IN SEARCH OF A MARKET (Joe Watson)
(Note by Gene Helms: George Heilmeier championed the opportunity
for TI in Artificial Intelligence (AI) during the 1980’s when
an exciting wave of new activity rippled through the academic community.
He directed a large part of TI’s Computer Science Laboratory
under Floyd Hollister toward software research in this area. The
computer language LISP had become the "language of choice"
for AI researchers, and MIT had sponsored research in computer architecture
optimized for LISP processing. TI acquired licenses from MIT plus
an early design start-up at Western Digital. Joe Watson led an Austin-based
program based on these acquisitions that designed and produced three
generations of the Explorer high-end workstation optimized for LISP
processing, and later as a UNIX server. At one time, there were
only two high end workstations in existence that were considered
suitable for AI research and development: TI’s Explorer and
a comparable product from Symbolics. Following is Joe Watson’s
account of the story.)
It became clear early in the 1980’s that the IBM PC had set
the standard for personal computing. But was there an opportunity
for a TI high-end workstation for technical applications? Several
believed that such an opportunity might be a good fit for TI’s
strengths. UNIX-based high-end workstations were proving popular
for engineering applications, including integrated circuit design---an
area of principal importance to TI.
An MIT project led by Stephen Ward had developed the NuBus, which
appeared to be ideal for a UNIX-based workstation. Two of Ward’s
former students had begun a startup business within Western Digital
to develop and market the NuMachine, based on the NuBus. TI acquired
this startup, which had begun delivering NuMachines to MIT. After
30 or so had been delivered, TI opted to discontinue the NuMachine
rather than re-engineer the machine for volume production.
Concurrently with these developments, a rebirth in artificial intelligence,
or "AI", appeared to be emerging. Although AI had been
around for several decades, little significant commercialization
had occurred. MIT, Stanford and Carnegie-Mellon were the AI citadels,
where researchers were using primarily the LISP language as a basic
research tool. TI concluded that an opportunity existed for a rather
specialized AI workstation that would directly execute the Lisp
language. There were two suppliers of these "LISP machines"
in the market. It was reasoned that if there were to be significant
AI commercial applications, as it seemed, there would be significant
Lisp machine markets for developing and delivering these applications.
To address this perceived opportunity, the TI Explorer, a LISP machine,
was developed. It was based on the NuBus and used the MIT LISP software.
Several Explorers were delivered to MIT as part of the license agreement.
A later version of this product was called Explorer II. Explorer
III, the final product in this series, retained the NuBus but was
re-engineered for use as a UNIX server.
A companion product for TI’s "AI thrust" was a
PC software package called the "Personal Consultant",
which provided development and run-time tools for "expert systems".
Such "knowledge-based" and "rule-based systems",
were the most popular areas for AI application. The idea was that,
with PC-based software, a much broader base could be established
for AI applications, with the Explorer being available for the more
computationally intense requirements.
Probably one of the best known commercial expert system applications
from this era was the Campbell soup story. One of their long-time
experts, who had intimate and detailed knowledge about the soup
processing factory flow, was about to retire. Soup was cooked in
the can after sealing, and it was essential that a carefully controlled
time and temperature profile be followed in the cooking process,
or the soup would be spoiled. Whenever one of their many soup canneries
developed a flow problem, the result was a lot of bad soup, so the
expert would be consulted to quickly correct the problem. In order
to preserve the expert’s knowledge base, a group of TI "knowledge
engineers" was called in to capture his knowledge, codify it
and make it available through an expert system. The company reported
that this project was very successful. There were other projects
that were notable, some of which met operational needs for several
years. One airline company did scheduling on a rule-based system,
running on TI Explorers.
As it turned out, the pure AI market proved to be very limited.
AI techniques were absorbed into the more general set of tools used
in software development supported by UNIX and, later, Windows environments
that became de facto standards. While LISP machines, and specialized
AI PC software, were a relatively short-lived market phenomenon,
the NuBus survived for awhile, appearing in the Apple Mac II, but
was soon overtaken by newer technology.