reasoning

AI (general)

The process of drawing conclusions, making inferences or solving problems by combining pieces of information through intermediate steps. Researchers distinguish deductive reasoning (conclusions that follow necessarily from premises), inductive reasoning (generalizing from cases) and abductive reasoning (inferring the best explanation), and speak of mathematical, causal, spatial, temporal and commonsense reasoningcommonsense reasoning, n.: The everyday background knowledge and reasoning people take for granted — dropped glasses break, hungry people eat, nobody is in two places at once — which machines long struggled to match. and of planning. Symbolic AIsymbolic AI, n.: An approach to AI that writes knowledge down as human-readable symbols and rules, such as 'every bird has wings', and reasons by manipulating them step by step instead of learning patterns from data. implements reasoning explicitly with logic, search and theorem proversautomated theorem proving, n.: Getting computers to prove mathematical or logical statements by themselves, by searching for a chain of valid steps that leads from basic assumptions to the statement.; probabilistic AI reasons under uncertainty with tools such as Bayesian networksBayesian network, n.: A diagram of causes and effects with probabilities attached — say, how rain and a sprinkler can each make grass wet — that lets a computer update its beliefs about hidden causes when it sees evidence..

Language models

Multistep problem solving by large language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code., typically by generating intermediate text before an answer (chain of thoughtchain of thought, n.: Getting an AI to write out its reasoning step by step before giving an answer, the way a student shows their working. This often makes answers to math and logic problems more accurate.). Since 2024, reasoning modelsreasoning model, n.: An AI model trained to work through a problem step by step in hidden 'thinking' text before it answers. Letting it think longer generally makes it better at math, coding and planning. trained with reinforcement learningreinforcement learning with verifiable rewards, n.: Training an AI by letting it attempt problems whose answers can be checked automatically — a math result, a passing program — and rewarding it only when the check succeeds. to produce long reasoning tracesreasoning trace, n.: The written-out 'thinking' an AI reasoning model produces before its final answer — notes to itself full of attempts, checks and corrections, which users may or may not get to see. have made large gains in mathematics, science and coding, spending more test-time computetest-time compute, n.: The computing effort an AI model spends while answering a question, as opposed to while being trained. Letting a model think longer or try many answers can make it markedly smarter. on harder problems. Whether this is genuine reasoning is contested: some researchers point to success on novel problems, others to brittleness under small changes to problems, reliance on patterns resembling training data, and traces that may not reflect the computation actually behind the answer (chain-of-thought faithfulnesschain-of-thought faithfulness, n.: Whether the reasoning an AI writes out really reflects how it reached its answer, rather than being an after-the-fact story that leaves out what actually drove the decision.).

The model got the right answer, but its reasoning trace skipped the step that actually mattered.