After John Searle: what does the Chinese Room say today?

The American philosopher John Searle died on 17 September 2025 in Safety Harbor, Florida, at the age of 93. Born in Denver in 1932, he had taught at the University of California, Berkeley, from 1959.

In an assessment published this August, Berkeley's student newspaper called his legacy "complicated": his widely accepted contributions to the philosophy of language and mind and the serious allegations he faced at the end of his career are named in the same sentence.

A news site has to record both. What follows takes up why his thinking is still at the centre of the argument over artificial intelligence.

Speech acts#

Searle's first major contribution was in the philosophy of language. Speech Acts (1969), which systematised the line begun by J. L. Austin, stops treating language as merely a tool for describing the world.

The basic observation is this: to speak is to act. When you say "I promise", you are not reporting a fact; you are creating an obligation. "I hereby open this meeting" does not describe the meeting as open — it opens it.

Searle classified these acts and spelled out the conditions each must meet to count as valid. The framework is used from linguistics to law, from AI design to communication theory.

The Chinese Room#

What carried Searle's name beyond philosophy, though, was a thought experiment from 1980.

Imagine a person in a room who knows no Chinese. Sheets of paper bearing Chinese characters come in from outside. A thick rulebook tells them which character to answer with. The person applies the rules and passes the papers back out.

To a Chinese speaker outside, the room speaks perfect Chinese. But the person inside understands not a word.

Searle's conclusion: manipulating symbols according to rules does not produce understanding. Syntax alone does not generate semantics. So a computer, however well it runs a program, does not thereby understand.

The objections#

The argument has been intensively debated since it appeared. The main replies:

The systems reply. The person in the room may not understand Chinese; but what understands is not the person, it is the person-plus-rulebook-plus-room. Searle answered with a variant in which the person memorises all the rules: the system is then inside the person, and still there is no understanding.

The robot reply. Add sense organs and bodily action to the room, and the symbols connect to the world; meaning follows.

The brain simulator reply. If the program simulates the operation of the brain neuron by neuron, and the brain understands, then so does it.

Searle wrote detailed responses to all of these. The argument is not closed.

Where it stands now#

With the spread of large language models, the Chinese Room has returned to the agenda — for defenders and critics alike.

Those who find the argument strong say the models perform statistical pattern matching at enormous scale, and that meaning does not follow from that. Those on the other side argue that "understanding" cannot be defined independently of behaviour, and that the argument therefore begs the question.

David Chalmers's 2026 work, reported on this site, tries to pass between the horns: by attributing "quasi-beliefs" and "quasi-desires" to a language model, it proposes a way of talking about its mental life without entering the argument about souls.

Whichever side is right, the ground on which the argument is built still belongs to a paper written at Berkeley in 1980.

---

Note: the harassment allegations against Searle and the university's response to them are covered in detail in the Berkeley source cited above.