We are headed toward a powerful revival of entrepreneurship. Yes, I think it's close, and it will truly unfold in the next year or two.
This is Greg, a founder of OpenAI, which created ChatGPT. Over a billion people use it every week.
I'm a decent programmer, but I don't write code anymore. I send everything over and it writes the code, but I give it a lot of feedback and prompts.
If you're building your own product, how do you strategize so it doesn't become obsolete when new AI models are released? Let me show you something I made recently. I am quite ambitious in my work with artificial intelligence.
Well, I think the answer to whether there's enough ambition will always be "No, because you can always do more."
If someone wants to level up in working with agents and only has an hour, where should they start this weekend?
I would say, first of all,
Let's talk about practice. I recently spoke with a Nobel laureate in economics, and we kept coming back to one theme: the models are already very powerful, but there are few real-world applications yet. What are you launching this year that could change that and show people what AI is truly capable of?
Well, first of all, it's very interesting to see how many people use ChatGPT. We have 1.2 billion weekly active users. And 300 million ask ChatGPT about health or solve personal issues with it every week. And yet, most of these scenarios are just the tip of the iceberg. We can see it in the metrics. If a person uses ChatGPT for three different tasks—meaning they realize ChatGPT can do three different things for them—they become a power user; they stay, keep using it, and want to use it more. But reaching those three scenarios is very difficult, and in a way, that makes sense. Here is an AI that can do everything, but what exactly can it do? Yes.
Yes, exactly. Where is the magic?
It's about finding that specific magic that hooks you personally. But at the same time, I think there is a huge opportunity here, a huge space, almost a void. You talk to the AI, and the AI knows what it is capable of. It could suggest: "Listen, you asked me to do this, but actually, I can do more. I have an idea. Let me make a whole presentation or a spreadsheet." That is exactly what we are exploring. But I think this is important. Right. People need simplicity from AI. They don't need an interface with buttons, sliders, or model selection. None of that. That's not the AI we were promised. We need someone to talk to, someone who solves problems, helps achieve goals, and even takes initiative. It can say: "Hey, you have an email in your inbox you haven't replied to for a couple of days. I looked into a few things. Here is the right answer to the question. I've drafted a response. Do you approve?" And by the way, the use case I just mentioned isn't just a theoretical hypothesis. This is my real-world example of working with it. Right before we started this podcast. So, I think we've reached a point where it really can be much more proactive, take that burden off you of having to figure out what the system is even capable of, and actually help you achieve what you want.
What else can we do with AI now that seemed impossible a year ago? What should people try?
Well, I think there are a lot of them. There is a whole spectrum of these models. I believe the best way to get familiar with them is to experiment with the system or see what it's capable of. And when it comes to programming, these models now write software better than almost anyone I know. I'm not bad at programming, but I don't write code anymore. I send everything to the AI, it writes it, and I provide a lot of feedback and prompts. So, it's always fun to create a small game and get instant gratification just by seeing how long it took to implement, and then ask it to make something more complex. I think we've reached the point where it's so good at using Blender, so good at creating 3D models, that it becomes interesting to ask it to create models from photos and the like. It uses programs like Microsoft Paint to sketch from a photo, but you can ask it to create more realistic things too. For example, a friend of mine wanted to build a guest house, and with its help, he made CAD drawings, took them to a contractor, and now the house is being built. Yeah.
This really expands the possibilities. Usually, you would have had to hire someone. I don't know how to do CAD drawings, but now we have intelligence in our pocket that can do that for us. I personally love trying AI in different areas of my life. I think voice mode is what makes AI very accessible. You just switch to voice mode and start talking to it. Now, voice mode can be connected to various tools, so you can ask it to do everything you would in a chat. If you've connected your email and calendar, you can ask: "What's my schedule like today?" Or, for example, "I want to free up time to think about how to do this." And it actually sees what's happening in every part of your calendar. So for me, it's a practice of stopping and thinking about something a bit more ambitious that I can ask of my AI than what I've asked before, and seeing if it works.
Yes, I actually want to show you one thing I recently made. I think anyone can build their own productivity app now, or even several. And here is what I did. I asked artificial intelligence to write an app that prepares me for a podcast. I specify who the podcast is with, how long it will last, and what my success metric is. Basically, YouTube views.
And then the app starts working, gathers information about the guest, and suggests topics for conversation. It turned out great, but how do I make it learn from my experience? And where can it be developed further? I will record our conversation. It will go in here, and the app will analyze it. Right now, it's checked my past interview here. What do you think is missing? And do I have enough ambition in working with artificial intelligence?
Well, I think the answer to the question of whether there's enough ambition will always be: no, because you can always do more. That's the beauty of it. AI. It's about solving problems, about helping you achieve what you want. There's no ceiling to what you can imagine, right? So it's all about realizing that once you've succeeded at one thing, it means you can dream a bit bigger. In my view, yes, this is a super cool app. First, in terms of background research, and when you were building it, you went through all the interviews I've ever given, didn't you?
Yes. It gathered all the background info, went through it. See, it gathered all the background info. The only thing is it doesn't connect to social media yet, but I think that can be solved with an extra plugin. And then here are all the past OpenAI press releases, and here is a tab with questions. I like it. We have a questions tab. It already has titles and covers here. And here are the questions I can just copy.
Yes, I would be interested in how much it helps with processing and the actual production of the final video once everything is recorded. After all, this thing has all the context. About me, about OpenAI, about what you covered in previous episodes. And then the question: how do you want to edit it? I've heard many cool examples of how AI helps with video editing, quite proactively. So I would really be interested in whether it's possible to make an app where you are only required to do one thing: basically look at the questions, make sure you like them, and if not, give feedback, and it will make it better. You show up, record the interview, and everything else is already taken care of. How much work goes into the process from the interview to the moment the video goes live on YouTube?
Oh, quite a lot. I have a large team. We've also digitized a lot of things. I've been doing YouTube for 12 years now, so we have all these files to draw from. That's why it's easier to structure the work with AI now, but each episode still takes at least 40 hours.
Wow! And if you could reclaim that time, so the mechanics ran themselves, and you could focus more on what the interview should be and everything else involved, or think about scaling and doing way more interviews—what would you do with that time?
That's exactly what I want to ask about small businesses that are already using artificial intelligence in their daily operations. What is the owner's role now? What should they be thinking about more, and what is actually holding businesses back right now?
I think the very foundation is changing. First off, we see how many small businesses are launching because people feel more confident when they have ChatGPT. To start a business, you need a ton of different expertise, which is hard, but now you have PhD-level intelligence or a world-class business consultant. All of that is in your pocket and it's free. If you have the desire and the means to pay for other AI models, you can get much deeper analysis and far more computing power to solve any task you want. So, starting a small business is easier than ever. And we see this in concrete examples. When you talk to small business owners, you can see how many tasks they are offloading. And sure, there are tasks they already knew they had to do, but there are a ton of things that nobody would have touched before. Like gaps in a business strategy or, for example, how to even create this specific document. And there are still small business owners who don't even realize what they could do with AI. One of our staff members said he was picking out a watch at an antique watch shop and started talking to the owner about AI. And the owner was like, "Oh, I don't need AI. I run a watch shop." And he said, "No, no, AI can help everyone. Let me show you." Then he described some of his problems and added, "Well, yeah, he doesn't actually even have a website." So, that's the simplest thing.
Exactly. And you just tell voice-mode ChatGPT: "Make me a website, make me a great website." And then the owner says: "Listen, every watch of mine has its own story." Right? And then everything becomes very simple. Tell the story of every watch. And you get a beautiful visualization of the entire product. And you realize that this is how technology comes to life. Usually, for many, it is almost inaccessible. Therefore, I think it's important to ask yourself: if you had the best programmer in the world on your staff, what would you have them do? Now that is truly possible.
You once said somewhere that small businesses actually need to work on their ambitions first. Yes.
Yes. Yes. And I think that, I think that's a very interesting thought.
