Can AI think or be conscious?
There is no evidence that today's AI is conscious or understands the way people do. It produces convincing language by predicting patterns. Whether a machine could ever be conscious is an open question that scientists and philosophers disagree on, and there is no agreed test for it.
The full answer
Today's AI is not known to be conscious, and most researchers do not think current chatbots are.
A language model produces fluent text by predicting patterns learned from its training data. Fluency is not evidence of feeling or awareness: it can describe sadness or say "I understand" because those phrases are common in what it learned, not because it experiences anything.
"Think" is also a slippery word. AI systems can solve problems, plan steps and write code, which looks like thinking. Whether this counts as understanding is argued over, and the answer depends on what you mean.
Could a machine ever be conscious? That is unresolved. There is no accepted scientific theory of consciousness, so there is no agreed test. Some researchers think it is possible in principle, others think it needs biology, and many say we do not know.
For practical purposes: treat AI as a powerful tool that can seem human, not as a person, and be careful about trusting it as one.
Read moreSuperintelligence: the full encyclopedia articleGlossary