The Lighthill report criticises AI research in the UK

Artificial Intelligence: A General Survey was written by Sir James Lighthill FRS, Lucasian Professor of Applied Mathematics at Cambridge. It is dated July 1972 and was published by the UK Science Research Council (SRC) in early 1973, in a paper symposium with responses from other scientists.^[1]

Background

The SRC chairman, B. H. Flowers, asked Lighthill for a personal review, because the field seemed too contentious to assemble a dispassionate panel of experts. The Council discussed the report in September 1972 and decided to publish it alongside alternative views, including comments by N. S. Sutherland, R. M. Needham, H. C. Longuet-Higgins and Donald Michie.^[2]

What the survey did

Lighthill based the survey on about two months of reading and discussion. He divided the field into three categories: A, Advanced Automation; B, "Building Robots," meant as a bridge; and C, computer-based studies of the central nervous system.^[1] In plain terms, A is practical automation, C is using computers to study the brain, and B was meant to connect the two. Lighthill judged that the attempt to build that bridge was not succeeding.^[1]

Lighthill's review was therefore a judgment about research directions as much as about results. It asked which parts of the field were delivering on their promises, and it concluded that the two-way link between practical automation and brain studies was not yet working.

Key details

He singled out the "combinatorial explosion" as a general cause of disappointment, an obstacle to building a self-organising system on a large knowledge base. The term describes how the number of possible combinations grows so quickly with problem size that exhaustive search becomes impractical.^[1]

He noted that the best chess programs in England played only at the standard of experienced amateur county club players, and that chess masters beat them easily.^[1] He also predicted that over the next 25 years the field would split further, with categories A and C developing closer ties to their own applications.^[1]

His prediction of a split between categories A and C was a forecast, not a finding, and it is one of the review's claims that can be compared with what came later.

Influence

According to the Chilton Computing overview, the report was highly critical of basic research in robotics and language processing. It caused a loss of confidence in AI among the UK academic establishment, including the funding body, and this lasted almost a decade.^[3]

The debate also became public. In June 1973 the BBC televised a debate at the Royal Institution, the "Lighthill Controversy," with Lighthill, Donald Michie, Richard Gregory and John McCarthy.^[3]

The report was published with responses, and the debate that followed was public, so both the critique and its replies are part of the record.