Edward Feigenbaum and Raj Reddy receive the 1994 ACM Turing Award

Background

The A.M. Turing Award is given by the Association for Computing Machinery (ACM). The 1994 award went to two researchers who had each turned artificial intelligence from a research ambition into working systems: one taught programs to reason with an expert's knowledge, the other taught machines to follow continuously spoken speech^[1][2].

The award

The ACM lists Edward Feigenbaum and Raj Reddy as joint recipients of the 1994 A.M. Turing Award^[1]. The citation reads: "For pioneering the design and construction of large scale artificial intelligence systems, demonstrating the practical importance and potential commercial impact of artificial intelligence technology."^[1]

Two words carry the weight here: scale and practice. What the ACM honoured was not a proposal but the building of systems large enough to be useful, and a demonstration that such systems could matter in commercial terms^[1].

Feigenbaum and expert systems

Feigenbaum is known as "the father of expert systems"^[1]. An expert system encodes what a specialist knows — the rules a physician uses to narrow down a diagnosis, for instance — and applies it to a narrow problem. It does not try to reason about everything; it tries to be as good as a trained practitioner inside one domain.

Feigenbaum led the Heuristic DENDRAL project and the medical expert system MYCIN, and EMYCIN was the first portable general expert system "shell"^[1]. The word shell describes a separation that turned out to be important: the reasoning machinery is kept apart from the domain knowledge. Keep the inference engine, swap in a knowledge base for a new field, and you have a new expert system without writing a new program. That division of labour is what made the approach reusable.

Reddy and continuous speech

Reddy pioneered the construction of systems for recognizing continuous speech^[2]. Hearsay I was the first system capable of continuous speech recognition, and Hearsay II, Harpy and Dragon followed^[2].

Continuous speech is hard for a reason worth stating plainly: unlike speech dictated one word at a time, ordinary talk has no reliable pauses between words, and the same sound is pronounced differently depending on its neighbours. A recogniser therefore has to weigh many candidate word sequences at once, using acoustic evidence together with grammar and meaning to discard the wrong ones.

Reddy's systems also produced the "blackboard model" for coordinating multiple knowledge sources, which has been adopted across the spectrum of applied artificial intelligence^[2]. The idea is a shared workspace: independent specialists post partial results on the blackboard, and a control component decides which of them should contribute next. No single component holds the whole solution; the answer emerges from what is written on the board.

Where they stood at the time

At the time of the award, Feigenbaum was Professor of Computer Science at Stanford and had just been appointed Chief Scientist of the United States Air Force (1994-1997), while Reddy was founding director of Carnegie Mellon's Robotics Institute (1980-1992)^[1].

Why it mattered

The ACM page states that together the two joint recipients have been seminal leaders in defining the emerging field of applied artificial intelligence and in demonstrating its technological significance^[2]. The 1994 award honoured two different routes to the same demand — that a system work under real conditions. One route pushed expert knowledge into rules a program could apply, and gave applied AI the portable shell; the other built machinery for weighing uncertain evidence and letting several kinds of knowledge cooperate, and gave applied AI the blackboard model, which has been adopted across the spectrum of applied artificial intelligence^[1][2].