Microsoft's Chief Science Officer: As AI gets stronger, where do the opportunities come from?

Will humans have fewer opportunities as AI becomes more powerful?

On February 25, 2026, during a conversation at the Stanford Graduate School of Business, Eric Horvitz, the Chief Scientific Officer of Microsoft, offered a completely different answer.

This researcher, who has been deeply involved in artificial intelligence for four decades, placed AI on a rare time scale: In the history books 700 years from now, the 2020s - 2030s will be marked as an independent era.

Why does he say so? The comparison he provided is clear:

It took industries 100 years to recognize the full potential of the steam engine after its invention.

It took several decades for electricity to reshape society from its invention.

However, the penetration rate of AI is measured in months.

"This is a rare window of rapid change in human history."

Within this window, Horvitz sees not anxiety but opportunities.

However, these opportunities are not where most people are looking.

For example, managers focus on model upgrades, students worry about their skills being replaced, and white - collar workers are anxious about potential unemployment. These are not the sources of opportunities.

So, where do the opportunities actually come from?

Type One Opportunities: Integrating AI into Existing Businesses

The first type of opportunities lies in deployment.

Horvitz gave an example.

When electricity first emerged, factories used it in the following way: A large pulley was placed in the center, and belts were used to transmit power to each workstation. At that time, people never thought that each machine could be equipped with an independent electric motor. They simply regarded electricity as a more powerful form of steam and applied the old production methods.

It wasn't until several decades later, when someone redesigned the entire factory layout, that the true value of electricity became apparent.

Currently, AI is going through the same stage.

Everyone knows that AI is powerful, but people haven't figured out how to truly integrate it into organizations, business processes, and areas where it can generate value.

Horvitz said that this is where the opportunities lie.

Many startups he has seen recently are all doing the same thing: They choose a specific industry, delve deeply into it, study the specific business processes, figure out where AI can really be helpful, and then fine - tune the models.

These companies are not developing new AI technologies but are understanding how to make existing AI work.

Take healthcare as an example.

When many people talk about AI in healthcare, they think of applications in areas such as diagnosis and voice transcription. However, if you look at the entire healthcare system: payers, pre - authorization, responsibility allocation, the patient side, and the doctor side, it is a complex chain. AI can change not only the efficiency of a single link but the collaboration mode of the entire system.

The same goes for education.

AI can provide personalized tutoring, but its real value lies in redesigning learning paths and rethinking how education can be tailored to each individual.

Therefore, AI technology will become more and more powerful, and the models will become better and better, but technology is not the bottleneck. What is really lacking are people who can deploy AI in real - world scenarios and make it generate value.

To become such a person, the following abilities are required:

1. Delve into a field and understand its processes.
2. Know what AI can and cannot do.
3. Be able to design new ways for humans and AI to cooperate.
4. Be able to promote the implementation of such changes within an organization.

This is the first type of opportunity, and it is also the most easily overlooked one.

Because everyone is focusing on technological breakthroughs, but it is those who understand deployment that can truly transform technology into value.

Type Two Opportunities: Newly Opened Fields

The previous section discussed how to apply AI to existing businesses. However, there is another type of opportunity that comes from things that were previously impossible but are now achievable.

Horvitz used the term "edge of doability."

It means that there were some things that we didn't even dare to think about before, but now AI has made them possible.

This is a breakthrough from 0 to 1.

Let's still take healthcare as an example, but from a different perspective. He said that in our lifetime, we will witness a series of breakthroughs: Neurodegenerative diseases such as Alzheimer's, ALS (amyotrophic lateral sclerosis), and frontotemporal dementia, which have plagued humanity for decades, will take a turn for the better due to new insights and treatment plans brought by AI; even many types of cancer will become chronic diseases or be cured.

Why?

Because AI can do things that humans couldn't do before: design proteins, design molecules, and understand biological networks. In the past, scientists had to rely on intuition and luck to experiment. Now, AI can simulate countless possibilities and find combinations that humans would never think of.

Similar breakthroughs are also happening in materials science.

In the past, the discovery of new materials relied on the accumulation of generations of scientists. Now, AI can simulate atomic - level interactions and directly predict which combinations might be useful. What used to take decades of research can now be completed in a few months.

In addition to these scientific breakthroughs, there is another type of opportunity that comes from new challenges brought by AI, and authenticity is one of them.

As AI can generate content, imitate voices, and synthesize images, an urgent question arises: What counts as real?

The content credentialing technology that Horvitz's team is working on is redefining the verification mechanism. This involves not only technology but also the infrastructure of public life, industrial trust, and the media ecosystem. In the future, new verification standards, new trust mechanisms, and new industry norms will emerge in the field of authenticity.

Healthcare, materials science, and authenticity have one thing in common: They are all in new fields, not the old ones.

The old fields are highly competitive, and the strategies are well - established.

In the new fields, the rules are yet to be determined, and the landscape is still taking shape.

These new fields also have another characteristic: They are all at the intersections of different disciplines.

Horvitz said that many future breakthroughs will not occur within a single discipline but at the intersections.

People who can think across disciplines and connect knowledge from different fields are more likely to spot these opportunities and be among the first to enter these fields.

Section Three | But There Are Also Traps: Don't Let AI Weaken Your Thinking

The previous sections discussed where the opportunities are. However, in the interview, Horvitz also told a story about the risks behind the opportunities.

During Thanksgiving in 2023, his sister walked up to him, put her hands on her hips, and questioned him:

"What did you do to my students?"

His sister is a literature professor at the University of North Carolina. She found that no matter how much she warned her students, they were using AI to do their homework and not engaging in in - depth thinking. She blamed Horvitz for this. After all, he is the Chief Scientific Officer of Microsoft and a senior researcher in the field of AI.

Horvitz's reaction was interesting.

He didn't defend himself. Instead, he said that she was right. Learning how to write is part of learning how to think.

In his view, this is not just a problem among students.

As AI becomes more and more powerful, everyone faces the same temptation: Let AI write for you, think for you, and express for you.

However, the problem is that if you stop doing these things yourself, you lose the opportunity to think. Even worse, everything could be done by AI, and you may not even be able to tell what you came up with yourself and what was given by AI.

Recently, Horvitz asked a designer to create a small icon that says: 100% humaned (100% done by humans). He sometimes puts this icon at the end of his emails to tell the recipient that he wrote the email himself.

What does this detail indicate? When you habitually let AI do things for you, you will only passively accept and gradually lose your ability to think actively.

So, what should we do?

Horvitz said that humans must maintain the right of control. You need to know what you are doing and what to accept and what to filter.

For example, when writing a novel or exploring unknown scientific questions, if you are in control, the randomness of AI can help you find unexpected possibilities. However, if you just passively accept the answers given by AI, then you become the tool of the tool.

This made him think of a deeper question: Can we, on the contrary, design tools that actively cultivate human independent thinking? After discussing with his sister, he started to think about this question, but there is no answer yet.

AI is a double - edged sword. It can amplify human abilities but also replace human thinking.

When everyone is talking about the opportunities brought by AI, few people remind us that if you are just using the tool without maintaining your judgment and depth of thinking, you are not seizing the opportunities but being left behind by them.

Section Four | So, What Does It Take to Seize the Opportunities?

We have clarified where the opportunities are. So, what does it take to seize them?

Throughout the conversation, Horvitz emphasized just a few key abilities.

1. Curiosity

You don't need to master all the technologies from the start. What you need is curiosity.

Because curiosity determines whether you are willing to explore new paths and engage with things that were previously out of reach.

AI can do many things, but it only completes tasks and doesn't actively question why or what else is possible. This is unique to humans.

2. Find Your Interests and Delve Deeply

Find a field that you are truly interested in, delve deeply into it, and think about the relevance of AI.

Understand how AI can help with specific questions and directions, rather than learning AI in a general way.

3. Maintain Judgment

AI can provide materials and simulate possibilities, but the direction is always determined by humans.

People who can think independently and have their own judgment will not be left behind.

Ultimately, Horvitz always believes that AI will not weaken humans but enhance them. However, the premise is that the machines we build should serve humans, not take over human independent thinking.

What AI is accelerating is the possibility for humans to explore new fields, realize ideas, and break through boundaries.

As this process accelerates, opportunities will become more and more concentrated in the hands of those who can maintain a sense of direction, delve deeply, and connect across disciplines.

Because these are abilities that AI will never have.

Conclusion

As AI becomes more and more powerful, where do the opportunities come from?

The opportunities lie in two areas:

Firstly, deployment. Understanding how to integrate AI into existing businesses is more important than chasing after technology.

Secondly, new frontiers. In fields such as healthcare, materials science, and authenticity, which have just been opened up, there are no clear winners yet.

However, we also need to be vigilant against the traps: AI can amplify human abilities but also replace human thinking. What does it take to seize these opportunities? Curiosity, interest, and judgment.

AI is constantly evolving and upgrading, but the direction is always in human hands.

Technology does not determine the future; humans do.

Reference Materials:

https://www.youtube.com/watch?v=aWqfH0aSGKI

https://erichorvitz.com/ftp/boundedc.pdf?utm_source=chatgpt.com

https://www.microsoft.com/en-us/research/people/horvitz/

https://events.stanford.edu/event/a-conversation-with-eric-horvitz-chief-scientific-officer-at-microsoft-and-sarah-soule-dean-of-stanford-gsb

Source: Official Media/Online News

This article is from the WeChat official account "AI Deep Researcher". Author: AI Deep Researcher. Editor: Shen Si. Republished by 36Kr with permission.