Sam Altman: Why AI Will Surpass Human Intelligence by 2030

The lights come up on the scene in Berlin.

On September 25, 2025, local time, the Axel Springer Award ceremony kicked off at the Berlin headquarters. The star of the evening was Sam Altman, the CEO of OpenAI.

During this trip to Berlin, Altman clearly put forward a judgment:

By 2030, superintelligence is very likely to emerge.

Almost simultaneously, a more subtle signal emerged. Altman also announced on Twitter:

“Today, we're launching my favorite ChatGPT feature, called Pulse.”

Behind this feature update lies a major shift. What's being enhanced is neither chat fluency nor response speed, but rather, it's the first time AI has started to think actively for users.

Altman's description is straightforward: You might casually mention, “I want to go to Bora Bora,” and it might remind you to apply for a visa.

This isn't just an auxiliary tool; it's an outsourcing of the subconscious.

This isn't information recommendation; it's behavior prediction.

Superintelligence isn't just a talking robot; it's intelligence that starts thinking for you.

And Pulse is the first step.

This time, the turning point for AI isn't computing power or parameters; it's the shift from waiting for your questions to thinking for you in advance.

Section 1 | AI No Longer Waits for Your Questions; It Starts Thinking Actively

Altman said that Pulse is his favorite ChatGPT feature to date (as shown in the tweet above).

This isn't just a polite remark; it's a signal.

In the past, ChatGPT waited for your questions. You'd say, “Help me write an email,” and it would start working. Now, Pulse will actively analyze what you've recently talked about, the ideas you've mentioned, and the interests you've expressed while you're sleeping.

The next morning, it will present a list of things you might want to know.

It's not about what you asked; it's about what it guesses you'll want to know.

Altman gave an example during the conversation:

“For instance, if you mention in casual conversation, ‘I hope to go to Bora Bora one day,’ or ‘My baby is six months old, and I want to learn about the baby's development...’ In the future, you might receive relevant updates.”

This isn't answering questions; it's preparing for you in advance.

You don't issue a clear instruction, and it doesn't wait for your command. Instead, like a personal assistant, it remembers what you care about and thinks one step ahead for you.

Altman even directly referred to this as ChatGPT's “active mode”:

“This represents the future direction of ChatGPT: from being completely passively responsive to being truly active and extremely personalized.”

You no longer need to input prompts sentence by sentence. You just need to express yourself. And AI starts to think more, remember more, and do more for you.

In other words, it's not that AI has become smarter; it's that it starts to consider things for you before you even speak.

Section 2 | Altman: I Expect Superintelligence to Emerge Before 2030

At the interview, the host asked: When do you expect an AI smarter than all humans to emerge?

Altman didn't beat around the bush and directly responded:

I expect the development speed of AI to be extremely astonishing. By 2030, I'd be very surprised if a powerful AI that surpasses human capabilities hasn't emerged.

He wasn't being evasive; he provided a clear time frame.

Then, he added another sentence: I'd also be surprised if we don't see a similar progress rate to that of 2024 and 2025 by the end of 2026.

In other words, he isn't vaguely talking about some future day; he believes that AI will undergo a leap - forward upgrade in the next few years.

So, just how powerful is this particularly strong AI?

It's not about answering questions a few seconds faster or learning a few more languages. Altman used this sentence to describe its capabilities: It can make scientific discoveries that humans can't achieve on their own.

“Scientific discovery” means that it doesn't just have a large memory; it can also infer, break through, and create new knowledge.

Altman also gave an example:

“We've already seen that GPT - 5 has started to come up with new ideas that scientists consider valuable, especially in the fields of physics and mathematics.”

Not all of these ideas are necessarily correct, but it has indeed proposed some things that humans haven't thought of yet.

In his view, this kind of AI is no longer just a small tool for assisting in writing copy or generating images; it's an intelligent entity that can participate in research, innovation, and even discover patterns in advance.

Therefore, he started using another term to describe them: Extremely powerful and capable of doing things we can't do.

And this is what he calls “superintelligence.”

It's not about having emotions or subjective consciousness; it's about surpassing human capabilities.

It's not about becoming more human - like; it poses a question to humans:

Are you ready to work with this kind of intelligence?

Section 3 | Being Replaced by AI? The Most Important Thing is to Care About Others

Altman was asked a personal question: As a father, what kind of education do you hope your children receive so that they won't be replaced by AI in 30 years?

His answer wasn't to learn programming or get into a prestigious school.

It was four words: “Learn how to learn.”

He said that future AI can help you with many things, but there are some things AI can't do, such as:

“Figuring out what others want, creating truly useful products and services for them, and establishing a real connection with the world.”

These abilities, in essence, are empathy, insight, and understanding of people themselves.

Altman provided an important definition:

I've found that the most important ability of a person is knowing how to care about others.

This statement sounds more like a psychologist's view. But precisely in an era when AI is becoming more and more capable, this is the most needed ability for humans.

Because when AI helps you write PPTs, reports, find information, and extract data, what truly makes you irreplaceable isn't speed but your ability to understand what others want.

Altman also left a warm message:

“AI isn't the protagonist, but it can make your life more wonderful.”

Your life, your choices, and your goals are still up to you. AI is just a tool, an assistant, and a partner on standby. It can help you realize your ideas, but it can't decide your direction for you.

Only when you have clear goals can it cooperate precisely; only when you know what you want can it know how to help you.

He isn't preaching; he's reminding:

The more proactive AI becomes, the clearer humans need to be about what they care about.

Section 4 | Whoever Controls Computing Power Controls the Future

The more capable AI becomes, the more crucial the energy and computing power behind it are.

Altman said in an interview: Only by building sufficient computing power can AI truly realize its value.

This statement might sound like it's about technology, but in fact, it's about a very real issue: Who can build the foundation for AI to run?

In the past year, while developing models, OpenAI has been building AI factories globally.

(Source: TechCrunch, Five new Stargate data centers of OpenAI)

On September 23, OpenAI, in collaboration with Oracle and SoftBank, announced that they will add five large - scale AI data centers in the United States, located in states such as Texas, New Mexico, and Ohio, with an investment of over $400 billion in three years. By the end of 2025, the total computing power of the entire Stargate project will reach 10GW (equivalent to the total power generation of a medium - sized country).

Altman put it like this:

“Each of the data centers we're building now has tens of thousands of machines and thousands of construction workers on - site... And all these resources might only serve a small portion of ChatGPT users.”

Behind this statement lies the reality of AI: For ChatGPT to think for you in the morning and help you plan your life, it needs strong energy, electricity, and equipment support.

That's why governments around the world have started to pay attention to the issue of controlling AI infrastructure.

OpenAI has begun to promote the local deployment of the Stargate project in multiple countries. It has already launched in Norway and the UK to build sovereign AI infrastructure.

During this interview, Altman specifically mentioned the importance of the German market. Almost all young Germans are using ChatGPT. To this end, OpenAI has joined hands with SAP to launch a “German - version OpenAI” service to achieve local AI deployment and keep data within the country.

Behind this local deployment lies a deeper consideration. Altman emphasized:

This technology will become extremely crucial, and countries need to be able to truly rely on it.

Actually, this isn't about the model; it's about sovereignty.

From the model to computing power, from electricity to data compliance, from usage rights to access control, the closer AI gets to people's lives, the more it becomes a national strategic resource.

The Stargate project and the German - version OpenAI are just the beginning.

Ultimately, which people and which regions will have control over AI's capabilities is likely to be determined not by technological leadership but by infrastructure construction, resource mobilization, and cooperation methods.

Section 5 | Jobs Won't Disappear; They'll Just Be Done Differently

The more powerful AI becomes, the more people worry: Will many occupations disappear?

During the interview, the host asked Altman: Will positions from accountants to bank advisors soon disappear?

His answer wasn't a perfunctory one; he looked at it from a different angle:

A more useful way to think about it is to look at the percentage of tasks, not the percentage of occupations.

What does this mean?

He explained: A job consists of many different types of tasks. AI will take over some of them. Originally, you had to do five things a week, but now AI can help you complete two or three. Your role won't disappear immediately, but it will change significantly.

This is already happening.

• For example, doctors are already using GPT models to assist in diagnosis and write medical records;
• For example, legal advisors are using AI to help with preliminary retrieval and contract writing;
• For example, programmers have AI automatically generate a lot of basic code, and they mainly focus on “thinking about how to break down problems.”

Altman also mentioned a statistic: Approximately half of the jobs in society change every 75 years. This rule has existed even in the era without AI.

It's just that in the AI era, this pace will be faster.

New occupations will keep emerging, and old occupations will gradually disappear.

Regardless of how occupations change, Altman believes that one principle remains constant: AI is always just a tool. It can help us do more things, but we still have to decide what's important and what's worth doing.

What's truly reshaped by AI isn't just job titles but the way of creating value and solving problems.

This isn't a threat; it's an opportunity: You can finally hand over those mechanical, repetitive, and boring tasks and devote your time to more worthwhile things.

The key is that you need to actively embrace this change.

Conclusion | Superintelligence is Coming. Have You Figured Out What You Want?

Why is Sam Altman so confident in giving the clear year of 2030?

The answer lies in three signals:

• The growth curve of AI capabilities is extremely steep, and Altman doesn't see any signs of it slowing down.
• GPT - 5 has comprehensively defeated humans in math and programming competitions, and AI has started to come up with new ideas recognized by scientists.
• Most importantly, starting with Pulse, AI has begun to think actively without waiting for your questions.

From passive answering to active prediction, this isn't just a functional upgrade; it's a fundamental transformation of the intelligent model.

At this rate, 2030 isn't an optimistic prediction but a conservative estimate.

True superintelligence might not suddenly appear one day and announce, “I'm smarter than humans.” Instead, it will quietly permeate every corner of your life and think about things you haven't even considered yet.

It has already begun.

References:

https://www.youtube.com/watch?v=cC4Pm8NS4CA&t=2s

https://www.youtube.com/watch?v=I0002 - 2OBAU&t=6s

https://www.businessinsider.com/sam-altman-predicts-ai-agi-surpass-human-intelligence-2030-2025-9

https://www.axelspringer.com/en/ax-press-release/sam-altman-receives-the-axel-springer-award-2025

Source: Official media/Online news

This article is from the WeChat public account “AI Deep Researcher”, author: AI Deep Researcher, editor: Shen Si. Republished by 36Kr with permission.