Sam Altman's Latest Interview: What Mindset and Judgment Should You Have in an Exponentially Changing World

The rapid iteration of tools requires a mindset that breaks free from old frameworks to adapt to an era of exponential change.

Have you ever thought about a startup that was only established two weeks ago, managing to completely redesign a suite of mainstream office software—documents, spreadsheets, presentations—all centered around AI? This isn't something I made up; it's what Sam Altman (co-founder and CEO of OpenAI) said in a recent interview after meeting such a company. Ten years ago, we could almost predict what a ten-week-old startup would look like. Now, if a ten-week-old startup resembles those from a decade ago, it indicates that it has already fallen behind.

This interview was packed with insights, discussing entrepreneurship, how OpenAI has evolved over the years, and how he personally handles pressure and makes choices. I’ve compiled some of the most striking segments, adding my own interpretations for you to read.

Most People Still Choose the Easy Battles

Altman mentioned that this moment in time is fascinating; costs are rapidly decreasing, and the time required to get things done is shrinking quickly, which is precisely when startups have the greatest advantage. This phenomenon is occurring across many fields, and theoretically, it should be the best era for entrepreneurship. However, he observed a rather contradictory phenomenon: most startups are still focused on the same task—creating AI agents for specific industries. This path is viable and can even be quite profitable, but it is unlikely to produce companies that will be truly remembered in this era.

What struck me was his subsequent statement: despite the tools having completely changed and the capabilities of models continuing to rise, people are still hesitant to bet on something that is currently unachievable but could be possible in two years. He said this temptation is particularly strong; it’s easy to use today’s agents to solve immediate problems, which is understandable, but that’s not the path he would choose.

Upon reflection, this is fundamentally a matter of patience and belief. Being willing to lay the groundwork for something that is not yet achievable boils down to betting that models will continue to improve and that one’s judgment about the direction is correct. Most people cannot do this, not because they don’t understand the trends, but because they cannot resist the urge to see returns immediately.

Believing in Exponential Growth is Harder Than It Seems

Altman mentioned a method he has always used: whenever he meets someone new, he mentally plots a coordinate for that person, assessing where they currently stand. The next time they meet, he observes how far and how quickly that person has moved forward. He said this is based on the same underlying belief he has about the progress of model capabilities—his deep trust in exponential growth, whether it pertains to a person, a company, or a model.

He stated that if he were still advising entrepreneurs, this is the one thing he would want them to truly understand. Moreover, he noted that this concept is difficult to accept broadly because the market itself has not yet adapted to the fact that model capabilities will continue to advance exponentially. Therefore, starting projects that require smarter and cheaper models is entirely reasonable.

I find this segment particularly worth pondering. Believing that a curve will continue to rise sounds like a simple idea, but basing all decisions on this belief requires more courage than one might imagine. Most people’s intuition about exponential growth is flawed; they either underestimate the accumulation of the previous years or begin to doubt whether the rapid ascent will continue during the fastest growth phase.

Enduring Chaos is Something You Can Only Learn Through Experience

One part that left a deep impression on me was when Altman said that no matter how much someone understands the reasoning, certain abilities can only be developed through repeated experiences—operating amidst chaos and believing that you will ultimately figure it out. He said this won’t cost you your life; even if you don’t know how to solve it now, you will find a solution. He emphasized that this is something that can only be learned, not taught, and he believes this is the biggest shortcoming of many young founders—they haven’t gone through the process of gradually learning to coexist with chaos.

He also shared a straightforward analogy: the first time you encounter a potentially catastrophic issue for your company, it feels like the sky is falling. By the time you’ve survived the tenth instance, you’ll think, "I’ve made it through the first nine; this one probably won’t be that bad." He later realized that bad things will always happen; instead of resisting them, it’s better to learn to accept this uncomfortable process. He said most people think the opposite of a bad experience is a good experience, but in reality, the opposite of a bad experience is no experience at all. In the not-too-distant future, you will inevitably enter a phase where nothing significant happens, so even terrible experiences are worth being grateful for.

This statement left me momentarily stunned. We are too accustomed to viewing pain as something to be avoided at all costs, but if the opposite is not comfort but rather emptiness and numbness, then enduring chaos seems not just a cost but a part of being alive.

What Promise Does a Trustworthy Company Offer to the World?

When discussing mission, Altman mentioned something he cares deeply about: one of his biggest concerns regarding AI risks is that a small group of people or a company might feel they should control the entire world, which he refers to as AI authoritarianism. Therefore, what OpenAI aims to do is make intelligence extremely abundant and cheap, placing it in everyone’s hands rather than in the hands of a few. He emphasized that they do not intend to create products in every vertical field themselves; instead, they want to enhance the foundational capability of intelligence, allowing the entire economic ecosystem to grow various things on that foundation.

There was one part I found particularly interesting: he said that the components necessary to create abundant intelligence—chips, energy, data centers, robots—are precisely what humanity will need immediately after intelligence becomes abundant. Even if ideas and creativity become worthless, we still live in a physical world where things need to be genuinely produced. Thus, energy and robots are not just stepping stones to that goal; they are also things that will be needed right after.

When I read this part, my first reaction was that this logic is quite simple. Ultimately, no matter how intelligent something is, for it to have any impact in reality, someone must move the material. However, upon further reflection, this also reminds us not to view intelligence as something too abstract; no matter how advanced a model is, it ultimately relies on a set of very heavy, very physical infrastructures to be realized.

The Invention of the Company is More Important Than Many Technologies Themselves

Altman shared some thoughts from his childhood; he has always been curious about the Industrial Revolution, where a bunch of technologies coincidentally emerged around the same time and expanded at a similar pace. He has always wondered which technology was the most critical. Looking back from today’s perspective, he believes the truly pivotal invention is the joint-stock company. Before that, businesses relied on trust among acquaintances, consisting of family businesses without the concept of shareholders. After the emergence of joint-stock companies, sovereign states granted this new entity an unprecedented status—not the power of a state, but capabilities far exceeding those of individuals, enabling capital accumulation and allowing for high-risk, highly speculative ventures, with different companies specializing in different aspects and collaborating with each other.

He referenced a graph I’m eager to look up, showing the decline in the proportion of extremely impoverished people in human history and the decline in infant mortality rates. If we stretch out the entire history of humanity and mark the time point when the joint-stock company was invented, the shape of the curves will look markedly different afterward. He stated that this represents a remarkable and extraordinary performance of capitalism in human society.

I find this segment particularly enlightening. When we talk about startups, we often discuss products, financing, and growth, but we rarely take a step back to consider that the organizational form of a company itself is a technological invention that binds the interests of a large group of people together. Thinking this way, entrepreneurship is essentially about leveraging this invention and adding your own elements on top of it.

Believe in a Few Things and Keep Everything Else Flexible

When discussing how to make long-term plans, Altman said he doesn’t usually work backward from the future to the present. His more habitual approach is to first identify a few directions he firmly believes in, then step forward from the current point in time, clarifying what can be done now and what can be done this year. Only in rare cases does he plan for five or ten years ahead. He mentioned having seen too many people hold onto a multitude of beliefs about the future, only to be constrained by their rigid worldview. You might see some rocket companies suddenly pivot to AI, which is such a situation. The truly useful approach is to hold onto a few deeply believed principles while keeping everything else flexible, firmly maintaining the core.

He mentioned a friend’s company core value, which is called "critical path," meaning to always focus on the biggest stumbling block in front of you, move it aside, then look for the next one, and repeat this action continuously. He said he has been very clear over the years that his critical path in life is to make intelligence abundant; as long as there isn’t an unusual concentration of power, he believes this will lead to tremendous prosperity. He stated that he is rarely tempted by other ideas and seldom thinks about whether to change his goal. Recently, however, he has started to seriously consider what comes next if superintelligence is indeed on the horizon.

I found this quite moving; for someone to focus on a single critical path for so many years without being distracted by other opportunities is more challenging than any planning methodology. Ultimately, effective planning is not about how accurately you can calculate but whether you can maintain belief in a few things over the long term while filtering out the surrounding noise.

Dare to Get on the Plane in a Risky State

Altman mentioned that he has always adhered to a principle: when in a somewhat risky state, you should get on the plane. He recounted the story from the time just after ChatGPT was released, when leaders around the world were very anxious, with some questioning whether this technology might spiral out of control. He sensed a storm brewing. Following the advice of Brian Chesky (co-founder of Airbnb), he decided to embark on an intensive tour, originally planned for about eight cities, but they ended up covering 28 countries in 35 days. He said he basically lived on planes during that period; it was a strange experience—though the flights were comfortable, traveling itself is still exhausting, and he missed the time zone, his own bed, and his office.

He also mentioned an interesting criterion for distinguishing whether something is a real trend or a fake trend. A fake trend looks like this: a lot of people are excited about it, but those who actually buy it lose interest after a while and don't design their lives around it, leading to it gathering dust. He used VR as an example. A real trend, on the other hand, is something that continues to appear in your daily life. For him, ChatGPT is almost used every day; sometimes for three hours, sometimes not at all, but it remains a constant part of his life. He said this judgment method was something he summarized after observing many startups at YC (Y Combinator). If one is willing to spend time analyzing the data, one can truly see a lot of things.

I really like this method of distinguishing between real and fake trends because it is simple enough. There’s no need to look at complex growth curves; just ask one question: has this thing quietly embedded itself into your daily rhythm, or did it just excite you for a while before being cast aside?

Asking for the Impossible

Altman shared an example about Codex (OpenAI's programming agent application), saying it was a memorable experience of asking for something seemingly impossible. At that time, they were clearly falling behind Claude Code (Anthropic's programming agent product) in the programming space. Normally, in such a situation, trying to turn the tide in a category where someone else has already taken the lead is considered impossible. Most people would accept this and move on to the next direction. However, they felt this matter was too important to give up easily, so they assembled a team and assigned them this nearly suicidal task. As a result, this team achieved a rare success in business history; among the best programmers around him, the most used programming tool is this product.

He stated directly that if they hadn't proactively asked for help and assigned this nearly impossible task to the team, none of this would have happened. His own explanation is that programming is too important for RSI (Recursive Self-Improvement), not to mention the economic value behind it; they couldn't convince themselves to abandon this track.

When I read this part, I thought about how asking for something can seem simple, but in practice, it is a strong counter to a very powerful social default. Everyone assumes the winner has already been determined, and trying to compete is destined to fail. But what Altman did was refuse to accept this default, first assuming it was possible, and then trying.

Projects That Need to Be Killed and Teams That Need to Start Over

During the interview, a painful question was raised: how do you make the decision to pull the plug on a project that has already been invested in for over a year, spending a lot of money, computing power, and effort, with users who are using and enjoying it? Altman said this is not something that can be decided in a single meeting; it is more like a slowly accumulating awareness. Gradually, one realizes that these resources, these people, and this product direction could create greater value elsewhere, leading to the painful decision to pivot.

He gave two examples: when GPT-3 (OpenAI's early language model) was truly operational, they shut down an exciting robotics project at the time and consolidated all resources onto it. Recently, after the programming agent became fully operational, they also shut down Sora (OpenAI's video generation product) and the browser, both of which were promising directions, and focused entirely on programming. He said this does not mean Sora was not doing well; if continued investment had been made, it could have been successful. It was just more important to allocate resources and energy to the programming agent at that time.

Regarding how to get the team to accept this shift, he said everyone understands the mission and the trade-offs behind it. Even if it is difficult at the moment, the team knows why this is necessary. Some may be unhappy, but more will say, "I understand why we need to do this; it is right for the mission."

I think the hardest part of this segment is not the decision itself, but how to get a group of people who have already invested a year of effort to believe that the next thing is also worth going all in on. This requires not only judgment but also strong communication and team leadership skills.

Focusing on Your Strengths and Finding the Right People for the Rest

When discussing how to become better at something, Altman cited Johnny Ive (former Chief Design Officer of Apple) as an example. He said the most important lesson he learned from Johnny is that truly great design comes from thoroughly studying the problem itself, rather than having a flash of inspiration for a solution. If one rushes to an answer or locks oneself into a solution too early, the result is usually not very good.

He also candidly admitted that he is not good at product design. He disagrees with the notion that one should only hire people who are deeply knowledgeable in their field. He said he doesn’t understand design at all, but after just thirty minutes of conversation with Johnny, it becomes clear that this person is genuinely talented. His principle is that instead of forcing himself to improve on inherent weaknesses, he should focus all his energy on areas where he is already strong, making those strengths even stronger.

There’s also a personal detail he mentioned: during the time he was working on Sora, he deliberately made himself addicted to TikTok to understand the product experience. Initially, he just wanted to learn, but later he found he genuinely enjoyed it, going from ten minutes before bed to an hour, and then to three hours on Saturday afternoons on the couch. He said that feeling was enjoyable in the moment, but he clearly knew it was not good for him. Eventually, he turned off most app notifications, including those from messaging apps, and deleted TikTok because he felt it was too powerful for him to control.

When I read this part, I was quite surprised that someone who is constantly creating more powerful AI products could also be affected by the very things they create, needing to resort to the most primitive methods of simply turning off notifications and deleting apps to regain control. This reminds me that judgment is not something that can be established once and for all; it requires continuous self-management, even for those who understand the design logic of these products best.

My Thoughts

Listening to the entire interview, I feel that what runs through it is not a specific methodology, but a mindset in the face of uncertainty. Believing that indices will continue to rise, enduring chaos until it is no longer frightening, holding on to a few deeply held beliefs, asking for what you need, letting go when necessary, and focusing your energy on what you truly excel at. These principles may not seem fresh when viewed in isolation, but the challenge lies in being able to do all of these things simultaneously in an environment where all assumptions are being overturned.

Altman concluded with a statement that left a deep impression on me: he said that most startups today still resemble those from ten years ago because the so-called correct practices are still being taught in the same way. At most, they have changed the wording, saying they hire fewer people and spend more on tokens, but that is far from enough. I think this statement serves as a reminder for everyone: the tools have completely changed, and if the way of thinking remains in the old coordinate system, then no matter how radical the words may be, the outcomes are likely to still be old.