Why I Don’t Believe AI Is an Existential Threat to Humanity

There is growing concern that artificial intelligence will eventually become so intelligent that it surpasses humanity, escapes our control, and becomes an existential threat.

I don’t find that argument convincing.

My view starts with a simple premise: AI is fundamentally derived from human intelligence. Modern AI models are trained on information created, discovered, organized, and expressed by human beings. Their understanding of language, mathematics, science, engineering, philosophy, and reasoning comes from the accumulated intellectual work of humanity.

AI can process that knowledge at a scale no individual person could match. It can analyze enormous amounts of information, identify patterns, combine ideas from different fields, and explore possibilities at extraordinary speed.

But I think we make a mistake when we confuse speed, scale, and breadth with a fundamentally higher form of cognition.

AI Accelerates Human Cognition

A calculator can perform arithmetic faster than a person. A computer can search billions of records faster than any researcher. Simulation software can test thousands of engineering possibilities long before humans could physically build them.

AI takes that principle much further.

Suppose an AI-assisted research system discovers in three years something that human researchers might otherwise have discovered in thirty. That would be enormously significant, but I would describe it as an acceleration of discovery rather than evidence that AI has transcended human intelligence.

The destination was still reachable through human reasoning, experimentation, and accumulated knowledge. AI simply shortened the path.

That distinction matters.

AI may produce ideas or discoveries that surprise us, but surprise alone does not demonstrate a form of intelligence beyond humanity. It may simply mean that a machine explored possibilities faster and more broadly than humans could on their own.

The Ceiling Is Humanity, Not an Individual Human

When I say AI has a human intellectual ceiling, I am not claiming that AI cannot outperform individual people.

It obviously can.

No single person can read millions of books, remember everything humanity has written, or maintain expertise across hundreds of disciplines at once.

But that is not the relevant comparison.

AI is drawing from the collective intellectual output of humanity.

In that sense, AI can be far more capable than an individual while still operating within a framework ultimately derived from human knowledge and reasoning.

A system that can combine the insights of millions of people may look superhuman compared with any one person. But greater breadth and processing speed do not necessarily mean that it has crossed into an entirely different category of intelligence.

The Superintelligence Argument Makes a Large Leap

The existential-risk argument often follows a progression like this:

AI systems are becoming more capable. Therefore, they will eventually surpass humanity's intelligence. Once that happens, they may become uncontrollable.

The first statement is clearly true.

The second and third are much larger claims.

Increasing computational power does not automatically prove that intelligence can increase without limit. Processing more information does not necessarily create a form of cognition independent of the intellectual foundations from which the system was built.

That is where I think much of the fear surrounding runaway superintelligence becomes speculative.

There is an important difference between saying AI will become extraordinarily capable and saying AI will become an intelligence that exists beyond meaningful human cognitive boundaries.

I see strong evidence for the first. I do not think the second follows automatically from it.

AI Can Still Be Dangerous

None of this means AI is harmless.

AI can make fraud easier, scale misinformation, assist cyberattacks, produce convincing false content, disrupt jobs, and amplify poor decisions. Governments, corporations, criminals, and ordinary people can all misuse increasingly capable systems.

Those are legitimate concerns.

But most of these dangers are not entirely new categories of harm. They are existing human capabilities amplified by technology.

Humans already deceive, manipulate, attack computer systems, spread propaganda, build weapons, and make disastrous decisions.

AI can make some of those things faster, cheaper, and more scalable.

That is a serious problem, but it is fundamentally different from the idea of an independent superintelligence emerging and turning against humanity.

AI Is Leverage

I think the better way to understand AI is as cognitive leverage.

Machines gave us physical leverage. Computers gave us computational leverage. The internet gave us informational leverage.

AI gives us the ability to apply human knowledge and reasoning at unprecedented speed and scale.

And like every form of leverage, it can amplify both good and bad intentions.

A scientist can use AI to accelerate research. An engineer can explore designs faster. A programmer can build software more efficiently. A criminal can automate fraud. A government can expand surveillance.

The tool magnifies capability.

That means the most important question may not be whether AI itself becomes humanity's enemy.

It may be what humans do once they possess that much cognitive leverage.

That is where I believe the real risk lies.

AI may dramatically increase the speed at which humanity thinks, discovers, builds, and acts. But I do not see that acceleration as evidence that AI has escaped the boundaries of human intelligence.

I see AI as an amplifier of humanity.

And whether that amplification ultimately benefits or harms us will depend far more on how we use it than on the idea of some future intelligence existing beyond us.