouTube)
00:52:40 — May 30: AI Poses “Risk of Extinction” According to AI Leaders; Statement on AI Risk | CAIS
00:54:06 — June 5: Apple announces Vision Pro; Introducing Apple Vision Pro: Apple’s first spatial computer
00:55:51 — June 7: DeepMind Announces AlphaDev Breakthrough in Sorting Algorithms; AlphaDev discovers faster sorting algorithms
00:59:02 — June 8: Cohere announces $270M Series C; Cohere Announces $270M Series C to Bring Generative AI to Enterprises
Read the Transcription
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: think there is a high probability that these jobs could not only be impacted in changing the way they work, but that they even are needed.
[00:00:08] Paul Roetzer: So hope that that is not the case. we still have time to take action. really what it comes down to is each individual organization committing to that human-centered approach to the adoption of this technology.
[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.
[00:00:39] 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 51 of the Marketing AI Show. I'm your host, Paul Roetzer. This is a special edition, a little different than our usual weekly format. Mike and I are both traveling, in and out of the country for the next couple weeks. So we thought we would do a couple of special shows. This week's is going to be interesting because what we're doing is taking a look back at the first, almost six months of the year.
[00:01:12] Paul Roetzer: So we're sort of at the midway point. A lot has happened in artificial intelligence this year and we thought it would be really valuable for ourselves to stop and take a look back and hopefully to our listeners. Many of you have been joining us as the year has gone on and maybe haven't had a chance to go back and hear some of the earlier episodes.
[00:01:31] Paul Roetzer: And then I know there's a lot of you who've gone back and listened to a lot of the earlier episodes. because I love connecting on LinkedIn and learning about how much you're enjoying the show and getting value out of it. I thought it would be really helpful to just take a moment and look back at what has occurred and try and consider what that could mean for the next six months.
[00:01:51] Paul Roetzer: So the way we're going to do today's show is, we went through, the episodes starting in January and we started looking at what were the major things that happened, and we curated that list town to 30. My original goal was 10. There was just no way. There was a, a single week in March where you could pick 10 from.
[00:02:09] Paul Roetzer: So I'm going to move really quickly through these 30 items. But just to give you some context, when we switched the marketing AI show to this weekly format where Mike and I picked three main topics and then rapid fire items, we started that in October of 2022. So October 14th to the exact episode 20, was when we switched to this weekly format in October of 2022, our podcast had 433 downloads the entire month.
[00:02:39] Paul Roetzer: In May of 2023, it was over 24,000. Each episode is now averaging about 1700 downloads in the first 24 hours. So the interest in artificial intelligence, the amount of news and information that is occurring on a weekly basis is kind of hard to comprehend. But just seeing the growth of the podcast and then hearing from so many listeners each week, you know, really tells us that there's something going on here.
[00:03:08] Paul Roetzer: And the awareness and interest in this topic continues to explode. And as we talk about kind of some of these major things that have happened throughout the year, I think we'll all start to realize the context of what's going on. Even for me personally, you know, we put this list together and I've honestly just scanned the list.
[00:03:24] Paul Roetzer: I haven't really thought deeply about any of these or what I'm going to say. More just to kind of approve that. Yeah, these are all 30 and I don't know what else to cut to try and just like simplify this. So I'm going to just kind of go through 'em pretty quickly. Before I get started quick, thank you to our, podcast episode sponsor BrandOps.
[00:03:43] Paul Roetzer: Did you know brand health, media monitoring, social listening, listening, competitive intelligence share, a voice and review tracking could all be done with the same tool? When we sat down with the BrandOps team, it was remarkable to see what one AI-based platform could do. Having a complete view of brand marketing performance and instantly knowing where to focus to increase impact helps businesses unlock faster growth.
[00:04:08] Paul Roetzer: Visit BrandOps.io/marketingaishow to learn more and see BrandOps in action.
[00:04:16] Paul Roetzer: All right, so let's get into the top news items, from the year so far, 2023. And again, I'm recording this around mid-June. It's going to come out. We got, I don't know what day that's going to be, 12, 13, 14, I think June 14th. This episode's going to be coming out, so you might be listening to it that week.
[00:04:35] Paul Roetzer: And we thought about trying to reforce rank these. We were going to go like, what are the top 10? And like, count down to one. And then we realized there was just no way to do that. And so what we decided to do is go chronological and we decided that actually builds a better story because it helps you realize how quickly things are changing and sort of the sequence with which these different innovations and news items occurred.
[00:04:55] Paul Roetzer: All right, so here we go. We're going to start January, 2023. We actually took the first two weeks off of the year. We did not do an episode, so January, I think January 16th, around there was our first episode. So we had on January 16th ChatGPT Disrupts Education. There was a New York Times article that came out, said ALARM by AI Chatbots, universities start revamping how they teach.
[00:05:20] Paul Roetzer: So again, keep in mind ChatGPT comes out November 30th, 2022, right before winter break for a lot of these universities. And honestly a winter break for a lot of businesses. And so in December there was a lot of scrambling to figure out what exactly is this technology, what impact is it going to have?
[00:05:41] Paul Roetzer: And so already within 45 days we had people talking about the disruption to the entire education system. Like it was that quickly understood that this was very different technology. January 17th, just a day later, we had news that Getty images announced that it has commenced legal proceedings against Stability ai.
[00:06:03] Paul Roetzer: So Stability AI is one of the companies building these foundational models, specifically Image, generation is the main thing they're involved in. They now have a language generation model as well. They raised a hundred million in October of 2022. They were recently in the news because, there was, I think it was a Forbes or Fortune article that was sort of questioning some of the legitimate legitimacy of the founding of the company.
[00:06:28] Paul Roetzer: So, you know, it's like with a lot of the stuff with ai, there's layers and layers to these stories. It's not as simple as company raises a hundred million. It's an amazing company and it's going to be, you know, runaway success. But the issue here that we have talked about throughout the podcast is that these foundational models, so again, a large language model, you know, the foundational models that are built that enable ChatGPT and these other technologies.
[00:06:52] Paul Roetzer: In the case of image generations, same deal. You have these foundational models that are trained on all this data so that they can generate images or generate videos. And it seems pretty clear that a lot of these models were probably trained on copyrighted materials that they didn't have permission to use.
[00:07:09] Paul Roetzer: Now they're going to argue for the years to come that it was fair use. There will be lots of legal proceedings, lots of money made by lawyers, lawyering this stuff. Probably Supreme Court cases. So it's just the very early stage of this. But in this case, what was happening is images that were being output by Stability, ai, their, their models had elements of the Getty watermark in them.
[00:07:33] Paul Roetzer: So it was very apparent that they did in fact, use copyrighted materials to train their model because it was coming out in the outputs of it. So I am no lawyer, but my assumption is some, some pretty hefty penalties are going to get paid at some point. And this is just one of many, lawsuits that are already in the works in this space.
[00:07:56] Paul Roetzer: So definitely throughout the rest of the year, we're going to see continued legal cases being brought against some of these companies. And this is one to watch. So, January 28th. So January wasn't huge. I've got three items here in January. This is number three on our list. I wrote something called the marketing, the Responsible AI Manifesto for Marketing and Business.
[00:08:17] Paul Roetzer: And we dedicated, I think, a whole episode to this one in January. the premise here was that, you know, when we wrote our book, marketing Artificial Intelligence, it came out in summer 2022, we looked around and tried to find brands that had great responsible AI guidelines and principles to, to help lead their company.
[00:08:38] Paul Roetzer: What we found was most of the time it was the tech companies that had these responsible AI guidelines. And so Google and Adobe, I think are the two that we used in the book as examples. So theirs were more about the building of this technology, and what we were struggling to find was where's the guidance for marketers and business leaders about responsibly applying this technology in their organization for their brand, for their people.
[00:08:59] Paul Roetzer: And so this was like one of those spur of the moment things. I'd been thinking about this for a couple years. I got home, I think it was like a Friday, got home from the gym and I just wrote this thing. And so I sent it to the team that day and I said, here, we're going to put this up like this weekend. If anybody has any feedback, throw it in.
[00:09:16] Paul Roetzer: Otherwise we're going live with this thing. And so you can go get this, we'll put it in the show notes. What I wanted to do was put out a creative commons responsible AI principles guide and let people start with that as a base, because I've talked to enough leaders, enough organizations, enough marketers to know people don't have these things.
[00:09:34] Paul Roetzer: Generally speaking. And so I thought if we just put out a template, it's not going to be perfect. We'll have to evolve it ourselves. It'll keep moving forward. But hopefully if we open up to the community, other people will build on top of it, edit it, revise it under the same creative commons attribution, share like license, and then hopefully we can move the industry forward.
[00:09:52] Paul Roetzer: So there's a lot of really, IM important elements to it. There's 12 principles that are outlined. But the basic theme of it is it has to be human-centered. So we're not, we can't wait for government, to come along and apply regulations to protect us against this stuff. So each organization has to think about the impact it's going to have on their people, on the consumers that they reach, on their communities.
[00:10:15] Paul Roetzer: And they have to put these principles in place that say, this is what we believe about AI and how we're going to apply it. And again, the whole theme is, our feeling is it should be human centered. We're not trying to remu remove humans from the loop. We're not trying to put them out of jobs. Some of these things may be an outcome of it, but that shouldn't be the goal of it.
[00:10:35] Paul Roetzer: And so that's really what it is. So it's a, it's a really good starting point. If you don't have this or you're kind of new to thinking about some of the ramifications of aI would suggest, you know, go and check that out. And again, it's, it's free and accessible on our site. Doesn't even require, an email address.
[00:10:48] Paul Roetzer: It's not a download, it's just on the, on the site. So check that out. Again, responsible AI Manifesto for marketing and Business January 28th. So that rubbed up January. Again, not a, not a ton happening. Sort of a slow build, as everybody was trying to figure out what ChatGPT was and the impact it was going to have.
[00:11:05] Paul Roetzer: I was having lots of conversations at that time with SaaS companies who were scrambling to figure out what this meant to their product roadmap and what generative AI was and how it was going to change, you know, what they were doing in business. I thought a lot of venture capital firms started getting, Worried that the SaaS companies they were investing in had no idea what they were doing.
[00:11:24] Paul Roetzer: And that is a pretty accurate, assessment largely of where the SaaS industry was late in 2022, early 2023. So there was a lot happening behind the scenes. I was having lots of conversations with universities, school systems, boards, venture capitalists, CEOs, product leaders. So you could tell qualitatively that everything had changed, but we weren't seeing it necessarily yet.
[00:11:47] Paul Roetzer: Play out in the news. So then February rolls around right at the start. February 1st. Ub s announces that based on their data, ChatGPT hit a hundred million users in January. So fastest growing consumer technology in history, on consumer replication in history. And that came out in a, a Reuters article.
[00:12:07] Paul Roetzer: So we now knew again, it was users like it could be, people tried it once, but a hundred million is a, obviously a very significant number. And I was starting to feel that I was getting, you know, text messages from family members asking what this stuff is and how it works. Again, lots of calls from teachers and, you know, professors and just people I wouldn't normally hear from that were testing the technology and trying to understand the impact it was going to have.
[00:12:33] Paul Roetzer: February 6th, we had Google Bard announced. So Google said an important step in their journey and talked about the impact it had on search. But that Google Bard was coming, they were going to take one of their language models. Again, Google has thousands of AI researchers at the time, they had two AI research labs.
[00:12:50] Paul Roetzer: They had Deep Mind and Google Brain. Those two labs later in the year would merge together to form Google DeepMind. But Google is not new to ai. They've been working, aggressively on artificial intelligence for decades since the founding of the company. And so Google Bard was announced as I, you know, I guess the simplest way to think about it is their version of ChatGPT.
[00:13:12] Paul Roetzer: So if you haven't tried Google Bard, it is basically the ChatGPT, it is now available. You can go to bard.google.com and you can test it for yourself. A lot of, I've, I've tested it in many use cases. Generally speaking, GPT-4 in ChatGPT seems more powerful. Better reasoning capabilities, more creative outputs.
[00:13:34] Paul Roetzer: But we also know that the current version of Google Bard is not even close to the most advanced version that Google has access to or could put out into the world. So expect a lot in the second half of this year with more advanced versions of not only ChatGPT, but Google, Bard and some of the other major players.
[00:13:53] Paul Roetzer: So that was where we started to see Google taking action. Though, you know, there was a lot of talk early in the year about Microsoft sort of caught and open eye, caught Google off guard by releasing ChatGPT into the world and Google sort of scrambled to figure out what does this mean and what do we do about it?
[00:14:09] Paul Roetzer: And this was where we started to see the the next steps they were going to take. So number five on our list, Google Bard is announced. Number six, Microsoft announces AI powered Bing an edge. So in essence that their, you know, GPT-4 was being brought to, the search. And that has been a recurring theme is Microsoft's infusion of this, open AI technology into the product.
[00:14:31] Paul Roetzer: And again, Microsoft wasn't new to ai. Thousands of AI researchers been working on AI for decades at Microsoft. But they found it was going to be faster and more impactful to partner with open AI to bring their technology to market. In exchange open AI gets all the compute power they need and all of Microsoft's expertise and distribution.
[00:14:49] Paul Roetzer: You know, meaning they have obviously millions of customers. Number seven, this one was an interesting, one sort of flies under the radar to start, but meta publicly releases llama under a non-commercial license focused on research use cases. So they gave a 65 billion parameter, large language model, just means relatively big, large language model.
[00:15:12] Paul Roetzer: To researchers, so it, it flew into the radar because while they were opening this model to people as an alternative to working with like a GPT-4 or something, they didn't release all the details to the public of how it was built, what the weights were, what the training data was within 10 days.
[00:15:29] Paul Roetzer: All of that information leaked out though. And so what happened was it accelerated access to advanced open source large language models. And so this, the, in retrospect, this will probably be a very important moment, not only just in the half, first half of this year, but maybe in the history of the diffusion of this technology.
[00:15:49] Paul Roetzer: It was really the first time we had a model like this that people could start building on. And so you've started seeing innovative applications, built on top of this within weeks, once the, all the information was released about, how it was built and what data was trained on, what the weights were.
[00:16:06] Paul Roetzer: So a, a really important thing, and again, if you're new to this, it may be weird to think about like Facebook or Meta having a play here. They have one of the top AI research labs in the world run by Yann LeCun, and they formed it back in and it was like 2012 or 2013. So to me, the, like we talk so much about Google and Microsoft and Open AI and some of these other players.
[00:16:28] Paul Roetzer: The dark horses in all of this, to me, to the general public and and business world are, are probably going to be meta and apple. I don't think they get talked about nearly enough with what they're doing, in part because Apple doesn't talk about what they're doing. Meta does, but they just don't get much credit for it.
[00:16:49] Paul Roetzer: So, you know, I've historically not been the biggest fan of Facebook as a company. I generally still have some issues with the company, but their AI research is top-notch and they continue to pump out research papers and models and I don't think that's going to stop. I really think that, within the, you know, certainly the second half of this year, we're going to start to see meta play a much larger role in all of this as it starts to get infused into their, technology and obviously have massive distribution with over a billion users or so, Facebook.
[00:17:23] Paul Roetzer: So if they choose to put this stuff out into WhatsApp and Instagram and. Facebook, like they can have a, a major play here. So keep an eye on meta and keep an eye on Apple, which I'll talk about a little bit later. Okay, so that was February. So we had what, four items in February up number seven March is when stuff just got crazy.
[00:17:44] Paul Roetzer: So March 2nd I wrote, an article on our site, which again we'll link to in the show notes called The Law of Uneven AI Distribution. And what happened here was I was starting to think about how this technology was diffusing. So we were having conversations with companies across every industry, we're, you know, studying what's going on, reading the research papers, and you're realizing like this technology is going to move so fast and the capabilities are so significant.
[00:18:13] Paul Roetzer: And yet there was companies in industries that were just shutting off access to it. Do you have like financial services companies blocking ChatGPT Access? You had educational systems that were not allowing students to use ChatGPT and trying to pretend like it didn't even exist. I was seeing really interesting consumer applications come out that I would never use because of the amount of data you would've to give up to get them.
[00:18:35] Paul Roetzer: And so I was trying to process like, how is this going to play out in business and then in our personal lives. And so I wrote the law of uneven AI distribution, which basically this is what it sums up to, the value you gained from AI, both personally and professionally, and how quickly and consistently that value is realized is directly proportional to one, your understanding of the technology.
[00:18:56] Paul Roetzer: So you have to understand it before you can use it, your access to it. So do you work at a company where you're not even allowed to use these tools and your acceptance of it? Meaning, am I willing to give up the data I have to to get the benefit of this technology? And so what I realized is this going to be completely unevenly distributed across industries and then across individuals?
[00:19:16] Paul Roetzer: Because we all make these, these choices. So in our personal lives, AI has the potential for personalization, convenience, abundance, fulfillment, happiness. You can get all those benefits out of it in our professional lives. It can be efficiency, productivity, creativity, predictability of outcomes and behaviors, success.
[00:19:34] Paul Roetzer: But to get all of those things, you have to be willing to, one, understand the technology, learn it. You have to have the access to it, and then you have to accept what it's going to give you. And so this was, again, it was just more of like a, Thinking out loud moment where I was trying to comprehend this, but I've used this idea when I do a lot of public speaking to explain, depending on the industry I'm talking to, why they may not see the disruption and the impact as quickly as some other industries because they're just less likely to do it because even they don't have access or they're not accepting of what they have to give up, whether they're not allowed based on, you know, legal or, or privacy or technology reasons.
[00:20:14] Paul Roetzer: So just again, it's a good read. We have a, a podcast episode where we talked about this in early March, so you could check that out. So that was number eight, the law of uneven AI distribution. Number nine on March 6th, HubSpot. Announces chats Spott in public Alpha and they also announced content assistant in private beta.
[00:20:33] Paul Roetzer: So now, if you don't know my background, I used to own a marketing agency. I sold that agency in 2021, but we were HubSpot's first marketing agency partner back in 2007. So I built my agency around HubSpot and helped kind of form the partner ecosystem that today makes up a significant portion of their revenue and has played a major role in kind of building that company over the last, you know, whatever, 15 years or so.
[00:21:00] Paul Roetzer: So we stay pretty close to HubSpot. We use HubSpot at the institute to power our marketing. So. Dharma Shaw, the co-founder and c t o, personally was building Chats spott. And the idea here is it's built on, from my latest understanding, built on open AI's APIs specifically, I think they're using GPT 3.5 because it's faster than GPT-4, to infuse, language generation capabilities right into Hubspot's CRM, so you have it in your email, social blogging, wherever you're creating content within there.
[00:21:34] Paul Roetzer: And then I think he's also using DALLE-2's, API to build image generation capabilities in. They may have diversified, he may, he may be using other. Models. An API is not just open eyes, but that was the original direction. And so the key here is this is the first point where we start to think about, oh wait, I don't have to go look for generative AI technologies.
[00:21:56] Paul Roetzer: Now, as a marketer or business person, the tech stack I am currently using is just going to infuse them right in. So if you're a HubSpot customer, you may have gone and got image generation tools and language generation tools, and now you start to wonder, do I even, am I going to need those or am I just going to be able to use HubSpot and all that's going to be baked right in.
[00:22:15] Paul Roetzer: So that was number nine, March 6th. HubSpot releases, chats, spott or announces chats Spott. March 7th, the next day, number 10, Salesforce announces Einstein GPT, enclosed pilot, and same deal again. If you're a Salesforce customer, you're now going to have Einstein in Slack in developer in marketing and sales and service.
[00:22:36] Paul Roetzer: So anywhere within that, that platform that you use, your, your employees are using the technology. In theory, this generator AI capabilities will be right there. Now, as of June, 2023, I haven't heard anything about this opening up to all customers. I believe it is still in some form of a closed beta. But maybe you have access and you're testing these technologies.
[00:22:58] Paul Roetzer: But again, e early March we're starting to see a lot more happening already. So that was number 10, number 11, was obviously a big one. and actually let me go. I'm going to, now I'll stay in order. I don't want to confuse this. So on March 14th, GPT-4 is announced. So Open AI releases GPT-4.
[00:23:16] Paul Roetzer: If you've heard me give a talk recently, I tell the story of how March 14th I was traveling to, the West Coast from Cleveland. And in the time I took off from Cleveland to the time I transferred in Denver. The next, four or three items I'm going to talk about all happened. So we had GPT-4 s announced, which changed everything.
[00:23:40] Paul Roetzer: I assume by now most people have experimented with GPT-4. My experience so far is most people have not even scratched the surface of what GPT-4 can do. A lot of people assume it's just basically a writing replacement or just a writing aid. It is an insanely advanced strategy tool. And the plugins we're going to talk about a little bit, it expanded its capabilities, but for me, I actually don't use GPT-4 as a writing replacement at all.
[00:24:09] Paul Roetzer: I use it for ideation, for summarization, for strategy, a assist in my creative thinking. So it is, it is a very powerful thing, but it, it is important to keep in mind. On March 14th, when GPT-4 came out, it was already almost seven month old technology. What that means is, generally speaking, these major AI research labs are about a year ahead of what we're seeing.
[00:24:34] Paul Roetzer: So there are more advanced versions. So if you use GPT-4 today and you're in shock at what it's capable of doing, it's almost a year old technology already. Actually it might already be a year old. What I mean by that is there's a lot more coming that is very difficult to comprehend. But GPT-4 was certainly a milestone moment in, the evolution of ai and certainly this year.
[00:25:02] Paul Roetzer: Also that day, actually earlier that morning at number 12 on March 14th, we have Google announces Palm, which is one of their language models, was going to be opened up to developers to build on and that they were going to infuse it into Google Workspace. So if you are a Google Workspace customer using docs and sheets and slides and.
[00:25:22] Paul Roetzer: Whatever their meeting thing is called. I think it's just meeting, Google meeting the AI is going to be infused right into it. Now, this is in, some form of a beta right now. I know there are people who have access to this. the thing that's interesting me about this is we said for the last seven years this was coming, like this is one of those ones where people.
[00:25:42] Paul Roetzer: Just didn't realize the progression of this AI technology. Like we wrote about this in our book that this, that your experience with a lot of these language models was, Google Smart Compose. You know, it was finishing sentences for you in Gmail for the last couple years by predicting words in a sequence.
[00:25:59] Paul Roetzer: You had social media networks where it was recommending replies to posts, text messages, same deal. So we saw the efforts by these AI research labs for the last decade to solve human language. We knew that they weren't stopping at writing sentences and emails. They were trying to get to writing paragraphs and entire documents.
[00:26:19] Paul Roetzer: So in the book we talked about this inevitable outcome of they will write papers for you. It just wasn't reality until chat, c p t emerged. So to see Google here again at number 12 announcing this language model. The key here was it was sort of the realization of what appeared for years to us watching to be an inevitable outcome.
[00:26:39] Paul Roetzer: But this was a complete shift in business strategy for them, a willingness to put these models out into the world, to let other people build on them, to infuse them right into Google workspace. This is a very important moment to, for AI and for me to sort of see this decade plus of studying this space to start realizing these things we theorize would happen to start happening.
[00:27:01] Paul Roetzer: Also that day, again, still on March 14th. And at number 13 we have Anthro announce as Claude, which is their language model. Fast forward, well, actually back up a second. So Anthro is a very important company here. They're one of the companies building language foundational models. They were founded in 2021.
[00:27:22] Paul Roetzer: So I'm, I'm Wiki. This is Wikipedia now, 2021 by former senior members of OpenAI. So they left OpenAI. It was actually, siblings, Daniella and Dario. And they, Daria was actually the vice president of research at OpenAI there. From everything I've read and heard, there was a disagreement about the direction OpenAI was going, especially with their relationship with Marcus, Microsoft, and moving from a nonprofit to a for-profit company.
[00:27:50] Paul Roetzer: And so they left to, found Anthro at which in on May 23rd, so fast forward a little later in the year, a couple weeks ago, they announced 450 million in Series C funding. So you can go, request access to Claude now. One interesting thing is you can have up to about 75,000 words of context and a prompt.
[00:28:14] Paul Roetzer: So, you know, if you think about giving GPT-4 a prompt about what you want it to do and the output you want, it's kind of limited in the information you can provide to it with this language model for anthro, you could. Our, like you give it our entire book. So our book is the marketing Artificial intelligence book is 60,000 words.
[00:28:32] Paul Roetzer: You could use that as the prompt and say, start asking questions about the entire book. So you're going to see this with these language models, that context window, the amount of information you can provide within a prompt and a preset is going to continue to expand. You know, it'll eventually be in theory, millions of of words.
[00:28:51] Paul Roetzer: But that's, that they're a major player. So anthro is one to keep an eye on. So that was March 14th with the three big items on March 14th. On March 16th. This was a pretty important one. So this is the number 14 on our list. The US Copyright Office releases generative AI guidance. In this guidance, we, we spent an episode talking about this.
[00:29:12] Paul Roetzer: We've written about this, so I'll, I'll kind of be brief here. But the simple, thing is you don't own what AI creates. So this has come as a shock, even when I do talks now as recently as the beginning of June, when you explain to people that if you use AI to write something, create an image, generate a video, create audio code, you have no copyright protection for it, and you cannot request copyrights for it.
[00:29:36] Paul Roetzer: So this is a, is it a very important item because there's a lot of fear about AI's just going to replace all writers. Well, I mean, if you care about owning a copyright to what they create, then that can't be true at the moment. So, and for a lot of other reasons as well, but to stick with the copyright purpose.
[00:29:53] Paul Roetzer: So the copyright office released this guidance. They said a human has to author something and a prompt is not authorship. So straightforward. And they even get into examples of prompting and how it doesn't, meet requirements. I will say here, talk to your IP attorneys. If you use outside agencies or freelance writers, make sure that what they are creating for you and providing to you under work for higher agreements, you can actually get a copyright to, and keep an eye on this space because the copyright office also announced listening sessions.
[00:30:24] Paul Roetzer: I do expect this to evolve, but it's probably going to take a, a long time. So again, very important context. Keep in mind at the moment, you cannot get a copyright for any language, image, video, audio code, or anything that AI creates that doesn't have significant human involvement in adapting whatever the AI generates.
[00:30:48] Paul Roetzer: Number 15 on our list still only halfway through March. We are March 16th. Microsoft introduces Microsoft 365 co-pilot in one of the best product demo videos I've ever seen. So, there's about a minute 30 video on this. You can go back and watch. We'll include the link to the post in the show notes.
[00:31:07] Paul Roetzer: I've shown this video now to a number of, people when I'm doing presentations. If you show this video to a group of CEOs, you have to pick their jaws up off the floor. Like once you watch this video, you realize how. Integral generative AI is going to be into every aspect of knowledge, work across all elements, all business functions.
[00:31:29] Paul Roetzer: So marketing, sales, service, ops, hr, finance, legal, it, everyone is going to be impacted by these capabilities, and I have yet to meet an organization that is preparing their, their workforce for this. So Microsoft, 365 co-pilot is available now in some closed beta. They have some customers that are using this.
[00:31:50] Paul Roetzer: If it works anywhere close to what they show in the demo video, it will transform knowledge work by the end of the year for people who have access to it. So if you think about Google Workspace, so we talked about HubSpot and Salesforce infusing generative AI right into their platform. Now, if you have Google Workspace and Microsoft 365, by the end of this year, We are talking about almost all knowledge workers having access to generative AI capabilities in every application they're using to do their job each day.
[00:32:21] Paul Roetzer: There's 132 million job full-time jobs in the United States. A hundred million of those are knowledge workers. People who think and create for a living, all of them are going to be impacted by this. Nobody is ready for that. So this was where things really started changing for me. Like I just started thinking like this.
[00:32:39] Paul Roetzer: I don't think we're ready. Well, I know we're not ready as a society, certainly in the business world, certainly in education. So, you know, it was just one product announcement, but definitely one that really made me sit back and think about the bigger near term impact this is going to have. Then on March 20th, number 16, in our list, more than halfway through now, runway, which is one of our kind of favorite AI companies we've been following for the last three or four years.
[00:33:04] Paul Roetzer: It's runway ml.com. They announce Gen two, text to video. Which they had introduced as a breakthrough in storytelling. Now, this one, it is available. I actually just tried gen two this week. You can get it. You can go to their website, we'll put the link in. But on June 7th it was released to web and mobile.
[00:33:27] Paul Roetzer: And so you can, in the mobile app create four second videos from a simple text prompt. And one of the cool things I saw in there is it'll actually recommend ways to improve your prompt before you hit generate. So I would go experiment with this technology. It's very early, but it's going to rapidly improve both in quality and length.
[00:33:47] Paul Roetzer: So four seconds is just a taste of what it's going to be able to do. So really cool stuff. Very interesting company. They raised 50 million, I believe in December, 2022. And I have to guess they're about to raise a boatload more, or, I would think they're a really interesting acquisition target for like a Pixar, Disney, Adobe, like.
[00:34:10] Paul Roetzer: I mean, it's just an amazing company. So a and they were, they played a major part in stable diffusion, the introduction of that language or that image generation model. So their team, played a major role in the grad that Stability AI kind of took credit for. But it, it actually wasn't Stability AI that, that built that.
[00:34:29] Paul Roetzer: So, okay, so that's runway announces gen two, really cool tech. Go test it out. Number 17 on our list, March 21st, we are still in March. Again, it was a wild month. Adobe announces Firefly, their generative AI capabilities. So the interesting thing here is Adobe is actually one of the software companies that has been investing in AI for years.
[00:34:50] Paul Roetzer: There are a lot of publicly traded and influential SaaS companies that were, Caught asleep at the wheel, I would say on November 30th of last year that I assumed had been doing more with AI than they were. And they have been scrambling ever since to try and figure out their roadmap and their point of view and, you know, what they're going to do about it.
[00:35:11] Paul Roetzer: Adobe had been investing heavily in ai, but amazingly took a little while to figure out the generative AI play. They, they apparently weren't ready, or, or they, they just weren't publicly saying they were ready. But the outward perception is they scrambled to put this together. Firefly is interesting.
[00:35:31] Paul Roetzer: You know, the trick here is it, it is supposedly trained on only images that they had approval for, like, that they had license, license licenses to train on, which means legally it's probably a safer bet, but it's nowhere near as good as like a mid journey. Is, and that's probably because Mid Journey trained on a bunch of stuff it wasn't supposed to train on, so it's a better product.
[00:35:57] Paul Roetzer: It's the challenge of like, where, what are the legal and ethical lines here about which technologies we use? But again, if you're an Adobe user, you, you, you know, fly or Fly was the first introduction. They then in May, late May introduced Generative Phil, which was this really cool technology where you take an image and then just, you know, use a text prompt and fill around it and it's, it's pretty sweet.
[00:36:19] Paul Roetzer: There was actually some really cool Twitter threads about that. So Adobe gets in the game on March 21st, March 22nd, number 18 on our list, AI experts release Open letter to pause AI Development. This came from The Future of Life Institute. I'm not going to spend a ton of time on this. We also had another one, a, a couple weeks ago, saying that from the Center for AI Advocacy.
[00:36:44] Paul Roetzer: Is it something Center for ai? Something. Basically there's, there's this movement of some very influential AI researchers that, worry that we are moving too quickly and that these really advanced models like G B T four and whatever comes next are a threat to humanity. It is, I've talked about this at length on the podcast, which why I'm not going to go into great detail here.
[00:37:07] Paul Roetzer: It's something to be very aware of that there are a lot of important people who've spent their lives in this space who do think that there is a significant, risk of harm and downsides to this. In the near term. My general feeling is I'm really happy that there are, very smart and accomplished people thinking about this and working to solve it.
[00:37:29] Paul Roetzer: I think some of the way they're messaging this is distracting from the near term issues that we face in our, election cycles. The disruption to our education systems, the workforce, the things that are very obvious that don't take a leap forward in, your assumptions of where AI's going to go. If we just focus on where AI is, it's already going to have major disruptive impact on economy, workforce, education systems, our ability to know what's real and what's fake online, you know, synthetic media, all of that.
[00:38:03] Paul Roetzer: You don't need any advancement in AI capabilities for those to be very real issues. And my, my main concern is there's just too much focus in the mainstream media and even within AI research community on these long-term, you know, threat to humanity kind of stuff. Again, not that it's not, doesn't matter, and not that it's not a possibility.
[00:38:23] Paul Roetzer: I just feel like it's disproportionately being spent on stuff that we don't actually know how it would occur versus the stuff that is right in front of our face and we don't seem to be solving fast enough. So, number 18, our list March 22nd, the Future of Life Institute letter. Yeah, lot more on that on past episodes, but I'll, I'll, I'll pause there.
[00:38:45] Paul Roetzer: My thoughts on that one. Number 19, March 23rd, open AI announces ChatGPT plugins. These are things that pop into ChatGPT, that expand its capabilities, the ability to take actions, connect to browsers, analyze data, it seems like it's been a slow build. I heard a quote recently from Sam Altman, the co-founder and c e o of Open Eyes saying they don't have product market fit for these plugins yet.
[00:39:08] Paul Roetzer: That people actually just want chat G B T capabilities in the platforms they already use. Like HubSpot and Salesforce not to come in there and build a plug-ins. So it, it looks like it has the potential to be like an, like an iPhone app ecosystem kind of moment where everybody builds all these apps to be on that platform.
[00:39:29] Paul Roetzer: But it doesn't sound like that's materializing yet, in large part because a lot of the plug-ins are useless. And so it's really hard to go in there right now and find ones that are actually like worth your time or that you're willing to give up the data. You have to, again, go with the law of uneven eye distribution.
[00:39:45] Paul Roetzer: I don't even know who's building some of these plug-ins. So am I willing to put data in to have it analyzed things to get value out of it? Have no personal information. I don't even know who they are. So I just, I feel like it still has a ton of potential, but it seems like it might be a slower build versus a fast takeoff for the plug-in ecosystem.
[00:40:05] Paul Roetzer: Number 20, March 30th. The date this actually occurs is kind of hard to nail down, but it seems March 30th is around the date auto GPT. So you may have heard this term. The basic premise is rather than generative AI agents that just create outputs, these are a AI agents that take actions on your behalf.
[00:40:24] Paul Roetzer: So imagine, you know, going and, completing a booking for you, reserving, a table at the restaurant or making a purchase of your grocery list or sending emails for you, like whatever, this whole premise of action. And we talked, we had a episode, of the podcast where I talked about my theory of action, transformers and I, and why this, AI leader Andres Kapai, went from Tesla, left Tesla as the head of AI there, and went back to open AI to work on these action transformers.
[00:40:51] Paul Roetzer: All these research labs are working on the ability to give AI agents action capabilities. So again, rather than doing just outputs, they can go do actions on your behalf. So this is an inevitable outcome of where this goes. Auto GPTs are, are very, very early and most instances of them being talked about, it's like, yeah, it works sometimes.
[00:41:16] Paul Roetzer: So people are building some interesting things, but don't feel like you've missed this boat if you have no idea what I'm talking about with auto GPTs. But the idea of AI agents taking action is a very real near term component of what is going to happen that will have a significant impact on knowledge work and it will affect your personal life.
[00:41:36] Paul Roetzer: So almost fast forward. I think, you know, if you look at a goal of what a lot of people are building, think of it as a true AI personal assistant, you know, in your personal life and then in your, in your business life where you have. Almost this like super powered intern. That can not only provide information for you, but it can actually go take actions on your behalf, like an assistant or an intern.
[00:42:01] Paul Roetzer: That's, that's a very real thing that these labs are trying to build. And I do think we'll probably see some more milestones in this space before the end of the year is up. It's hard to kind of predict who, where they're going to come from and what they're going to look like, but there's no doubt that the ability for AI agents to take action is something that is believed to be possible and is being worked on.
[00:42:24] Paul Roetzer: March 31st, number 21, Italy Bans ChatGPT, they have since I believe, removed that ban, which is interesting cause I'm doing a talk in Italy in the near future. And so I wasn't sure like if, if chat b t was even going to be allowed when I was doing the talk. So the um. There was talk about, other el other areas in the EU banning ChatGPT as well.
[00:42:48] Paul Roetzer: It seems like that's backed off recently, but I will say at the moment, to my knowledge, Google still is not allowing their generative AI tools to be used in the eu. So it's four 50 million, consumers in the EU that don't have access to Google Bard unless they go through a V P N. So there's some battles being waged behind the scenes.
[00:43:08] Paul Roetzer: The EU is trying to push through a AI act that would be pretty restrictive on some of this technology. Google doesn't appear to agree with that act, so they're just saying, fine, we just won't even let our technology be accessible in eu. So keep an eye on on that. All right, so that wraps March. We're at number 22 now, moving into April.
[00:43:28] Paul Roetzer: There's just a couple items in April. Things slow down. Apparently. We have April 13th. Amazon announces generative ai and a w s interesting thing here is what does Amazon do? They're the everything store. So they decided let's be the everything store for language models. So they partnered up with a number of language model companies, including AI 21 Labs, an Thropic we talked about earlier, Stability, ai, who we talked about earlier.
[00:43:52] Paul Roetzer: I don't think coheres in there, but at least at the time of the announcement it wasn't. And then they also announced their own foundational model called Titan. So their premise is, rather than going to Microsoft and getting Microsoft slash OpenAI, you could come to Amazon and you can kind of build a symphony of language models based on whatever your needs are, what industry you're in.
[00:44:14] Paul Roetzer: So it's an interesting play. I haven't heard honestly too much about it since that time. But that was their first foray into, you know, making a major play. Now obviously Amazon, AW w s is like the dominant player in the cloud, and they have like dozens of pre-trained models. Again, if you read our book, chapter one, we talk about the significance of Amazon in this space and the play they've made to have all these pre-trained models.
[00:44:38] Paul Roetzer: So again, Amazon is not new to AI at all. It powers their own business. They've been building models and sharing models for the last decade. Generative AI is the kind of the new space here for them. Then on March 20, or April 23rd, number, April 19th, number 23 on our list, I published something, where I said how AI could impact millions of knowledge workers in the next one to two years.
[00:45:05] Paul Roetzer: So again, this goes to, I've just as we have these conversations these week, we do the podcast episodes. I'm on the road talking to lots of different audiences and. Having Zoom meetings with all kinds of different leaders. You're trying to kind of piece together what is going on and what could happen.
[00:45:20] Paul Roetzer: And so I'd sort of settled on this belief that I think there's a greater than 50% probability that we lose millions of jobs in the next like 24 months. And so to say something like that, I feel like you have to have a really good reasoning. And so I went through and kind of played out why I thought that was possible, what, what I thought could happen to prevent it, and then what actions organizations should be taking.
[00:45:50] Paul Roetzer: So there's a whole blog post dedicated this, there's a whole episode dedicated to this. If. If this is an interesting topic to you, I would suggest going and listening to that episode and reading the post because it really plays out why I think this, and again, I'm not obviously cheering for this. I do think that it is not inevitable.
[00:46:09] Paul Roetzer: I believe that there are things that can occur, even the macroeconomic perspective, geopolitical perspective, that could prevent this from happening. But I think it, there is a probab a high probability that these jobs could not only be impacted in changing the way they work, but that they even are needed.
[00:46:28] Paul Roetzer: So I hope that that is not the case. I think we still have time to take action. But really what it comes down to is each individual organization committing to that human-centered approach to the adoption of this technology. So go check that out. It was, it was an important moment for me because it just helped me sort of center around.
[00:46:48] Paul Roetzer: What I thought the future could look like. And I think about this post, this thought process every day and try and make sure that we're taking whatever actions we can to ensure the most positive outcome for, for humans in this scenario. All right, then we jump into May. So we are almost kind of, kind of home coming down the home stretch here, May 1st, number 24 on our list.
[00:47:10] Paul Roetzer: Geoff Hinton announces he has quit his job at Google, so Geoff Hinton is godfathers of modern ai. His team sort of set off this deep learning movement that has led to the generative AI phase we are currently in, and he's been at Google since he sold his company to them. I think it was back in 2011 or 2012.
[00:47:29] Paul Roetzer: There's an amazing book, genius Makers by Cade Mets of the New York Times. I would highly recommend reading, if you want to kind of go into the history of the last 12 or so years of deep learning, starting with Geoff Hinton and him selling his company to Google. It's just a fascinating story in Cade had inside sources at OpenAI and Google and Microsoft and I do, and all these amazing AI companies, and it's just, it's just a phenomenal story.
[00:47:57] Paul Roetzer: So if you want to know more about Geoff Hinton, kind of the history of why him leaving Google and saying AI could destroy humanity and I regret in ways my life's work. You'll understand more if you read the Genius Maker's book. And Cade was also the keynote for our marketing AI conference in 2021. Yeah. So, and that keynote fireside chat is actually on our YouTube page, so you can go listen to the interview about it.
[00:48:23] Paul Roetzer: I'm not going to say much about this one. Basically like this is part of what's set off this whole ai existential threat to humanity thing, that we have talked about at length and I referenced earlier. It is significant. He is significant. It is one perspective. And I think what you have to do with all of this is you have to take in all these perspectives.
[00:48:45] Paul Roetzer: There is no one person who actually has a clue where this goes or what happens. You have to think about and listen to the perspectives of a lot of the different parties, inspe in some ways, the protagonists who annoy me significantly. I listen to what they're saying though, when you try and find the thread of where they're coming from and why they're so strong in their opinions.
[00:49:09] Paul Roetzer: And you try and synthesize that and find the middle ground and say, okay, what is everyone actually worried about here? What are their incentives to be talking about the stuff they're talking about? And then, and then you kind of form your own opinion. So hopefully our podcast helps you do that. That's what we try and do is give this like very balanced middle of the road, here's what everybody is saying and thinking and present that information in a balanced way so that you can kind of synthesize it and figure out for yourself how you feel about these topics.
[00:49:35] Paul Roetzer: Number 25 on May 10th, Google announces generative search experience. So again, now we're starting to see Google's urgent response to open a Microsoft back at the end of 2022, starting to find its way into their core offering of search and ads. And we don't know where this is going. I get asked all the time about the future of search organic traffic.
[00:49:59] Paul Roetzer: It's going to change. I don't know that. Google and Microsoft know exactly what that's going to mean or how it all plays out. So you have to pay attention to this space, especially if you're involved in marketing and search it. You're how you do your job and the KPIs you report on and how people find your brand and consume information and buy your products.
[00:50:19] Paul Roetzer: It is all going to change and change dramatically in the next two years, but nobody knows what it looks like. So you just have to stay kind of at the forefront of this. Pay attention to these major announcements and then try and figure out what does it mean to you and your company. Number 26, May 16th, Sam Altman testifies before Congress.
[00:50:41] Paul Roetzer: Again, I'm not going to go into great depth here. We did a podcast, there was actually four hearings that week. The one with Altman and Gary Marcus, and the executive from I B M got all the headlines, but it was probably the least significant of all of them. My general belief in what's going on in White House and Congress and the US government is they don't understand the technology largely.
[00:51:02] Paul Roetzer: They're trying to figure it out and they're trying to determine what the public cares about so that they can message around it in this election cycle. So I don't, I don't, not that I don't think there's some altruistic reasons why they're doing this and that they don't think regulation is important and that they don't need to figure out how to apply existing laws to stuff.
[00:51:21] Paul Roetzer: Like, I think all of that is true, but when we're seeing hearings like this, I think most of it is trying to comprehend what's going on. Does the public care, which parts of it do they care about? And then, because the reason I think is it's so bipartisan right now, like, No one is taking a side on any of this really.
[00:51:42] Paul Roetzer: And in part it's because they use the tech in their election campaigns. So a and they, they see our ability to advance this AI tech as a differentiator in the global battle for AI supremacy with other countries. It, it's just a really complex topic that has lots of threads to it. So nothing is what it seems is, I guess what I'm saying here, like these headlines you read thinking you see Sam Altman in front of Congress, that all of a sudden we're going to have some regulations.
[00:52:12] Paul Roetzer: That is not what's going to happen. This is largely a pr play to, for one of the tech companies to have a say in whatever future regulation comes, and two, for these elected officials to figure out how to play this out in the upcoming election. I don't, I don't know what else to say. I think that basically covers my overall feelings on it.
[00:52:36] Paul Roetzer: You can go back and listen to past episodes and kinda Yeah. Hear more. Okay. Number 27, May 30th, our last one in May, we have the, statement on AI risk, again, that AI poses a, a threat to humanity. Okay. So Center for AI Safety, that was the organization I was trying to think of earlier. The statement was 22 words.
[00:52:58] Paul Roetzer: It said, mitigating the risk of extinction from AI should be a global priority alongside other societal scale risks, such as pandemics and nuclear war. That is the entire statement. A bunch of really important AI researchers signed it. They know more about this topic than I do. I think they believe this to be true.
[00:53:18] Paul Roetzer: Generally, it's really hard for any of them when they've been asked point blank to explain how exactly. This happens. How is this on power with pandemics and nuclear weapons? They're playing out a future. They're looking at scaling laws of this technology and saying, well, in five to 10 years it's going to be able to do this and then we're going to be in trouble.
[00:53:36] Paul Roetzer: It can become super intelligent, destroy us. That's kind of how they're thinking about this. So again, it matters. I'm glad they're thinking about it. I'm glad organizations are working on solving for it. I'm just more concerned with near term stuff, so I just wouldn't get caught up in these headlines and like be losing sleep over at the, at the moment.
[00:53:58] Paul Roetzer: But they're very real risks that we will have to face and I think it's important that some people are working on this right now. Okay. And then finally we're going to wrap with June. A few items snuck in. Before I recorded this podcast, we had June 5th at number 28. Apple announces Vision Pro. They're not, metaverse, never used the word metaverse in it.
[00:54:20] Paul Roetzer: Not even ar vr, it is spatial computing. The reason it makes this list is because there is an insane amount of artificial intelligence that makes this product possible and it, it looks interesting. I don't know if you're going to spend $3,500 on it. I think it comes out actually publicly available in early 2024, I believe is when it's going to be available.
[00:54:42] Paul Roetzer: But I was actually more interested in the AI side of it and everything that goes into it. I do, as I mentioned earlier, think Apple is sort of the dark horse in all of this. There is, I don't know that it's going to happen, but to me, if Siri becomes what it was supposed to be and we think a future where AI personal assistance are everywhere and everyone has one, that to me is the leading potential player in it.
[00:55:07] Paul Roetzer: So if Apple, because again, they quietly build all this stuff. They never talk about anything they're building, if they're building their own foundation, foundational language models within the organization and they're planning on enabling action transformers to occur right on your phone. I really think that Apple could end up being the most important company in this story that no one's talking about.
[00:55:28] Paul Roetzer: They have thousands of AI researchers, they are genius at product innovation, and they have the most significant distribution of anybody given the iPhone that we all carry around in our pocket. So just interesting. I follow Apple closely. Unfortunately, there's just not much to follow because they never talk about what they're doing.
[00:55:51] Paul Roetzer: Alright, June 7th, 2000, 23 number 29 on our list Deep Mind, which we referenced earlier. Google Deep Mind is the AA research lab within Google, run by Demi Sabba, who I talk about a lot on the show. They announced, this one's kind of more geeky, but it's interesting announces Alpha Dev breakthrough and sorting algorithms.
[00:56:12] Paul Roetzer: I'm just going to read real quick why I think this matters. It says New algorithm algorithms will transform the foundations of computing. Digital society is driving increasing demand for computation in large part because of AI and energy use. For the last five decades, we relied on improvements in hardware to keep pace, but as microchips approach their physical limits, it's critical to improve the code that runs on them to make computing more powerful and sustainable.
[00:56:40] Paul Roetzer: So basically, to build these really advanced models, we're thinking about these scaling laws aren't going to work out in just the hardware side. There has to be smarter ways to do the underlying algorithms says this is in especially important for the algorithms that make up the code. Running trillions of times a day.
[00:56:57] Paul Roetzer: In our paper published in Nature, we introduce Alpha Dev, an AI system that uses reinforcement learning to discover enhanced computer science algorithms, surpassing those honed by scientists and engineers over decades. and again, this goes to like. We talked about what are the, what are the things that excite us about artificial intelligence?
[00:57:16] Paul Roetzer: In a recent episode, scientific discovery on breakthroughs that humans haven't been able to solve is really, really interesting and exciting. What that could make possible, and this sort of breakthrough, this sort of research paper, is the kind of thing two, three years from now, you will see a 60 minutes episode about, and the impact it had on all these different industries.
[00:57:37] Paul Roetzer: So then this ends with, Hal Alpha Dev uncovered a faster algorithm for sorting, for sorting, a method for ordering data. Billions of people use these algorithms every day without realizing it. They underpin everything from ranking online search results, search results, and social posts to how data is processed on computers and phones.
[00:57:59] Paul Roetzer: Generating better algorithms using AI will transform how we program computers and impact all aspects of increasingly digital society. So again, The average person will never know what Alpha Dev is, won't have ever heard of it. Hopefully now, because you listen to the show, you can impress your friends and talk about Alpha Dev.
[00:58:18] Paul Roetzer: I, they just, deep mind comes out with the most insane stuff and like, I'll go give talks and ask who knows who Demis Hassabis is? And no one raise their hand raises their hands. This is like every time no one knows what AlphaGo is. They don't, they don't know Alpha Zero. Like they have no idea. And yet I think Demis and what they're doing at DeepMind is maybe some of the most important work in human history.
[00:58:41] Paul Roetzer: So again, if nothing else from this, like go follow DeepMind and Demi Saba and his team Shane Laggan, the people at, at DeepMind, they were working at the forefronts of like the frontiers of advanced AI and biology and energy. Like just amazing stuff. So yeah, that's my like plug for Denis and DeepMind.
[00:59:02] Paul Roetzer: And then the last one, that makes the cut here is cohere, which another language model company we've talked about a lot on the show announced a 270, 200 70 million series C, which I'm sure values them well over a billion, probably a billion and a half to two. I'm not sure what the number is. They announced that.
[00:59:20] Paul Roetzer: I think the key here is they're making an enterprise AI play. So I've said before that I think large language models are going to be as critical to organizations in the next decade as CRMs have been in the previous decade. So if you think about how core the CRM is to everything we do in marketing, sales, service, and beyond, that's what I think is going to happen with large language models.
[00:59:39] Paul Roetzer: That every company, especially every enterprise, is going to have custom built and trained versions of large language models that power. The generative AI within their organization where all the data can stay private and secure and and proprietary to them. And then they can use that in marketing, sales, service applications and beyond.
[00:59:57] Paul Roetzer: And that's kind of what Cohere is hinting at here with what their direction is. So it's said in the release, the platform is built to be available on every cloud provider deployed inside a customer's existing cloud environment, virtual, private cloud, b, bpc, or even onsite to meet companies where their data is.
[01:00:12] Paul Roetzer: This empowers businesses to transform existing products and build the next era defining generation of innovative solutions while keeping their DA data s secure. So again, I think just a prelude to where a lot of these companies are going to go is they're going to move into the enterprise, they're going to custom build things.
[01:00:28] Paul Roetzer: So you're not going to be just going and using ChatGPT and having all your employees randomly using stuff and putting a bunch of confidential information into these models. You're going to have your own model where you know all your data is secure and that you can build on top of. All right, so that, that is it.
[01:00:43] Paul Roetzer: That is our 30 top things. It is kind of crazy to look back at just these first, not even full six months yet of the year, and think about everything that has already transpired and to try and consider what could possibly be coming next. It, it's really just amazing time. It is amazing to be living through this phase in human history and certainly the impact it's going to have on businesses and society and education.
[01:01:10] Paul Roetzer: And, appreciate you being along for the journey. I mean, it's every week we're just trying to figure this stuff out and hopefully us trying to make sense of it out loud is helping you figure it out as well. So join us, next week for episode 52. It's going to be another special edition cause again, Mike and I are sort of on, I don't say summer break.
[01:01:27] Paul Roetzer: We're both on speaking tours, mixing in a little vacation along the way. But 52, our Chief Growth Officer, Cathy McPhillips, is going to join me. And what we did is we've been teaching this intro to AI for Marketers class since November of 2021. I think we've done 26 episodes. We've had more than 11,000 people register for this free class.
[01:01:47] Paul Roetzer: I do it like every three or four weeks roughly. And the best part of it each week for me is I present for 30 minutes, or each time I do it, I present for 30 minutes. Then we do about 30 minutes of q and a and we're averaging almost a thousand registrants per class now in 2023. And so we've been getting, you know, close to four to 500 people showing up for each class.
[01:02:09] Paul Roetzer: We've been getting dozens of questions at the end of each episode for each class. And so what we did is we curated the questions from 2020 threes class. I think we've done seven. In 2023. We took all the questions and we curated them and kind of summarized 'em. Used G PT four actually analyze 'em, and we settled on 15 common questions that everyone is asking.
[01:02:30] Paul Roetzer: And so episode 52 is Cathy who moderates the class for me and does the q and a in the class. She's going to moderate the Q, the q and A with those 15 things. So hopefully those are really valuable. So hopefully both these episodes are just packed with a ton of value for you, really give you a good perspective on what's going on, where we are in the state of ai, and then helps you sort of make your plans for the next half of the year.
[01:02:53] Paul Roetzer: So thanks for being with me. After episode 52, Mike and I will be back, episode 53 for our regularly scheduled weekly, and we'll try and catch up on. Any of the items, news items, and major milestones that we've missed along the way. So thanks again for being with us. As always. I will talk with you soon.
[01:03:10] 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.
[01:03:31] Paul Roetzer: Until next time, stay curious and explore AI.
Cathy McPhillips is the Chief Growth Officer at Marketing AI Institute.