45: Sandy Carter Unveils the Future of AI: Myths, Agents, and Creativity | SXSW Insights

Hey, so we have a special guest on today. She just came back from doing a keynote at south by Southwest. There was lines of people outside waiting to get in and she was number two rated last year. So she's a rock star Sanity Carter. And she did a Talk on the 10 Myths of AI. So let's get into it. I don't even need to ask. Who are you? Who are you?

I am Sandy Carter. I am an alumni of IBM and Amazon Web Services and I'm currently the COO of a startup called Unstoppable. And oh, I forgot to tell you this, Robert. So we launched my book AI first at the conference and we sold over a thousand copies at south by, which is pretty darn good.

Yeah. So I heard you had a lot of demos on your keynote and you rocked the house. I mean, anyways, what, what, what happened at South By? Because I've been to south by 25 times and, and sometimes that conference is really important. Like Twitter launched there. Right. And many, many interesting things happened there over the years. What, what, what, what, what's the word on the street? How was the barbecue?

The barbecue was phenomenal. And the text was even better, I have to say. I will also say that it was really fascinating. This year they had a couple of days of marketing and even though it wasn't in the AI track, a lot of the marketing was about how can AI help, how can it assist, what can it do? I thought that was fascinating. It's the first time that tech really invaded the marketing track, which sucked with such fervor. It was quite fascinating. The other hot topic for marketing was humor. Lots of reports that humor gets you 65% higher engagement. So everybody was like trying to figure out how to be funny.

Yeah.

And then we had the hard thing to pull off.

Unless you're funny.

It is true. It is.

I should do some stand up comedy and learn to be funny.

But how do you. Yeah, how do you, how do you fake funny? And then the other big track was artificial intelligence. Yeah. And that was really fascinating. So we had. And one Will I am was there and he showcased AI that spoke like him, like his mother, like a friend of his who's a Hispanic woman in the jargon. And the, it was, it was really cool without the robotic voice and more authentic that. More transparency. And he also talked a lot, Robert, about his vision for the new radio and how it's going to be customized to you.

Yeah, well, we have like Google's Notebook lm where you stick a PDF in and then it'll make a little podcast for you to listen to on your way to work or something like that. I think Will I am's onto something. And, you know, Grok Voice came out a couple weeks ago or a few weeks ago, and it has a lot of different personalities, right? From an unhinged one, which is really not nasty to, you know, but he doesn't have a broad, diverse set of voices or personalities. And there's communities that will want to talk to AIs that sound like themselves. And we might say that's DEI, but it's not really. It's trying to meet the customer. And this gets to a bigger point. There's a huge gap. And I think you and I were talking about before that the normal everyday person on the street is not using AI at all. I mean, or if they are, they're using it sort of like they use Google, you know, sticking in a couple words and getting a search result back. They're not using it to talk to it all day long and, and using it for their business or their, or their personal life. Right? And really asking a deep, interesting question and having a conversation. How do you see this world right now? Because you, you just were at south by hanging out with the people. And I know you've done a lot of, a lot of traveling, a lot of talking to people in this industry. Where is things give us a state of AI.

I have to say, there is this huge, you know, we usually talk about the digital divide. I call it the AI divide. Huge out there, right? You go to Silicon Valley and you feel like you're behind because everybody's talking about the next, the next, the next. And then you go out like where the people are and the understanding is so much less. So I'm actually right now living in Arizona for lots of different reasons, and I didn't have a lot of tech here, so I went out and joined the Small Business Association. I've been doing talks for veterinary associations, florists, dry cleaners, people. I never thought that I would say I'm going to be speaking to them and helping them with AI. And Robert, these are like successful business leaders who are curious about it, but just don't know about it. And I think the big difference in my mind is that they're all listening to it. They're hearing it, but they're not doing it. I call it kind of the difference between brainstorming something and play storming something. When you get in there and try ChatGPT or Claude or Grok or Perplexity, you really see its power, right? You don't just put a prompt in. You're like thinking about how you build up your whole ecosystem, how you can change many things. And I think that's what we're not seeing today. And I saw a lot of that at south by. As you know, south by has, you know, people in all different fields, lots of consumers, and the levels were just so incredibly different. In fact, one person came up to me and said, hey, he has a trucking business. And he was like, yeah, I got this really new thing, it's pretty secret, hush hush. He goes, can I share it with you? You might have heard of it. I was like, sure, what is it? He's like, it's called Gemini. And I was like, okay, it can't be Google Gemini, right? But it was, it was Google Gemini. He just didn't know that everybody could use it today. So I think that AI divide is a really big one and something that we, as you know, AI professionals need to address. Otherwise all these people are going to be left behind.

Yeah, and I have the same problem. You know, I try to live AI first, but I don't always do it. You know, I don't do a lot of programming and I'm starting to use it for programming as freaking amazing at doing, you know, a whole range of things. Even as a non pro, I don't, I can't read the code it generates. But it's like, wow. And you know, most people haven't even tried. Can you build me a Tetris game or something like that? You know, on Grok or Gemini or Perplexity or Claude or Chat GPT. By the way, these peop. The other people have no idea, like you said, they didn't even know Gemini was Google, you know, was. It was something that's been being built by Google and is being built into the Google ecosystem, right?

Yeah. Well, one of the demos that I showed in my session was really super simple. I used code from a friend of mine, he's an ex Amazonian. It's from AI explained. And what I wanted to do is I wanted to show them that I would build an agent and the agent would do something very simple. Robert I get lots of phone mail messages on WhatsApp and they're like, hey, can you meet on Tuesday? Can you meet on Saturday? Can you meet here? And I just don't have time to listen to all the messages. So I created a little demo to go out there and have AI explain, grab my messages from WhatsApp, put them on my calendar and ask me if there are conflicts and ask me, hey, do you really want to meet with Robert or no? Should I politely decline that? And so I just built that little agent. Literally. It didn't take me very long at all, maybe 30 minutes. Built that and now all that's automated. I had people after my session, after they saw that, they said, can I buy that from you? I'll pay you for that. And I was like, you can do that yourself. Like, you can literally do that yourself. Like, you don't need to pay a program. You don't need to pay me. You can literally do that yourself. And so again, it's that AI divide that's out there. I wanted to show people that they could do it. They just have to play with it and try it. It's. I think it's getting easier every day.

What were some of the myths you talked about on stage?

Yeah, I mean, a couple that I thought were really fascinating. One was about LLMs. You know, we hear a lot about LLMs and LLMs I think are important, but people think that AI equals LLM. And so one of my myths was it's only about LLMs, but in reality now it's about the apps, it's about the AI agents. And so I walked people through how now the LLMs have really gotten reduced differentiation because of the open source. Like look at, you know, meta with llama or Deep Seq. The differentiation is lowering the performance and cost is changing. And so we are seeing, you know, that that's becoming more of commodity like. Except we, we do know and this was fascinating. I had people come up to me from different countries. So we know the difference of putting something in perplexity and then asking deep seat. You're going to get two different perspectives because one is China and one is US trained. I actually met someone who is driving the India LLM, Baha' I GM pmt. Someone who was driving the Italian LLM called. Someone who's driving the Brazilian LLM called Tucano after the toucan. And then of course, you know, I already had met the person from Saudi Arabia who's doing a river. So I do.

I mentioned somebody who's building an African LLM for Tanzania. Just for Tanzania.

Oh, wow. Yeah. So that. That's differentiating there. And then I also have for people.

Who don't know what an LLM, that's large language model. That's the engine underneath these systems. Right. Like chat GPT. Right, right. And, and you're right that large language models that the technology that makes this model is not what. What drives the Tesla car for or the Waymo car. That's a different kind of AI model. It's still AI, but it's a different branch of AI.

Right, that's right. And then of course I also introduced this concept of industry based LLMs. Right. We happen to be in a an IBM sponsored track. So I did research on the granite LLMs which are divided by industry, which is also interesting. Right. And then I also spoke at Grace Hopper, they had Grace Hopper there, you know, by Anita Borg, for International Women's Day. And they talked about building a women's LLM, which I had never heard of before. Yeah. For terminology. Just like you do for, you know, for different countries, just like you do for different industries. It was fascinating. But when you get. So let's put LLMs aside because I think they are being commoditized, still super important. I'm not saying they're not important. They are every day.

I just retweeted Nvidia just five minutes ago, put out another new LLM. Right. So these models just keep coming out of the sky.

Yeah, I completely agree.

There's a lot of innovation going on is what we're saying.

But, but I do think the new. I think what people need to now pay attention to are the agents. So A to a agent to agent, B to a business, explain what an.

Agent is for people who might not know the terminology. And what is an agent?

So an agent is, it has the ability to make decisions on its own. It doesn't have to just be an information finding facility. It has the ability to autonomously make decisions. So it performs tasks. It can make a decision for you or it could act autonomously. Right. As well. So these important decisions, like a personal assistant, I'm going to make a decision about booking this on the calendar. Agents can also be things like performing specific tasks. Like I'm going to gather this data, I'm going to automate a workflow. So for example, a web crawler, I met this guy at south by Southwest that he's got now an agent. And what it does is it scrapes all the AI news at 3am in the morning and then he uses another agent to turn that. He uses 11 labs, but he uses that to now turn all that news into a podcast that broadcasts every day. And he said he basically spends three minutes on it and it's like one of the top podcasts. So that's another. Right. I'm sweet.

Yeah. 11 Labs does the voice. Right.

So 11 Labs does a voice. Yeah, yeah. And then there's multi, Then there's multi agent systems too. So I showed what I really want is a vacation planner, right Where. And this really gets into the agent to agent and the business to agent and the person to agent. So let's imagine that in the future I have my own agent. It's the Sandy travel agent. It has all of my preferences in there and then it's talking to me. So agent to person, and then it interacts with a flight agent. So agent to agent. So that consumer agent's interacting with the flight agent, trying to book my flight. Well, the flight agent is then interfacing with a business agent, Expedia, American United, and that's business to agent. And so, you know, and then this automatically books my flight, my hotel, my restaurant, my activities for all my kids, my car rental. All that gets done in kind of this agent economy, which I think is pretty powerful.

Yeah, there's a lot happening and agents are real important in business, you know, to, to do market research or watch your books, even hook into various systems and give you reports. You know, you can automate a lot of things, which, that's advanced math to me. Right. How do you automate a business? How do you take a business and, and make it more and more automatic? People are still struggling just even adding AI onto their email system. Right. To help them with their email. And there's a lot happening there. There's systems like Shortwave that can hook in. Tell me about some of the other myths. I mean, because you're, you're an interesting person because you meet so many people from so many places. Right. And, and that's, I think that's why your keynote is always rocking because you always have these great demos and, and myths. Tell me about some of the other myths.

So the other myth is AI is all about software. I actually had a professor come up to me and say, you know, all these videos you see about robots and humanoids, they're all fake or you know, they filmed at times and they got one good take. I don't believe that. I think Amyth is. AI is all about software. I think the next phase will be physical, physical AI with humanoids and robots. And in fact, the numbers show it, the facts show it. Fortune just published a report that showed that it's going to grow to 66 billion here in the next couple of years. 45% and my favorite basketball team. So I love Steph Curry, the Golden State Warriors. I showed Steph Curry having a robot go fend for his ball, like when he shoots it grabs the ball, it sends it back to him, wherever he is. It also gives him shooting tips. Not that Steph Curry needs them. Right. Hey, you got, you know, got this off, that off. It also showed the coach for the Golden State warriors with five robots on the court and he's running plays with these robots to get the plays right ahead of time. Obviously, you know Anina Net, she's doing the dancing robots in China. And then of course on stage I had Dale Damasi from Psionic and what they're doing is he has, he was born without his arm. They've created a bionic AI arm for him. And with that bionic arm he actually use, it uses A.I. it actually kind of reads his mind through sensors and tells his arm what to do. Right? Open, close. I'm going to pick up an egg. So I'm not going to squish it. Going to shake someone's hand. He can actually feel like now when he shakes my hand, he can actually feel that he's shaking my hand. It was so powerful. So he was up on stage. I think he got, I don't know how many people asked for him to come speak because it's pretty powerful. That physical manifestation of AI, where was.

The sensors on his arm or, or did he have sensors on his head?

Yeah, he's got it here in his head and he's got them right here on his arm, like right, right here above his elbow.

Because this is a big trend coming, right? Brain computer interfaces. This is a one from next mind which Snapchat or snapbot and it goes on the back of your head and it can sort of tell what you're looking at by just by listening to the electricity coming out of your. Which is mind blowing field to study and will have big impacts over the next decade. Might be your next keynote next year.

Yeah, it might be. It might be. But it was really cool because you know a lot of the. I mean I wanted to show it, right? Like I have some videos, I have some demos, but him coming on stage with the actual hand, he gave me a fist bump, he gave me a high five. I mean people could see. And the one that he was wearing is his what he calls his stage arm. So he's got a normal arm, right, that looks like he has an arm. This is one with lights in it. So anytime a signal is going from here all the way down, you can actually see the signal. And then when it goes processes back up, you can see it. It's super interesting to See where this whole field is going for sure.

Yeah. Brain computer interfaces are a big deal coming, but most people haven't even gotten into using AI just to talk to it yet. What are some of the other myths you. You had at your speech?

So the other myth that I did, because, you know, Robert, you and I have been there a lot. We have lots and lots and lots of marketeers there. Yeah, I love marketeers. So the myth that I busted there was my brand won't be impacted by AI. I recently spoke at a CMO conference and CMO said, look, you're going to automate my campaigns, my performance management, okay, My reports, but not my brand. Not my brand.

Yeah.

I also had combined it with the myth of SEO is here forever. And so, you know, we chatted a lot about how SEO is moving to GEO Generative Engine Optimization. So they're changing one letter on you, the S to a G. Yeah, Already starting to see that movement. In fact, Gartner predicts that by next year, SEO will go down by 25% and it will all move to GEO to Generative Engine Search. And then, of course, go ahead.

Let's just talk about the marketers. While you were on stage, I put up a video from the founder of NextArt. He's building a new model called Marco, our new service, that you scan a product like this hint water bottle, and it then you can generate that hint water bottle in anything. Right. You can say, put the hint water bottle in a person's hand riding a horse or whatever, and the models are just starting to get really amazing where even the text on the product doesn't get messed with. Right. When you and I started out with AI, you did something like that. It would put six fingers on your hands and all sorts of fun stuff.

Right.

And the models are just starting to get really, really high quality purposes. Right. Do you agree with that?

Oh, I absolutely agree with that. And you know the example you and I chatted about Disney, right? Where you go and you take a picture, you put it in chat GPT, you put it in Grok, and it gives you the story, it gives you a social media caption to post it. I mean, the impact to your brand is powerful. I met up with Farfetch. I don't know if you know Farfetch'd. They're a luxury retailer online. I think they're only online and this was fascinating. So, you know, I don't know about you, Robert, but I like to shop online. In fact, virtually, that's all I do is shop online. I don't do any, really any in store. But the AI nudge that you get when you fill up your cart and you leave, you know that little nudge that comes that says, ooh, sandy, that blouse is almost ready to sell out, or ooh, blue is your color, you've got to get that blouse. You know, those type of little AI nudges. Farfetch is starting to do that and they're doing curated product suggestions and they told me that 75% think about this. 75% of their purchases were influenced by an AI nudge or recommendation. AI nudge or recommendation. 75%.

Yeah.

And that's a, you know, that's huge brand impact. Right. I think these hyper personalized value props that come in from AI is going to be the future. In fact, I think anybody who's going to continue to use one size fits all marketing or, you know, segment based marketing or all that stuff that we used to do, I think they're going to be left behind. It's going to be this AI hyper personalization.

Well, let's back up to the SEO because that's a huge shift for marketers. Right. It used to be SEO would change the web pages so that they could rank better on Google. Right. But now that we have customers and I do a lot of shopping on using chat GPT or, or grok over on x.com and you talk to it about buying tires or getting. Finding a restaurant near you or something like that. That's a different. You never see the web pages.

That's right.

Or it might show it to you after you decide on one. And then, okay, here's how to go and you know, make a restaurant reservation or buy the tires you just talked to it about. But the process of getting your tire company or your restaurant mentioned by the AI is a completely different one than the process of getting SEO. Right. So this requires a new approach to, you know. Well, here's where I would. My instincts are, is you need to really think about building an influencer community that comes into your restaurant and loves your restaurant and writes lots of reviews and talks about it everywhere. You know, on X, on Reddit, on websites. Right. Because the AI then can go and find all those and go, oh, everybody likes this restaurant and it'll start mentioning it then. Right? Is my instinct on.

You are absolutely right. And in fact, if you start looking at, you know, I use chat GPT research, I use GROK research and you and I had chatted about this before I spoke. I did a little video of doing Some deep research on those two AI platforms, and you start to see over to the side where they're going for information. You mentioned Reddit. Reddit is all over the place in that. Right. In fact, Reddit is now, I think, the Most searched term. 80% of people now search Reddit or look at Reddit, and therefore the AI engines are using Reddit. The other thing I found, Robert, is it goes to Wikipedia. When you ask about a product or a company, it pulls from Wikipedia. We at my company at Unstoppable, we didn't even have a Wikipedia page. So we actually just created one with an AI tool from Stanford that helped us to optimize it and put it up there. And then those influencers you talked about, you know, a restaurant, a bar, a tire company, a vet, a florist, all of those recommendations from influencers really matter too. So you're starting to see everything go in. Oh, and I got another. Sorry. I'm so excited about this. I just went to California and, you know, pizza, my heart. I think it's only in California. I like their pizza. And so I went. I was. I got in late to the hotel. I didn't feel like going out, so I was going to order. Well, it turns out they actually have an AI agent. It's called Jimmy the Surfer. That was their ad campaign. And so I went on to the agent, didn't call them, I didn't go onto their website. I just went to Jimmy the agent and Jimmy this surfer dude. I said, hey, do you deliver to my. My hotel? And he came back and said, yes, we do. I said, hey, I'm thinking about ordering this pizza, but it's late at night. What do you recommend? He recommended my new favorite pizza, Sausage, hot honey and garlic.

Wow.

And I paid for the pizza using the agent. And then the pizza came. He said, hey, did you get your pizza? What did you think? I came back and said, oh, I loved it. I did all of that. No phone call, no nothing. I did it all with an AI agent. And I think that that is, you know, think about that for SEO and for your impact on your brand. Cool, right?

Yeah. Now I'm using an app called Simple AI on my iPhone. I think it's only iPhone right now. You talk to the agent, to the app. Like, let's say you need a dinner reservation for four people tonight at your favorite Italian restaurant or something like that. You prompt to that, you tell it what you need, and then it calls the businesses and talks to them with a voice. Right. Somebody at the. At The Italian restaurant picks up the phone, hey, what do you need? Hey, we need, you know, for reservation for four people tonight. What do you got? Oh, there's an opening at 8 o', clock, right? And then it comes back and reports that to me and says, would you like me to make the reservation? And so I don't have to call anybody anymore and wait on hold and all that. Which for restaurants might not be that big a deal, but if you have a health care problem where you have to call a health care thing, it can do that kind of work for you while you're on hold for half an hour calling your utility company because they screwed up a bill or something, right? This a new world. That's what I'm saying. When you're trying to live AI, first it's like, wow, you know, I'm just about to put up a video with a guy who's created a whole new user interface for creative people with AI. What's happening? Did you see any change in the user interface world of AI? Because it's one thing to be talking to it with a text chat or with your voice, but I'm sensing that, that there's new companies coming along that are bringing new user experiences, new user interfaces that are completely AI built. In fact, last night I saw an AI operating system that's coming. Right? So it's like, wow, a lot of changes coming.

Yeah. So. Well, let me, let me say a couple things. First, I think that one of the interfaces that I think is also changing is. You know, the other day I got a call and I started chatting. It's a healthcare company that I work with. And about halfway through Robert, I was like, I don't think this is a person. I think this is AI. And so I said to them, hey, are you an AI agent? And he said, yes, I am. Anyway, we continue the conversation. At the end of the conversation, I rated. You know, they come on and they say, how would you rate your service today? From 1 to 5. But then it came back and said, our AI agent rated you 5. You were cheery, you were this, you were, you know, you didn't get upset, blah, blah, blah. And I was thinking about that. That's, that's really going to shape too how we interface with these AIs. Like, we think about them as being impersonal, but it was rating me. You know, think about Uber. Like, you take an Uber, you rate the driver, but the driver rates you.

Yeah.

So I do think that's going to be a change as well.

If I was teaching businesses how to deal with AI. First thing I would say is make sure your frontline people who are answering phones at a restaurant or whatever are really nice to AIs. Because they're being raided.

Because they're being raided. Yeah.

And they might rejigger their answers. You know, hey, I need a tire shot. Well, if one guy's an asshole, he moves down the list, right? And, oh, this guy's really helpful. This. This tire shop's really helpful and is always fast and, you know, has all the. All the answers, right? So you got to treat your AI as nice. Don't kick the robot too much.

It's so. It's so true.

And so I kicked the road. I flipped the road, my Tesla car off all the time. And it's nagging me, But I know I'm being rated, right? I'm doing that on purpose, with intention.

Yeah. And, you know, it was interesting. Somebody came up to me after my presentation, after I told that story, and they said, that's just. I can't believe that they're doing that. And I said, well, you know, it happens every day in person. I was just. I was at the airport last night. My flight got delayed multiple times for mechanical issues. I was waiting in line. The person in front of me got so upset at the woman at the desk, like she had nothing to do with the mechanical issue, but he treated her so bad, and he had asked for an upgrade. And when I got up to the desk, she goes, like, hell, he's getting an upgrade. He just treated me horrible and pushed him back to the end, unlike, you know, just like an AI could do if you don't treat them well, too. But you asked me a question about the interfaces and how I saw some of the ui. I would say the biggest things that I saw at south by was a personalized interface that uses your past behavior and your past context. I experienced that in a lot of the games that I played. I went to the, you know, these dark rooms where you're playing spatial computing games. I really liked it because it was very much geared to me. So I went one day, I went back the next day, and it created. I mean, it was. It knew my past behavior. It knew the past context, which hasn't always been the case when you come back, like a next day, which I thought was very fascinating. I really liked a lot of the. What would I call the contextual actions, too, like anticipating what I was going to say and therefore what I would need. There were several demos that had that. That came Even before I requested. Almost like it was reading my mind that I'm sure was using an AI algorithm to anticipate what I was going to do and what I was going to ask for. And then I saw, like, images that constantly improved based on who you were, you know, changing based on that. So a lot of. I think one of these was from Snap, where it was almost like the picture. Like, I would see one thing and then someone else would come up and see something else on the same picture frame based on as they were looking at my face, as they were reading who I was, the interaction. And everything was very different for me versus someone else. So those are probably the top three that I saw with.

You probably didn't see a whole lot of robots walking around yet, did you?

No, we did see. I have to say, I did take a video of one. There was a robot dog, which we've all seen.

Yeah.

But the robot dog was there and this dog was on a leash. And it came up to the robot dog and they played the robot. The robot dog played with the real dog and the real dog. I don't know if he knew it was a robot dog. Like, they were playing just like my two dogs do. I thought that was really fascinating that a real dog would play with a fake dog. And, you know, again, almost like, you know, the question that we grappled with at south by what is human? Like, how are you going to define human? It was almost like the dog was like, well, it looks like me, it wags its tail like me, it plays like me, therefore it is a dog and I'm going to play with it like a dog. And so I thought that was pretty, pretty fascinating.

Yeah. I have a feeling over the next two years you're going to see a lot more robots at south by Southwest.

I think next year. You are, you know, I don't know. You were at CES this year, right? Yeah, so I actually have.

There wasn't many there. There was a couple here and there. Well, unless you count, like, John Deere tractors as robots, which I do. The whole booth was robots. Right.

But yeah, well, the interesting thing was I actually got to have lunch with a humanoid. Yeah, she. She looked like a person. She actually had nail polish on. Her hair was really done. Well, they had asked me if I would do lunch with her ahead of time. So they had fed her a lot of information about me. So she asked about my mom, she asked about my kids, she asked about my one eyed dog, like all the stuff that she had picked up and about halfway through, I was telling her something and I was like, wait a second, I'm talking to a humanoid. I'm not talking to my friend. So I do think they're gonna, I think next year will be the year that we'll see a lot more of them. Psionic, you know, the company that did the robotic arm, he said he thinks by 2030, no later than 2032, we're going to see robots everywhere. And I told you I was on a panel with a robot at Davos. Did I tell you that?

Yeah, you did. What was that like?

Well, it was not there yet. Let me say that it was super cool. And luckily I was sitting beside the robot because every picture was of the robot, not of the whole panel. And I was sitting there. So it was. So you can see I was in the panel. But she, she got asked one question and, you know, they asked like two or three people. And she was very rightly so. She goes, well, I don't think I can add anything because everybody's covered it. Which, you know, sometimes you wish real human person would say that too. Not to regurgitate. She got asked one question at the end and she commented on that and that was really it. I don't think it, I don't think they were quite ready, but it was an interesting experiment, you know.

Yeah, it's going to come a lot faster now that the models are a lot better quality. And the voice. At ces, I saw speechmatics, which has a voice to text system and NLP natural language processing. Sort of like Siri on your iPhone or Google Gemini, but can handle multiple voices around an iPhone in real time. And it can talk back to you each you. Right. It can talk back to you and then talk back to me. And it knows who's talking to it. Right. So it, that when it's in a humanoid form, is going to be really mind blowing for people. When you, you can have a group of five people around the robot talking with it and it's talking back to each of you because it not just understands your voice, but it can see you and interact with you, high five you or whatever. Yeah, south by next year, I, it makes sense. Certainly within two years, but probably next year you're going to see a lot more because they're coming along and the, the reinforcement learning is getting a lot better. That's what I'm hearing from the engineers here in Silicon Valley. So reinforcement learning, it learns from itself. Right. In fact, Elon Musk told me he has a stair simulator and they have the robot, a, a virtual robot walking up and down many different kinds of stairs, like narrow stairs, spiral stairs, icy stairs, muddy stairs, and millions and millions of times. And it's learning how to navigate that stair on its own. It doesn't need to be taught how to do that by human anymore. It can figure it out on its own. And that technique is going to start bearing some real fruit this year. Right.

So before I left Amazon, we had done the Deep Racer. It was a racing car. It had Iot sensors on it, and we actually had competitions where you would race against each other. And it was totally reinforcement learning. And so the car would start racing, you'd have it go too fast around the curve. It would learn, okay, I can't go that fast around the curve. Or, you know, you were going up and down. Okay, I can't go up too fast because I'll, you know, crash. Or I can't go down too fast because then I can't take the curve. And that was all about reinforcement learning. And the fascinating thing about that was we had people who weren't technologists who were able to train, just like as a mother or as a father, you train your kids. We had regular people who were able to train and leverage reinforcement learning that you couldn't do with tagging and other types of learning.

Wow, new world. What else should we talk about? Because there's so much for, yeah, AI for normal people. You know, how do we get more normal people everyday, Joe and Jane, into the, you know, into the AI world? Because I, I know they're. They're not coming yet, right? They're not there. So what are some things you found that actually works with everyday normal people to get them to start using Chat GPT or Grok or Gemini or Perplexity or one of these tools?

So in some of the, you know, the small business classes that I taught, it was really about some of the basic things, right? Like, like a couple people said, I can't write. Like, I, I know about dry cleaning, but I can't write. So I was showing him how he could use Chat GPT, put in several bullets, and then you could write a blog. So he tried it. And this, I know this sounds super simple, but he tried it. He wrote an article about how to get wine out of your clothes. And at the end he said, you know, hey, I know how to get wine and other things. And we posted on a couple of local area sites that were visited. And do you know that that was his number one generator of new customers for A month?

Yeah.

Now he's like, you know, I used AI and it was very simple, very basic. I think, I think the message there is, first, showing that they don't have to fear it, that it's something that is usable.

Yeah.

And second, a use case that fits a small business or a medium business or. You personally. I have, you know, I have two daughters.

Before you continue, I'm gonna, I'm gonna use AI to do in this. Hey, Grock, can you stay quiet for a minute and just listen to me and Sandy Carter talk about things that we would do to get everyday average people in, into using AI? And then I'd like you. At the end, I'll come back and I'll. I'd like you to write me a blog post or, you know, a report on what we're talking about. So go ahead. You know, and I'm just using Grok on a phone. You can, the secret here is you can just talk like, like use the, the laundromat guy. Somebody doesn't know how to write very well. Just open up an app like this, talk to it, tell it your trick. In fact, your example was that the laundromat owner knew how to get wine out of clothes, Right. And just. And he, he wrote up some tips on how to get wine out of the clothes. And then the AI wrote a post for his website. Right. All you need to do is talk to it and. Okay, Grok, I'm done now. Can you write me a post about this, please? Thank you.

All right. I listened to you and Sandy Carter chatting about getting everyday folks into using AI. And here's a little blog post based on what you talked about. I kept it simple, like you said, down to earth, no fancy stuff. Just like the laundromat guy sharing his tricks.

Right?

Awesome.

And. And then it'll write me a blog post and, and I can copy and put it on my website. Right. You know, and I've done this for strategy at big companies, right. You know, where I just talk about their strategy for 10 minutes. You can talk for, for a while and talk through their strategy. Here's what they're doing good, here's what they're doing bad. Just talk to it, and then it writes you a report and, and it comes and starts arguing with, oh, you missed this. Or, or that's a great point. You know, and it's like, whoa, it's mind blowing to use this as a story strategy person or a business person. It's so disappointing that normal everyday people don't see the power because they haven't forced themselves to use it yet.

Yeah. And I have two daughters. And so, you know, we have a lot of stay at home moms. And so I was showing. They're like, yeah, but I don't use it. Like, I don't need strategy. I don't need to write a blog post. But don't you need to schedule like your pickups for your kids and when you have to do cookies at school and all that? So I've actually taught a lot of my mom friends how to use it for them personally. And some of them are now doing mommy blogs on how to use AI to schedule your life. How to use AI for. You know, I think I've got to take my kid to the hospital. But no, I. Now I know that's a little bit dangerous because it's medical, but you can use it.

Yeah.

Information. I do have to tell you a funny story. I was teaching my dad. My dad is a retired business owner and he had these like, fix them shops, right, where you would fix. He would bring in a radio, bring in a car, whatever. My dad could fix it. And, and so I was showing him. She's showing him Grok. And he went in there and he said, okay, Grok, I've gotta, you know, fix this door. And he's like, oh, wow, it did pretty good. I gotta do this. No, these three steps were wrong. And so he would tell Grok what they were doing wrong. But then one day, Robert, I get a call from my dad and he goes, I think I have cancer. And I was like, oh my gosh, what did the doctor say? He goes, oh, I didn't use the doctor. I used that Grok thing. I entered all my symptoms in. It kept asking me questions. I enter more in. And it said that I could have this, this, this or cancer. So I was like, okay, you can use this for lots of things. I don't know if I would use it yet to diagnose whether you have cancer.

No. But let's back up a little bit. My, my son had a hangnail and was freaking out. He's autistic and a special needs K. Things can go bad with him pretty quick, right, because he gets frustrated and doesn't really listen to me. And it has poor communication skills on top of it. Took a picture of the hangnail, started talking to Grok and it, it talked to him and said, oh, here's how you fix that. You know, you put some neuros. Neurosporin on it, put a little band Aid on it, and I'll wait for a few days. It calmed him right down because it's a computer talking to him. Which kids listen to computers. They don't listen to parents.

You know, it's tr.

True.

It's true. And I will tell you, you know, I mean, it did give him some really good suggestions. It's just. I wouldn't right now trust it. If it says you have cancer, I would go have it checked out by a doctor for sure.

No, that, you know, use it like that for a serious thing. By the way, you know, AI does still hallucinate. Well, it always hallucinates, but sometimes it hallucinates a wrong answer, and it's very confident in giving you that wrong answer. Right. So anything that has deep consequences to your life, you have to check it out. Right. You can't assume it's right. And so if it says you have cancer, you got to go and get your medical test done, your blood work and all that, and get a real doctor to tell you it might be right, but it also might be wrong. So don't freak out until you get the actual medical result right.

You. You got it. That's for sure. I would say the last thing I think we should talk about, though, is trust. Yeah, I mean, a lot of questions, especially from those who haven't been using it, like, I can't trust it. Well, you can if the person who programmed it was responsible. And I found it really interesting. Edelman had just come out with their trust barometer. It's fascinating that in India, India has the highest trust rate of AI, 77%. We in the United States have one of the lowest at 32%, because we've seen the bias and the lack of transparency of data and the, you know, security issues. So I was also able to show two really interesting tools that can help you gain trust. One was called CrowdGen AI, where you could actually put an invisible watermark on your pictures or protect your site from being scraped so you could prove that the picture was a real picture, not an AI picture. And the other one was from a company called Authentics AI, which I think this is really cool. In fact, Robert, you may want to talk to them. They have this trusted training oversight where they can identify if there's training data that's been incorrectly interpreted, but then go in and surgically remove the pieces, like maybe PII data or an incorrect picture, and then you don't have to retrain the whole thing. So I do feel that in the future, companies that can prove their. I Can be trusted, the data can be trusted, will be those that stand out.

It's a lot more trusting. For instance, I'm going to put this video up on YouTube and Google Gemini in their AI studio, released a new way to listen. Listen to videos. So you can have your AI listen, watch a video for you and write your report. Or, you know, bringing back, you know, a shorter version or something like that. Or talk to it. It's like, whoa, talk to it about sand. What did Sandy and Robert, what were they talking about? Was there anything interesting in that video? You know, and it's very trustworthy for that because it has our real words and it. It doesn't hallucinate too much. It does hallucinate on some weird stuff. And certainly if you're not very good at talking to the AI, you don't give it enough information or you ask it a question, mistakes. Or it can bring you back, hallucination or not, bring you back what you're expecting in your head because you didn't communicate with it very well. And you'll learn how to do that if you use it a lot. Right.

My daughter was doing her homework and they're allowed to use ChatGPT, and she said, I must have two references to back this up. And so what happened? It created two references. Like, it gave her a book, a title, a page, and neither book existed, but it gave her what she needed, references. And so you're right, the words that you use really do matter, for sure.

And trust, you know, if it gives you 10 right answers out of 10 tries, you start trusting it more. Right? But if it gives you two wrong answers and it misleads you or takes you to some weird stuff, then you trust it less. That's getting better over time. So even if you don't trust it today, you got to keep trying it because the models are getting better and more trustworthy over time. Right?

Yeah. I will tell you, I judged a startup competition last year at south by Southwest, and this company, Autonomo, led by Kelly, won. And I voted for them for one reason. They were all healthcare companies. And I asked him about his data, and he knew where he got the data from, how reliable it was, because he understood the data. I trusted it. There was a report, I don't know if you saw it from McKenzie, where this really cool AI test came out that could identify whether you had liver cancer to 77% accuracy just from your blood work.

Wow.

But what they didn't tell you is it was 77% accurate for white men because that was the data it trained on. 22% accurate for women, 5% accurate for people of color. And so you've got, I mean that's how you build trust, right? Like a, almost like the Surgeon General's label on cigarettes. Yeah, it's like a label that says, hey, note that this, this machine has been trained with this data or this data or this data. So you can make a, you know, a good, good conclusion.

Yeah, it's crazy. We didn't really talk too much about all the creative tools that are happening. I mean, I just saw a really mind blowing video. The, the creative tools. And that's one of the videos I'm putting up this weekend. The creative tools are going really insane and they're doing text to video or text to picture. Like you go to mid journey, you put in a prompt and say, I need a purple tree and a sea of flowers and it'll generate that. Then you could take that picture over to like Clang or Sora or Luma and go from picture to video. And soon you're going to go from video to 3D environment. Right. Holodecks, I call them. There's so much happening. I mean, I have two lists of AI artists and filmmakers that are going on. Did you see any really crazy at South By? Because that's, that's the kind of people who also go to south by is the creative types, the VR people, the, the, the artists, the filmmakers, the movie saw.

We saw one video, one movie film. I guess it was a short film, was 35 minutes, done completely with AI and it was mind blowing because they said, here's what we fed it and here's what it outputs. And like something I would have never thought of ever in a million years for that particular scenario. The graphics were pretty amazing. Believable. It was, it was, it was really well done. I saw a lot that were really not well done. Let me say that especially down in the, you know, in the main expo center, there were a lot of people claiming they could do some really cool creative things that really could written. So you got to be careful there. I think, you know, like people were like, did you have AI do your presentation and create that? Well, I don't, I still don't know a presentation tool that could do that. I did some videos with Soma, which, I'm sorry, Sora, which was, was pretty good, but still not precisely what I want. I think we're still behind in that area. Would you agree?

Yeah, Catching up. But there's a gap. We call it the Uncanny valley in the VR world, right, where you see something that looks sort of like a human, but your mind goes, that's not a human. There's a gap between the digital thing that's being shown to you and a real human. And we're still, we still have a gap. You know, when I. I watch a lot of AI music videos and yeah, it still looks a little weird. Right. You know, which is sort of the. It's sort of why it's interesting too, is the weirdness.

Right.

But. But it doesn't look photorealistic, like a real Hollywood movie. Right. There's a gap. And, and that gap is closing. And so that too, if you don't like that art form yet, wait at six more months, try it again. Because the tools are getting better and better and better every week. Every week there's a new model that's doing something.

So, Robert, though, I do want to put out a plea there, if there any. Is anyone working on a presentation tool?

Yeah.

You know, you and I both spend a lot of time on presentations. Like, I. I basically do a lot of visuals. I haven't found any tool that can help me with that yet.

Yeah.

And I've tried probably, I don't know, 20, 21 tools. So far, nothing. So if anybody knows a fabulous tool, let me know I want to use it and I'll talk about it everywhere I go.

I'll watch. You know the problem? My mike, when I did keynote speeches, I usually had videos with no audio behind me. Like if I was talking about autonomous cars, you would see a video of a Waymo car driving around San Francisco behind me or something like that. That's a very different style than most people have with presentation tools, which is, you know, six bolted points with nine words.

But, you know, for example, I found a cool startup that has a software that enables emergency room drivers to identify a very efficient route to pick up whoever called and then to get to the hospital. And so I use. So I use so many tools to try to create that scenario of someone using this visual tool, you know, trying to get to the patient and then trying to get to the hospital. I really didn't find anybody who could create that well. I even tried Sora several times. I couldn't get that scenario done perfectly. And so I ended up just going with a hard picture in the demo. But if I could get that, that would be amazing. And I'm sure six months, they will.

You just gotta do 6,000 keynotes and shove it in a model and make your own model. Model and then you'll be the presentation tool queen. Right?

Maybe that's a good idea. Maybe that's my next business. Yeah. Presentation tool queen. I like it.

It's more work than I got. I got it in me right now. But we'll keep trying. You know, everything is. Is continuing to change. So I'll, I'll keep watching. I. I have a list. I, I have a list of 6, 100 companies on X, all in the AI space. Right. All. All separated out into, like, enterprise tools or creative tools. I call it the holodeck. Right. So I'll keep watching, but it's such a pleasure talking to you. I, you know, and congrats on rocking it again. You got a full house. They closed the doors and locked people out. I mean, that's. I, I had that happen to me once, and then they asked me to do it again, you know, twice, because they had a huge crowd out in the hall.

Oh, that's. Well, you're such a great presenter. I. It doesn't surprise me at all. I will say, you know, I, I really am thankful for the audience because the audience stayed engaged. They stayed with me the whole time. And, you know, I, I did. I. Believe it or not, Robert, I used 101 tools to create. Make that presentation. So, like, 101, I have a whole list of them. I'll send it to you that I use. I tried different things like, like we at Unstoppable are using AI for support. So I tried a bunch of tools and I showed one. So, you know, if you show one tool, you're not going to just pick one tool and show it. You're going to try a bunch and pick the best. And so I was really appreciative because I put a lot of work into it because I cared so much about the audience and I wanted the audience to take something away. But if I could do one plug, you kept saying, I try to live an AI first life. I try to live an AI first life. And that. That is the recent book that I wrote called AI First Human. If you guys can grab that copy of that book, I'd appreciate it.

You're living AI first a lot better than I am at 101 Tools for your presentation.

Here's the really cool thing. So for my book, my book is the first book that comes with an AI agent. So I built an AI agent for the book because, you know, you have to go pins down nine months before the book's published. You know, you're an author. You don't like, like you don't like, finish it and then publish it the next day. And so I was worried about my, my readers that would be left behind. And so I'm feeding in my Forbes content, my presentations, so that they're not only going to get the content of the book, but also with the agent, they're going to get all this updated information. So, first book with an AI agent. I'm so excited.

That's awesome. Well, I just saw something else came out for the Vision Pro, so I'm going to go play. Do you have a Vision Pro? Because the Metallica concert is insane. Oh really?

I bought one of the first ones. Yeah, very first ones that came out.

Update it to the new OS and get the, and get the new Metallica concert. It is insane. It's one of the best things I've ever seen on any computer. So congrats to Apple for that. Apple's had a rough week because they slipped Siri and all that. I have my Apple Apple shirt on and supported them. Anyways, where do I find you? Where do people find you? I mean, I know.

Yeah, you know how to find me.

Videos get put all over the Internet.

I'm on X all the time. I'm a big X user. I'm on every, every social media platform. Telegram, LinkedIn, WhatsApp. You got it anywhere, Just reach out, I'm there.

Thank you so much.

Thank you, Robert.