209 | Altman shares everything, including what’s keeping him up at night, AI sparks an efficiency and entrepreneurship revolution, GPT-5 poised to drop any moment, and more essential AI news for the week ending July 25, 2025

0:00Hello and welcome to a weekend news episode of the Leveraging AI podcast, a podcast that shares practical ethical ways to leverage AI to improve efficiency, grow your business and advance your career. This is Isar Maitis, your host, and we have a very different episode for you today. We're going to talk less about technology and more about the impact of AI on the world. There are a lot of interesting things that happened this week, including a very long interview of Sam Altman, which is always interesting to listen to, as well as a paper by OpenAI calling Unlocking Economic Opportunity, a first look at ChatGPT Power Productivity, which is obviously very interesting to look at.

0:37some latest rumors about the release of GPT-5 and other topics that we need to cover. And we're going to focus on how AI is going to impact the world in the next months and years. This episode is brought to you by the AI Business Transformation course. It's a course that I've been teaching personally since April of 2023, at least once a month, in many months, twice. And I teach mostly private courses to companies and organizations. And we open a public course once a quarter. the next quarterly course that you can sign up for starts on August 11th, which is about two weeks from the time this episode comes out.

1:12So if you want to completely transform your career and or your business and have a very solid baseline on how to implement AI from a practical perspective in a business environment, this is the course that you need to join. And as a listener of this podcast, you can get$100 off the price of the course. So you can use the promo code leveraging AI 100 to get$100 off. That's going to get it down to$900, which is probably the best investment you can put in your career and in your business right now. It's completely negligible when you think about the future of your career and the future of your company.

1:52And within four weeks, eight hours, of course, plus open office hours where you can join me and ask me anything, you can completely transform your understanding of AI and have a hands-on understanding on how to actually implement it successfully. So come join us on August 11th, and now let's dive into this week's.

2:14Every time one of the leaders of the AI industry is interviewed on any stage or in any specific forum, it is fascinating because it gives us a glimpse to what they see, how they think, what they believe will happen. And this week, it happened to me, Sam Altman, he was a guest of the Federal Reserve event in Washington, D.C., talking to a lot of people from the finance world, from government. And it was a very long interview that dove into a lot of topics. And I want to cover at least the main ones, because I think it's fascinating and I think it's very important to know what Sam thinks. Sam started by talking about the rapid progress and the productivity gains that are being gained by the AI and its progress.

2:56So a few examples that he gave were that scientists are reporting two to three times increased productivity and that computer programmers are seeing 10x productivity boost compared to what they were able to do before using AI. Sam even shared a personal experience where he was planning to do something over the weekend, something that without AI would have taken him days. He was hoping to be able to complete it with ChatUpt in hours and he was able to finish it in minutes. More importantly, he said that those minutes that he was using AI to complete the task was probably less than a dollar worth of compute tokens, which led him to the concept of intelligence too cheap to meter, which is his twist on electricity, too cheap to meter, which is a phrase from earlier days where electricity became more common.

3:40And the idea is basically that we already have and definitely will have intelligence that down to the unit level will be more or less zero, which will allow us to do incredible things because we'll have vast amounts of intelligence at a very, very cheap price. He was talking about the fact that the cost of intelligence have went down in a factor of 10 every single year in the past five years. And he expect that to continue in the next coming years. And he summarized it with the following quote. I don't ever remember something that you know would have cost 10 ,000 of knowledge work a year ago, cost a dollar or 10 cents or whatever it would cost now.

4:21And he was referring to his own personal experiment saying he would have hired a professional code writer to do the thing he wanted to do. And now it cost him a few dollars or maybe a dollar to do the same amount of work just using AI to do it. This is obviously profound. And as he's saying, that is going to continue to happen. So if it's a dollar today, it's 10 cents next year, it's a cent the year after that, and so on and so forth. And that without any additional crazy breakthrough is just following the current trend. Sam was also asked, how much is this like previous revolutions and what is different?

4:51And he actually gave a great analogy to the invention of transistors, computer transistors. And he said when it was invented, it was a great revolution and people were talking about this. But then it got embedded into more or less everything we know. computers, cameras, TV screens, cell phones, microphones, basically anything electronic that we are using has transistors in it. But nobody thinks about the transistors. People just think about the thing that they're using. He assumes the same exact thing will come with AI technology. He's saying nobody will talk or consider the AI companies for themselves because we will expect to have AI and intelligence more or less in everything that we're doing.

5:29He also compare the transistors scaling laws to the AI scaling zone. With Moore's law for transistors, I said he doesn't have a name in AI, but it's the same kind of concept when we can continuously scale their abilities over the years. One of the most interesting points he was asked about is obviously what's going to be the impact of AI on jobs. And he said that he's going to, and I'm quoting, there are cases where entire classes of jobs will go away. There are entirely new classes of jobs that will come. But then he said that the most likely dominant trend would be, and I'm quoting again, the tools people have to use in their jobs will let them do more, achieve things in new ways.

6:08So he gave several examples on how AI will impact different industries. He was talking about computer developers, and he basically said that the world needs significantly more software than we can generate right now. So if saying if a computer developer can generate 10 times more code right now than they could a few years ago. Well, that means we can generate 10 times more software, but the world may need a thousand times more software than it has today. So there's still more demand than the AI can supply on its own. And then he gave the opposite example of where it's not going to work at all, which is customer service, which we talked about many times on this podcast.

6:42And he mentioned that historically customer service was a real pain. You would call a company, you will talk to four or five different people about the same thing. And in many cases, that wouldn't even solve the problem. And then you would have to call them again and literally spend hours and a lot of frustration trying to solve something. Anybody who called any large company like AT &T and so on knows how horrible that is. And now AI agents, you can call them. They have a great conversation with you. They speak perfect English or any language you want in any hour of the day. And if they're connected properly to all the company systems, they will most likely solve your problem much faster than ever before.

7:15So this was his example for an industry that will probably be completely eliminated when it comes to human jobs. And I agree, I don't think we will have the concept of contact centers, or if it will exist five years from today, it will be on a much, much, much smaller scale than it is today. And he also gave an example of doctors. And he said, ChatGPT today, not even a medical specific AI, but ChatGPT today can probably give you a better diagnostics than most doctors in the world today, at least on day-to-day regular stuff and probably beyond that as well. And yet everybody would want to go to the doctor.

7:50the doctor will probably use AI to give you a better diagnostics, but the chances that you won't go to the doctor and just have AI do the diagnostics for you may come in the future, but it will probably take a while. And he's saying that right now, if you go to the doctor, you usually wait, at least in the US, 30 to 45 minutes in the doctor's room before the doctor actually gets to you. That's a huge waste of time, which means we don't have enough doctors to deal with what we have. And having the doctors use AI in order to dramatically expand their capability to help additional patients and more accurately is for the benefit of everyone, which is absolutely awesome, right?

8:25So if you can go to the doctor and the doctor will see you in five minutes rather than in 45 minutes, everybody wins. You get better, faster service. You get healthier, faster. More people can get access to healthcare because it's going to be cheaper down to the unit level. The doctors themselves can serve more people, make more money. Literally, everybody wins. So with these different examples, you can kind of see where it's going to be an augmentation of current human action versus complete change and dramatic removal of specific jobs and the creation of new jobs, which are unclear yet, which is, again, one of my biggest problems that you know that I keep on mentioning.

9:01And then there's going to be the new jobs created, which he didn't talk about at all, which is one of my biggest problems. So I'm going to talk about two issues that I have with the way Sam addresses the issue and probably all the other ones as well. One is that they're always mentioning that it's going to create new jobs, but they never mention what these jobs are going to be and how many of them are going to get created versus the ones that are going to get eliminated. So that's problem number one. Problem number two is that they're referring back to historical revolutions that replaced human labor with human thinking.

9:33That's been every previous revolution that we had. And with every one of these revolutions, we became better. And every one of these revolutions was better for society because it allowed us to be more productive. It allowed the economy to grow and overall made the world a better place. The reality is this revolution is different. It is taking over human thinking. It's taking over intelligence. It's taking over the thing that we moved to when labor, when manual labor was taken away. So what will we do next? Is it more critical thinking? Probably in the beginning, that is the right solution. But over time, potentially a very short amount of time, AI will be able to do better critical thinking than we can.

10:16And then what? I don't think they have an answer. I haven't heard anybody ask them that tough question. And I really fear that this thing that everybody expects, that it will generate more jobs because previous revolutions did, is not going to prove correct this time, at least not at the same scale. at the amount of jobs that it's going to take away. Back to Sam in the interview, the next topic that he touched upon is risks and safeguards in AI adoption. He talked about that in the beginning, there were a lot of hallucinations and that that generated a lot of fear. He's actually claiming that the fear was significantly higher than the actual volume of hallucinations.

10:53He's saying that the actual hallucinations, even in early models, were around 0.1%, while a lot of people that were early adopters thought that 50 % of the information is hallucinated. Either way, I don't think it really matters. I think what matters is that it does hallucinate. And in some jobs and in some cases, that's just not acceptable. But since the entire audience was in the financial sector, he gave examples from the finance world. And he said that they had serious concern in the beginning that financials and governments will be very slow to adopt. And yet he's saying that it actually happened the other way around, where very early on, groups like Morgan Stanley and the Bank of New York became their early biggest enterprise partners and that have implemented AI across multiple aspects of the business, mostly because of their fear that other companies will do it.

11:39If they don't, they will stay behind, especially with newcomers that can build an AI first bank or at least banking experience that will leave them behind. And therefore, they work together with open AI and on their own in order to create an environment that can allow them to leverage AI in a safe way, despite all the data controls and limitations that they have in regulations and so on. So the good news is if a industry that is as regulated as financial institutions with so much sensitive information can do it, probably every industry can do it. Now, he was also talking about specific risks like prompt injections, which is the ability to create a prompt that seems to be doing one thing when in reality it's doing the other thing.

12:19While the models have a lot of private information about an individual, this can be exploited to get access to that information. He sounded a very serious concern specifically about the financial world, where he's saying a lot of banks are still using voice recognition as a way to identify a specific individual and to allow you to move a lot of money from one place to the other, when in reality, that aspect could be completely easily hacked by AI. And now quoting a thing that terrifies me is apparently there are still some financial institutions that will accept voice print as authentication. And then he gave a few other examples.

12:52And then the following sentence was, AI has fully defeated most of the ways that people authenticate currently other than likes of passwords. What that tells us is we either develop new ways to authenticate and or sometime in the not too far future, we will have to go back to the days of going to the bank physically and talking to a person or maybe a robot in this particular case. And in order to do any transaction that we can, because that's going to be the only safe way to actually protect our money. Sam was also asked what keeps him up at night. The question was something about the lines that all the sci-fi movies tells us that AI is going to destroy the world.

13:28So my personal thoughts and expansion on what Sam said is that from a truth detection perspective, we're in a very bad spot, maybe the worst spot in history of mankind. I've recorded an episode called The Truth is Dead very early on in this podcast. It was probably May of 2023. And I think it's just obviously getting worse and worse. I think the days that we can trust what we see in the digital communication, even from people we know, even from our loved ones, has to be questioned at any given point. We will not be able to tell who we're speaking, who we're speaking to. Is it a real person? Is it a fake person?

14:03And the same will be true for news, for personal communication, and so on, which is a very scary thought because communication is what society is built on. You need to have trust in what your government tells you, what your newspaper tells you, what your municipality tells you, what your employer tells you and so on. And most of our communication today is digital communication. And if that goes away, this whole thing comes crumbling down. And so that is my personal biggest fear when it comes to short-term impact of AI. But there are bigger concerns. And Sam was asked about what keeps him up at night.

14:35Basically, the question came from the aspect of most sci-fi movies have shown us that AI will take over the world and destroy us. What keeps him up at night? And Sam shared three scary categories that keep him up at night. One of them is bad actors misuse. So basically a hostile entity that is developing super intelligence first, meaning before the U.S. And he's using it to create some catastrophic outcomes, such as developing bioweapons that we can counter or really sophisticated cyber attack that can take the entire U.S. infrastructure and things like that. So he's saying in this particular case, we just need to make sure we're there first so we can have the right AI to protect against these kind of scenarios.

15:11obviously the other side thinks exactly the same way, which means we're in an arms race of super intelligence AI right now, which has its own scary aspects because there's less concern about immediate risk and more concern of a big catastrophic risk in the future and being there first. The second scenario is the sci-fi scenario. The classic way of AI becomes autonomous, gets his own understanding, tries to protect himself or itself and fights against humans because we become a problem for it to continue to thrive and we can't turn it off, etc. He's saying that because of the actions that they and other companies are taking, this is a relatively low concern.

15:49I tend to disagree because of the previous point. They're running very, very fast in this race and they don't always take all the steps. And even if they do, not all the players do to make sure that they are taking all the steps in order to prevent this from happening. But again, that's my personal opinion versus Sam's opinion, his opinion probably counts a little more. But then the last one that he said, which is really concerning to him, and now after I heard him, really concerning to me, is more or less what he calls an accidental takeover, an emotional takeover, over-reliance on AI. Basically, he's saying AI will become so deeply ingrained in everything that we do and so superior to us in its ability to think and come up with ideas that will become dependent in our decision-making on the AI without exactly comprehending why it's deciding or suggesting what it is suggesting and will become so comfortably numb that AI will basically run our lives.

16:42He gave short-term example of young individuals that are becoming emotionally reliant on AI as problem number one by saying in the long run, these will be the leaders of countries that we rely on AI to help them make decisions. And then we get to the point that AI basically runs the planet versus humans running the planet, which is obviously a very scary thought. He did not give any advice on how he thinks we can prevent this from happening. So he was also asked about education and how he thinks AI will impact education. And he gave two analogies, which were great and I think very helpful. And I use them myself as well.

17:15One of them had to do with the calculator, basically the invention of the calculator and how math teachers, when it was invented, thought that's going to be the end of math teaching. Because if you can do math on a calculator, why do you need to learn math? The second was his personal experience when he was learning when Google came out. And again, teachers and universities and schools were banning the use of Google in the beginning because they said, if you're not going to walk to the university and spend an hour in looking for the books and reading through the information, you will not actually learn.

17:44And we know that did not happen. If anything, the opposite happened. We have a lot more access to a lot more information. And he's saying that AI is going to have a similar effect where we're just going to get more relevant information faster, which will allow us to learn more things and do more impactful things while investing less time in the tedious aspect of research. Again, I do not fully agree because of what I said earlier. The previous revolutions, let's say the calculator and Google, gave us access to information and we had to think and analyze and figure out what to do with that information.

18:15And if we're going to defer that to AI and rely on AI to do that aspect for us, then I think our ability to learn and think critically is going to be eroded potentially very quickly. On the flip side, he talked about the fact that how AI is empowering small businesses and entrepreneurs, and it's going to create a new era for entrepreneurs being able to basically have an employee in every area, something that most entrepreneurs, at least in the early stages, cannot afford. So think about a solopreneur that can now have a team of somebody to do contracts and customer service and marketing and advertising and enabling basically small businesses to get worldwide impact with very few people because AI agents can take every task in the business without hiring additional people.

19:00On one hand, that drives amazing innovation. On the other hand, it doesn't generate as much jobs as it does today. So today, small businesses are the largest employer in the US and probably in the world because they employ most of the people. But if most small businesses will be really, really small because they won't have to hire a lot of people, yes, it's going to help with innovation and driving new products and services, but it's not necessarily going to help in creating more employment unless we have a complete explosion in the number of businesses that gets created. And I think that has to do a lot with human nature and how some people are more open to taking this kind of risks and becoming their own boss and some are not, or maybe that will just become the norm and everybody will just have to do it.

19:38I don't know yet, but it definitely provides an interesting opportunity. My businesses are a very obvious example of that. And another very important topic that Sam was asked about and was mentioned, which I really, really like, is he was talking about how AI can level the playing field across the globe and how it can provide access to high quality services, education, financial advice, medical diagnostics, and so on, everywhere around the world that has internet access, which is now most of the planet. So in areas and regions of the world where these kinds of services are currently limited or non-existent, AI will provide them access to the top level of services as everybody else around the world, which can dramatically improve the situation of people in those areas.

20:23I really like that point. I already saw several different experiments of AI teaching in schools in Africa that is doing magic to the results of people from those regions, allowing them a lot more opportunities in life and so on and so forth. Add medical to that and consultation and other stuff. And I think there's definitely a lot to be gained beyond, obviously, the additional scientific research and discovery that will allow us to potentially solve the biggest problems of the planet. So what's the bottom line? The bottom line is Sam, as always, is very optimistic. He gave a lot of interesting viewpoints of where AI is going, but he also didn't shy away from the risks and challenges ahead.

20:59So I definitely recommend listening to the whole thing instead of just my summary. I'll put a link in the show notes for that and you can listen to the whole interview. As I mentioned in the intro, OpenAI released a paper this week called Unlocking Economic Opportunity, a first look at ChatGPT-powered productivity. And in this paper, they covered a few important topics. Topic number one is AI is a fundamental transformative technology. So the exact quote from there is, throughout history, certain technologies have fundamentally changed what people can achieve. More than just simple tools, they provide a new kind of infrastructure that drives progress.

21:32The wheel, the engine, electricity, the transistor. AI is such a technology. What they basically mean is that it's not just a tool, it's an infrastructure for growth and innovation. And I completely agree with that. Probably the craziest one we've ever seen, again, because it is replacing, or if you want, enhancing intelligence. Then they're talking about the equitable distribution of the benefits of AI. So on one hand, they're saying that AI will dramatically expand the economic pie, But then they're saying that the crucial questions are, and I'm quoting, how will that expansion unfold and who gets what slice?

22:08And I 100 % agree with that. That's going to be a very, very big question and a very big concern. That's going to happen a lot faster. That connects to our discussions in previous episodes on job losses versus job gains, how much money is currently invested into AI companies and how much revenue they will generate, how will that be distributed, whether UBI is possible or not possible or how the economy will even run if 30, 40, 50 % of the population is unemployed. Again, I don't think anybody has answers, but they're at least raising that as a concern that needs to be addressed. The wishful thinking outcome that they want that would happen is, and I'm quoting, everyone to be on the up elevator of AI.

22:45And then to do that, they're saying that we need to, and I'm quoting again, democratize access, support workers through changing times and build economic systems that reward broad contribution and participation. So again, this is a cool vision that they don't exactly explain how they will do. What it hints to is that every organization, whether a small company or governments, have to invest resources in providing the knowledge, the skills, the tools that the employees or people of that organization or society needs to have in order to be productive in this new universe with AI. I could not agree more.

23:20I think the most important thing right now is AI education. hence everything that I'm doing from this podcast that you're listening to, my YouTube channel, the Friday AI Hangouts that are open to the public. And you can join us every Friday at 1 p.m. Eastern time, where we talk about AI and its impact or down to tactical things on how to do things in specific businesses or specific use cases. And obviously the AI Business Transformation course and my workshops to companies, all of it is because I'm a huge believer that the more people know how to use AI, the more people understand the risks and the impacts that AI can have, the higher the chances that we're going to have on coming up on the right side and enjoying the benefits of AI while reducing the risks and the downside of this revolution.

24:01This paper also relates to the rapid unprecedented adoption rate. So a few key factors that they shared, some we knew from before. ChatGPT is the fastest adopted consumer technology in history. They reached 1 million users in five days, 100 million users in two months, and half a billion daily users today. Its adoption within the first two years were twice as fast as the internet's adoption. Part of it is because infrastructure is already there, right? So the infrastructure for the internet did not exist in many places in the first years of the internet, and now it does exist in order to distribute AI, but it doesn't matter.