The Answer Depends on a Definition
Whether computers can think has no settled answer, partly because “thinking” is not a precisely defined term. It can mean solving problems, having experiences, understanding meaning, or reasoning about oneself. A machine might do well on some of these and show no evidence at all of others. Much of the disagreement comes from people quietly using different definitions.
What is well established is what computers do: they carry out defined operations on data, extremely fast. Whether those operations, arranged in the right way, could amount to thought is the question philosophers and scientists still argue about.
Turing’s Proposal
In 1950 the British mathematician Alan Turing published a paper titled “Computing Machinery and Intelligence.” He thought the question “Can machines think?” was too vague to be useful, so he proposed replacing it with a practical test, now usually called the Turing test or the imitation game. A human judge exchanges written messages with both a person and a machine. If the judge cannot reliably tell which is which, the machine has passed in some sense.
The test has been valuable because it turns a fuzzy question into something observable. It also has well-known limits. It measures conversational behavior, not what is going on inside. A system could imitate human conversation by clever tricks, and a genuinely capable system might fail by being too accurate or too different from people. Most researchers today treat it as a historical landmark rather than a precise scientific measure.
The Chinese Room
In 1980 the philosopher John Searle offered a famous objection. Imagine a person who does not understand Chinese, sitting in a room with a large rulebook. Slips of paper with Chinese symbols come in, and the person follows the rules to pass out other symbols that make up sensible replies. To an outside observer, the room appears to understand Chinese. Yet the person inside understands nothing; they are just following rules.
Searle argued that running a program is like this: it manipulates symbols according to rules without grasping what they mean. On his view, a computer can simulate thinking the way a weather simulation can model a storm without being wet.
Critics have responded in several ways. One reply says the understanding belongs to the whole system, not the person alone. Another says that a program connected to sensors and a body in the world might acquire meaning through interaction. Searle and his critics never reached agreement, and the argument is still taught as a way of clarifying what each side assumes.
The Functionalist View
One influential position, called functionalism, holds that what matters about a mental state is the role it plays, not the material it is made of. If that is right, then in principle a mind could run on silicon as well as on neurons, as long as it performs the right functions. Supporters point out that neuron signals are physical processes that can be described and, to some extent, modeled mathematically.
Opponents think something may be missing. Some argue that consciousness, the felt quality of experience, may depend on the specific biology of living brains. Others argue that it may depend on a kind of organization we do not yet understand. Science currently has no agreed theory of how experience arises in any physical system, including brains, so claims in either direction go beyond the evidence.
What Current Systems Show
Modern software can translate languages, write fluent text, recognize images and play games at a very high level. Many of these systems use machine learning, in which a program adjusts internal numbers to fit large amounts of example data. Their performance can look like understanding, and in narrow tasks it can be impressive.
There are reasons for caution. These systems can produce confident but wrong answers, can fail in surprising ways on inputs that differ from their training data, and generally lack the broad, flexible common sense that people use daily. Whether this reflects a temporary gap in capability or a deeper difference in kind is debated. Researchers hold a range of views, and no single measurement settles it.
It also helps to separate two questions. One is about capability: can a machine perform tasks that normally need human thought? The answer is increasingly yes for many tasks. The other is about experience: does the machine feel or understand anything? There is currently no accepted way to test for that, even in other people; we infer it from similarity and behavior.
Comparing Brains and Computers
Brains and computers differ in obvious ways. A computer stores information as binary code and executes steps one after another or across a limited number of processors, while the human mind arises from billions of neurons working in parallel, shaped by a body, emotions and a lifetime of experience. Differences in structure do not by themselves prove that one can or cannot think, but they remind us that an analogy between the two is a simplification.
People also tend to read minds into things. We readily attribute feelings to pets, cartoon characters and chatbots. This is a kind of cognitive bias, and it means that our intuitions about whether a machine is thinking can be unreliable in both directions.
Where Things Stand
A fair summary is that computers clearly perform many tasks that once required human thinking, but whether any of them understands or experiences anything is an open question. Different thoughtful people reach different conclusions depending on their views about meaning, consciousness and what a mind is. For now, the most defensible position is to describe what machines can demonstrably do, be careful about claims of understanding, and stay open to evidence from neuroscience, computer science and philosophy that may sharpen the question.