In the last post we introduced Minotaur, an AI architecture built on ELC that works out its answers by measuring rather than predicting. We showed it checking real regulations against records of how they were applied, finding which organisation broke a rule, by how much and where, and telling a broken rule from a badly recorded measurement.
Today's language models are admired for something broader: they can deduce, explain, diagnose, weigh a hypothesis, imagine what would have happened if things had been different, and change their minds when told something new. Minotaur's brain, the authors argue, does the main kinds of reasoning usually credited to large language models, including deduction, induction, abduction, diagnosis and reasoning about histories in which asking a question changes the answer. It does all of this with one operation on one network, on one ordinary processor core, at about a quarter of a second per question on a described system.
On blind tests written by people who never saw its code, Minotaur matched a current language model reasoning in words, on a police interview where the questions change the witness's memory, and on learning, diagnosing and predicting systems from their data alone. The post contrasts this with how a large language model is trained to predict the next word of text, which means it inherits both human strengths and human weaknesses, such as being swayed by believable conclusions and by how a question is framed.
The post also traces seventy years of earlier attempts at machine reasoning. Starting with Allen Newell, Cliff Shaw and Herbert Simon's Logic Theorist in 1956, these efforts worked up to a point and then stopped, which the author uses to explain why language models became dominant. Exact, checkable reasoning was the goal of most pre-language-model approaches.
The central claim is that Minotaur reasons by working on the structure of the system it is reasoning about, rather than by imitating how people write their thinking down.
Source: Hacker News · Summarized by HeadlinesBriefing