THE DIVIDED MIND OF ARIFICIAL INTELLIGENCE

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP90-00965R000100260007-9
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 5
• Document Creation Date: December 22, 2016
• Document Release Date: December 21, 2011
• Sequence Number: 7
• Publication Date: February 15, 1987
• Content Type: OPEN SOURCE

_l I
Si Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
IMAGE
15 February 1987
The Divided Mind of
ART
IF!
DIAL
INTELLIGENCE
In Silicon Valley, the best and the brightest
are developing weapons that think. Now some of these
high-tech wizards are beginning to think twice.
BY IAIt1.AEEBS 7
--~ Less than two years ago, Pierre Blais was busy at work on the CIA's central brain. A practitioner in the
esoteric realm of artificia in~tel g nce, Blais was part of a team contracted to build a computer system
that would automatically take in messages from sources all over the world, "read" them, reformat them
and then route them to appropriate desks inside the giant Central Intelligence Agency headquarters in
Langley, Virginia. The work was not only stimulating, it was lucrative: Blais was earning enough to settle into
East Coast upscale suburbia and assure his wife and three daughters a future of cozy affluence.
Today Blais has traded the sprawling CIA headquarters for a funky frame house in Santa Cruz where a
handful of ardently antimilitary hackers are trying to create artificial intelligence for the masses-a program
that will run on IBM PCs instead of CIA mainframes. For his contribution he is paid in "maybe money"
-maybe, if the company ever turns a profit, he'll receive a check. Blais and his family now live in a compact
tract home in a scruffy corner of San Jose-a house he may have to sell to stay ahead of the bill collectors.
Pierre Btais is quite sane. But he is not what most in defense-dollar-flooded, technology-happy Silicon
Valley would call normal. The problem is that two years ago Pierre Blais began to develop a nagging
conscience.
Blais started to study up on the CIA, and the more he learned about its government-toppling adventures the
less comfortable he felt about adding to its power. Later. Blais came to the conclusion that artificial
intelligence or Al-the attempt to create machines that emulate human reasoning-has been commandeered
by America's military establishment and is being enlisted to help create a new generation of ill-conceived and
macabre weaponry.
Blais is not alone in his concern. Based on interviews with other Al designers and researchers in the Palo
Alto area-what could arguably be called the artificial intelligence capital of the world-many in this rarif Ted
domain seem to be wrestling with their consciences. Each person interviewed hit on the same two themes: In
the high-tech world, cutting edge Al work can be the most exciting high of all. And these days it is very
difficult to find an Al company or research lab that isn't taking money from the Defense Department.
Therein lies these wizards' dilemma.
P Terre Blais looks the part of a "knowledge engineer"-40 years old and tall, with sandy hair receding a
bit at the forehead, wire-rimmed glasses and clothes that look to have come off the no-nonsense rack at
Sears. As a teenager in Montreal, he was a self-described juvenile delinquent who rode with a
motorcycle gang and fantasized about becoming a Green Beret. In 1966, a time when many American
youths were fleeing over the northern border to avoid the Vietnam draft, Blais, a Canadian citizen, headed
south to Plattsburgh, New York, to enlist with the Army's 101st Airborne Division.
David Beers is an editor at Pacific News Service. His last article for Image was -Tomorrowland" (January 18).
Continued
Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
 Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
"I wanted to see some action, and the Canadian armed forces
weren't doing anything. I saw it as a rite of manhood," he says in the
measured.. precise tone of a technician, a tone that Blais rarely
abandons.
By January of 1968 he was running search and destroy missions
smack in the middle of the Tet offensive. Blais earned a Bronze Star
with a "V" for valor for his part in ambushing a platoon of North
Vietnamese. But the romance was already gone. He remembers
arriving in Hue. a city on Vietnam's central coast, to find "thousands
of bodies, most of them civilians, stacked like cordwood along the
road, with rats running all through them." Fellow platoon members
tortured and shot a captured enemy soldier while Blais, paralyzed
with horror, stared at a confiscated photo of the man's family. Some
nights, as he stood guard outside his camp, Blais would find himself
crying and praying for deliverance.
Honorably discharged in 1969, Blais returned to Canada, studied
business and mathematics at McGill University in Montreal and
tried to reassemble his life. "What happened in Vietnam is that all
the myths about life had been shattered. All the myths built by my
family, my school, my society, nation, history, all of these were
shattered in one blow," Blais says. It left him with a philosophy of
"pure nihilism plus pure pessimism. But I wasn't ready for that. I
was too young."
One day some Mormon missionaries knocked on his door. Blais
liked what he heard about the New Zion, an idyllic society out west
bound by strong values of church, family and honest work. He joined
the church, met and married a fellow convert from Okinawa and
enrolled as an undergraduate in international relations at Brigham
Young University in Provo, Utah. Later he went on to complete a
master's program in Asian studies there.
At BYU Blais found himself part of a small but growing cadre of
academics convinced that international crises like the Vietnam War
could not only be modeled but predicted using a computer. His
mentor and graduate thesis advisor was Richard Beal, who had
designed computer systems for the CIA and who later went on to
become President Reagan's special assistant for national security
affairs before his death in 1984.
By 1980 Blais's eclectic resume was attracting job offers in the
southern California area, and he eventually went to work for
Logicon, a Torrance company heavily involved in AI work for the
government's military and intelligence arms. He was awarded a
top-secret clearance and assigned to the SAFE project-the new
computer-controlled nerve center of the CIA. Eventually Logicon
moved the Blais family back east so that Pierre could be minutes
from CIA headquarters. He was getting paid nearly $40,000 a year
to solve the most exhilarating technical problems he had ever
tackled.
But then the nagging doubts began. In November of 1984, Blais
helped organize a conference of Vietnam vets against the CIA
mining of Nicaragua's harbor, and made several speeches. Mean-
while, at work, the projects crossing his desk struck him as crazier
and crazier-including one two-inch-thick proposal to develop a
system for reading another person's mind by recording and
analyzing his or her every blink and twitch. He remembers
buttonholing the proposal's author and saying, "This is immoral, it's
1984 stuff."
"Don't worry, it's just for using on the Russians, in things like
arms negotiations," his colleague responded. When Blais shot back,
"C'mon, what makes you think the CIA has any compunction about
using this stuff on its own people?" the coworker clammed up.
In December of 1984, Blais was summoned by the CIA for a
polygraph test. He claims he answered every question truthfully,
sure there was nothing damning about his background-an arrest
for minor theft in 1964, the selling of some hashish in 1971-that
the CIA didn't already know after clearing him two years before.
One month later, news came that his clearances had been yanked.
For the next eight months he fought the agency, demanding to know
the specific reasons for its decision. Blair claims that in their last
telephone conversation, the CIA security officer handling his case
-a person he never met and whom he knew only by pseudonym
-candidly admitted that the clearances were pulled because Blais
was too much "a person of his own mind."
CIA representative Kathy Pherson would not confirm or deny
Blais's claims, saying, "The agency does not comment on the
security process or who has what clearances."
Blais says that by the end of the tussle he didn't care about the
clearances anymore, so disillusioned was he with defense contract-
ing. Shortly afterwards, he resigned from Logicon.
By last February, Blais had made a clean break from his past
-or so he thought. Logicon and the CIA were behind him. S6 too
was Mormonism. He and his wife had abandoned the church,
disagreeing with what he calls its "authoritarianism and weird
apocalyptic nationalism." There was relief and eagerness in Blais's
voice a year ago as he discussed his fresh start, a $45,000 position at
one of Palo Alto's leading commercial Al firms, Teknowledge,
which caters to industries like oil and finance much more than to the
Department of Defense. Now his days were filled with the task of
mastering an advanced programming language called LISP. He had
started a discussion group on science policy issues like the Strategic
Defense Initiative (Star Wars) and found over a dozen coworkers
open and interested.
What Blais couldn't foresee at the time was the extent to which
Pentagon money had penetrated even commercial Al companies.
Nor could he prophesy that he would become deeply disillusioned
with the field, and that six months later he would be out of a job at
Teknowledge-and possibly out of artificial intelligence altogether.
Just what is artificial intelligence? After 30 years of research,
specialists in the field are still asking that question. Stanford
computer science professor John McCarthy coined the term in
1956 to describe the first stabs at duplicating human thought
with a computer, including a working program that proved theorems
in logic. In 1957 two other Al pioneers, Herbert Simon and Allen
Newell, predicted that within a decade a computer would be the
world chess champion, discover and prove an important mathemati-
cal theorem and compose beautiful music. None of those predictions
has happened yet, but the field keeps expanding.
Among Bay Area firms claiming to be in the Al business are
those trying to develop machines that recognize human speech, those
attempting computers that can be programmed in plain English
rather than mathematical languages and those working on robots
that "see" or "feel" and react accordingly.
Teknowledge and rivals like Mountain View-based Intellicorp
build so-called "expert systems," perhaps the most tried and proven
examples of Al, if not the most sophisticated. One of their
knowledge engineers spends a lot of time with, for example, a
hotshot mechanic, observing and charting his or her decision-making
process, then programs that expertise into a computer. Less
experienced mechanics can turn to the computer for advice when
they are stumped. Stanford has developed a similar, far from
perfect, system for, diagnosing infectious diseases. All these endeav-
ors remain at such an experimental stage that a joke is making the
rounds among computer scientists: "If it works. it's not Al."
The Department of Defense has invested too much in artificial
intelligence research and development to find that one funny. Since
Continued
Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
 Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
the early 1960s the department-specifically its Defense Advanced
Research Projects Agency (DARPA)-has been the main source of
funding for Al explorers, and by the early 1980s it was giving them
between S20 million and $25 million a year, according to the
congressional Office of Technology Assessment. Most of that went
into university labs for very theoretical research, with the rare
real-life product usually something no more threatening than a chess
program. The Pentagon signed the checks, but gave little direction.
That picture changed radically in 1983 when the Reagan
Administration marshalled its Strategic Computing Initiative (SCI)
-a $600 million, five-year crash program to provide the military
with "intelligent" weapons. Suddenly Al research became much
more focused. The Air Force is to receive an electronic "pilot's
associate" able to converse in English and help with navigation and
picking targets. The Army is slated for an "autonomous land
vehicle"-in essence, a tank that drives itself. The Navy and Army
are each to have a "battle management system," which will keep
electronic tabs on friendly and hostile forces alike and give strategy
advice to those in command.
All of these projects not only represent a quantum leap forward
in Al programming, they require computer hardware many times
faster than today's on which to run it, so SCI has earmarked millions
for new computer-chip technology as well. In 1986, DARPA
awarded more than $140 million to SCI, and if anything, the
program is gaining momentum as it passes the midway point of its
first five-year phase. Harvey Newquist, editor of Al Trends, echoes
the consensus among Al watchers when he says "$600 million will
only be a down payment on billions more to follow if SCI is to ever
meet its goals."
Meanwhile, specific branches of the military are nurturing their
own pet artificial intelligence projects outside of SCI. The Army is
particularly enamored of Al, judging by a recently circulated sheaf
of proposal requests. It wants, for example, "terminal homing
munitions... intelligent projectiles and missiles which will search
for stationary and moving armored vehicles... Since the decision of
which target to attack is made on board the weapon, the THM will
approach human intelligence in this area."
It also wants AI-equipped robots for use in "NBC environ-
ments;" that is, areas contaminated by nuclear, biological and/or
chemical weapons. The U.S. Army Quartermaster School in Fort
Lee, Virginia, has put in a request for one that would roll through a
body-strewn battle site, "detect and identify NBC contaminants,
decontaminate human remains, inter remains and refill and mark
graves." If that seems a rather bizarre desire, it is shared by more
than one visionary in the ranks. In a 1982 Army War College report.
Lt. Col. Dennis Crumley forecasts that one day, "When evacuating
the dead, [robotic vehicles] could be loaded and, by merely activating
a switch, dispatched to the nearest mortuary."
And then there is SCI's much richer cousin, SDI, the Strategic
Defense Initiative. Star Wars research is projected to zoom to $10
billion a year by 1991-more than the cost of the MX missile or B-I
bomber in their peak years of production. Just how much of that will
go to Al is still uncertain, but even 15 percent would nearly double
the present rate of Al military funding, which this year will top $ 170
million, according to congressional estimates. By contrast, the next
closest federal funding source, the National Science Foundation, has
about six million Al dollars to play with.
Ask Joel Shurkin of Stanford University News Service how
much money the Pentagon pours into his university's Al projects and
he just whistles. "We have three robotics labs alone," says Shurkin,
who writes press releases about the latest Al developments on
campus. Shurkin estimates that Stanford's Al projects receive
"about S5 million" in defense money. More precise figures are hard
to get because many departments, from electrical engineering to
philosophy, are conducting Al work, and few are willing to give a
funding breakdown.
Stanford reflects the general trend in computer science research,
according to Al observers. Nationally, even counting big contribu-
13
tions from the IBMs and Hewlett-Packards of the commercial
world, "most academic computer science research is now directed by
military agencies," writes UC Berkeley computer science professor
Clark Thompson in a report published in the July issue of
Communications of the ACM, a technical journal of the Association
for Computing Machinery. By other rough estimates, three-quarters
of all the nation's computer research is now being funded and
developed for the military.
For those who dream of creating thinking machines, but find
thinking weapons a nightmare, the military's newly aggressive
Al program creates wrenching dilemmas. Lucy Suchman is an
anthropologist who studies how people interact with computers
at Xerox's Palo Alto Research Center, an Al think tank nestled
among rolling grassy hills and BMW-filled parking lots. It has
always been against her principles to accept military money,
Suchman says, but these days she funds herself helping with a partly
DARPA-funded project. Even though the work is so abstract as to
have no immediate weapons use, Suchman says it bothers her just
the same. When it comes to government spending, "we have gone
way beyond our real requirements in terms of national security and
defense, and we are sorely neglecting human services," she says. "I
don't want to legitimize the current structure of things by saying,
'It's OK, I'll take that money.'"
Suchman says it makes her sad to see very unwarlike colleagues
"organizing their work around military applications" in order to land
grants. But she understands the pressures. To receive National
Science Foundation funding, she says, "you have to write an
enormous grant proposal, which is subject to very stringent peer
review, and at the end of it you get a couple hundred thousand
dollars." At money-laden DARPA, by contrast. "a lot of it has to do
with personal contacts. There's no peer review, no clear set of
requirements-just describe what you're doing, suggest its relevance
and capture the imagination of someone in the Pentagon."
That much cash swirling around has created a powerful vortex,
pulling at all corners of the Al world. indicates the Office of
Technology Assessment's Fred Weingarten. who is directing a study
on the state of Al research: "We're finding a lot of Al designers who
say they can't do long-term research anymore; if they want to get
funded they have to work on short-term military applications." And
the new push for intelligent weapons, those designers tell Weingar-
ten, means "there's a lot of untapped potential that we're going to be
sorry we didn't explore. Maybe we'll be sorry in the marketplace
because other countries may develop key civilian applications first.
Maybe we'll be sorry from a social perspective because we've lost
opportunities to improve our education or health system."
Brian McCune, cofounder of Mountain View-based Advanced
Decision Systems, is one Al whiz who read the writing on the wall
and is prospering because of it. When I talked to him last summer,
he bragged that his seven-year-old company had just snagged a
multimillion-dollar contract large enough to move the firm from
seventeenth to tenth on the list of Bay Area Star Wars contractors.
Advanced Decision Systems is building SDI's "command and
control decision aid test environment," McCune claims. When the
missiles start flying, he says, "somehow you need to boil down
everything that is going on in a globally distributed computer system
for a few human decision makers." And that's a job for Al,
according to McCune.
Advanced Decisions System, which according to McCune has
grown 50 percent over each of the last three years and in 1986
grossed over $10 million, is also hard at work on the electronic brain
for the Army's autonomous land vehicle along with several other
SCI contracts. "We're the largest company dedicated to government
applications in Al," states McCune, who was well prepared as a
student. "A lot of my research at Stanford had been paid for by
DARPA, so I was already hooked into that network. If you want to
be in the research area at all, then you want to work on hard,
state-of-the-art problems. And defense applications are the hardest,
Continued
Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
 Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
most stimulating of all."
Meanwhile, other Al designers who read the writing on the wall,
and don't at all like what they read, are beating a path to the Palo
Alto headquarters of Computer Professionals for Social Responsibil-
ity, a nonprofit organization with chapters in other high-tech centers
like Seattle and Boston. Executive director Gary Chapman says that
a disproportionately high number of new members in the 280-strong
local chapter are Al experts. Maybe that has something to do with
the average Al expert's background, which commonly includes
philosophy, psychology and other social sciences as well as hard-core
"hacking." Al designers are paid to be introspective-like poets and
psychotherapists, they spend a lot of time trying to figure out how
and why people think the way they do.
Behind the moral musings, though, are some hard technical
doubts. A[ expert systems work reasonably well for car fixing, tax
accounting and pill prescribing because those are narrow, well
understood and predictable endeavors. But the chaotic world of war
fighting is quite a different matter. "Our concern is that increased
reliance on artificial intelligence and automated decision making in
critical military situations, rather than bringing greater security,
leads in an extremely dangerous direction," wrote computer scientist
Severo Ornstein, Lucy Suchman and Stanford computer science
professor Brian C. Smith recently in the Bulletin of Atomic
Scientists. Because firsthand experience teaches that no computer
system is perfect, least of all Al versions, "we argue against using
them for decision making in situations of potentially devastating
consequence."
Chapman summons another Frankenstein-like apparition of A[
run amok. He reaches in his desk and pulls out a public relations
photo of Grumman Corporation's experimental &obotic Ranger.
Atop four dune buggyesque wheels sit two bazookalike barrels, and
atop these is mounted a video glass eye that, according to the
caption, "provides autonomous target acquisition" and "IFF (identi-
fication friend or foe)." Chapman's translation: the machine is
supposed to decide on its own who the enemy is, then blow the target
away. Grumman representatives point out that a human monitor,
controlling the Ranger with a three-kilometer-long fiber optic
thread, would have the option of overriding the robot's decision to
shoot.
Chapman says that because such "killer robots" would remove
the human component in war making, they constitute the next
crisis in scientific ethics, on a par in historical -significance with the
development of nuclear weapons." He believes these types of
weapons are banned under Geneva Convention rules. After consult-
ing with experts in international law-including retired Gen. Telford
Taylor, who was chief U.S. prosecutor at the Nuremburg trials
-Chapman feels that his case is strong enough to take to the United
Nations and World Court, and he plans to mount a public campaign
soon.
Other battles against the Pentagon's Al schemes are being
waged quietly out of sight, within the walls of Al companies
themselves. A 26-year-old knowledge engineer at one of the leading
Peninsula Al firms who requests anonymity says he was unnerved to
hear that his company recently landed a S1.5 million piece of the
Army's computerized battle management system. "It's a weapon to
coordinate killing on a large scale," says the Al whiz, who claims
most of his colleagues don't want to work on the project any more
than he does.
Recently he and fifteen other employees met with higher
management and proposed some new ground rules: From now on,
before proposals are submitted to the Pentagon, all employees must
be informed and have some say in the matter. No more Al designers
should be given military security clearances. Finally, before any
person decides whether to work on a military project, he or she must
hear arguments from their coworkers. The young knowledge
engineer thinks there is a good chance the requests will be approved.
"After all, we're highly trained specialists, so we have a little more
bargaining power than the Watsonville canners."
The former philosophy major assesses the stakes this way: "We
can basically be really productive people or we can piss away our
lives building systems that at best will not be used and at worst will
add considerably to the misery of the world."
Unlike those A[ designers who grudgingly take military money
under certain conditions, Pierre Blais is a purist. And he has
paid for that.
By last summer, when Blais and I spoke again, his
enthusiasm about his new job at Teknowledge had turned to disgust.
He had never really fit in at Teknowledge, he said. The work was
challenging, but so removed. He was too preoccupied with the war in
Central America and. increasingly, with the military's grand plan
for artificial intelligence. When Blais considers the promise and peril
of the Al field today, he worries that "we're like the unquestioning
German scientists under Hitler. Whatever destructive technology
you create for the military is eventually going to be used-and that
goes for artificial intelligence. I'd already seen how we used
technology in Vietnam-agent orange and napalm." At the end of
July he was fired for unsatisfactory performance, a charge with
which he has no quarrel. Blais says he never was able to throw his all
into the technical tasks expected of him.
The first blow came when he was taken aside one day and asked,
"You know about the peace clause, don't you?" Blais said he was
aware of a provision in the company charter stating that weapons
contracts were to be discussed openly and that no one could be
required to work on one. It was one of the reasons he was attracted to
Teknowledge in the first place. "Yes, well don't broadcast it to
outsiders," Blais was told. "You know, we would lose government
contracts if they found out about this." "I just shook my head in
disbelief." Blais says.
Blais knew that Teknowledge had a small subsidiary in Thou-
sand Oaks that did nothing but military and intelligence contracting.
That didn't bother him, because as he understood it, the main
operation in Palo Alto was overwhelmingly geared toward commer-
cial applications. But in the spring. Blais learned of Teknowledge's
latest marketing triumph. "I found out we had done a dog and pony
show in front of a conference that was totally dedicated to building
Al systems for the military, and that we had won the highest award
for impressing the customer. Some coworkers I know were rather
horrified to learn we had such a symbiotic relationship with the
Defense Department." As it turns out, theee of Teknowledge's
leading experts were busy at work on a Pentagon-funded battle
management system component.
Blais says his hands never again felt clean working on Teknow-
ledge projects.
All the while, a gulf was growing between Blais and the leading
Al designers at work, a group he calls the "tekkies." Most were
single, at least ten years younger than Blais, and their abilities with a
computer were unquestionably dazzling. "They were the best and
the brightest, but so utilitarian. They seemed to have no conscious-
ness about the world outside." When Blais would want to chat about
politics, philosophy, anything but Al, the tekkies would give him the
distinct impression that he was wasting his time, he says.
That was a criticism to which he felt vulnerable. Immersed in a
strict and demanding course on Teknowledge's brand of Al, Blais
often felt on the verge of drowning. He had passed the first set of
Continued
Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
 Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9
S^
evaluations, but with no gold stars. Some tekkies implied he might
not have the right stuff. "One told me I had trouble forming mental
models," Blais says.
The crash came June 26, the day after Congress approved $ 100
million in military aid to the contras. Blais arrived early, flipped on
his terminal, tapped a message into the electronic mail system and
routed it to every Teknowledge employee. It said: "This is a day that
will live in infamy in the history of this country."
"Everybody yawned," Blais laughs acidly. "1 got one angry
message from a sales rep in Michigan who called me an appeaser
and compared me to Neville Chamberlain."
Blais next ventured out of his cubicle and started ranting at a
coworker whom he knew supported the contra aid. "I said, 'If you're
so sure, why don't you put your bed where your mouth is? Why
don't you grab an M-l6 and shoot a few people? Why don't you do
that instead of retreating into your moral cowardice? Why don't you
do that instead of advocating that our money be used for others to do
the killing for us?"'
After that, says Blais, he began to be assigned less interesting
work. By the end "it was something a twelve-year-old could do." On
July 31 he was terminated with two months severance pay.
Teknowledge spokesman Mike Ayers declines to comment on Blais's
job performance or explain why he was eventually let go, calling it a
private matter. Exactly a week later Blais joined with other northern
California Vietnam veterans in a protest against U.S. policy in
Central America, returning the Bronze Star he had won in Vietnam.
Nowadays Blais divides his time between organizing more
protests by Vietnam vets and making the rounds in his double-knit
suit in search of another job in high tech. He doubts he'll find a solid
niche in Al--everywhere he has interviewed so far is receiving
DARPA funding, and that disqualifies them as far as he is
concerned.
In the meantime, it's hard to call what he does for Sapiens
Software Corporation a job. Blais has worked for the Santa Cruz
start-up company without pay-on maybe money-for two months.
Twenty-eight-year-old Katherine Kelly, the firm's blond, beautiful
and savvy president, turns down job offers from big Silicon Valley
companies often. Vice president John Hare, 31, bespectacled and
elfin, is a software master who claims he has found a way to make a
souped up IBM PC design and run Al expert systems. A steady flow
of UC Santa Cruz grad students grind out nuts and bolts
programming. There is also a striped cat, which sleeps here and
there about the homey cottage that serves as corporate headquarters.
Blais drops in, when time and finances permit, to fine tune
programs and teach Hare what he knows about LISP, the Al
programming language. In return Hare teaches Blais about his
specialty, C, a powerful language for programming microcomputers
like the PC. Blais says one company practice keeps him coming
back, pay or no pay: whenever a request for a Sapiens product
arrives in the mail from a Defense Department-connected buyer, it is
swiftly filed away-the company does not accept military orders or
contracts. "We created this company for brilliant minds who don't
want to work for the military," says Kelly, who founded a Santa
Cruz chapter of Computer Professionals for Social Responsibility.
"We're also making a product that is marketable," she quickly adds.
Blais thinks she may be right "The idea is sound, even if there
are some big hurdles to clear before all the bugs are out"-but he
doesn't know if he'll be around when all the red ink turns to black.
Because his own finances are already shot, he is considering moving
to New Zealand, Japan or "some other country where they need
high-tech people, but are more serious about peace." Or he may
jettison high tech altogether, and seek a teaching job in the field of
his master's degree, Asian studies.
It would be high tech's loss, John Hare says: "Pierre knows
computer languages, he's been around the block. When I started
teaching him C, he learned it two orders of magnitude faster than
the average beginner [or a hundred times faster, in layperson's
language]. He's damn competent."
Damn competent, but does his obsession
with the military's agenda get in the way of
his talent? Some people might want to write
Blais off as one more traumatized Vietnam
vet. Says Hare, "Some vets are more shaken
by the experience than others. Pierre has
about the healthiest reaction you can imag-
ine. He became politically active, and he's
really dedicated to peace. The fact that he's a
Vietnam vet makes him an intense person.
Fine. Intense people are my stock in trade."
Blais himself says, "If so-called post-
Vietnam traumatic stress syndrome is moral
pain, then I have it." And he's all but given
up hoping for a cure that would also allow
him to be at the forefront of artificial
intelligence.
"All the time I've worked in the computer
field, it's been for some sort of military
project," he says, standing on the wooden
porch of Sapiens' headquarters. "Seems like
you can't get away from it. Today I just want
to cut the umbilical cord." 0
Postscript: As this story went to press. Sapi-
ens received a small but life-saving infusion
of investment money. allowing the company
to move into new. more upscale quarters.
Blais. for the time being, is a full-time
employee. making one-third less than he did
at Teknowledge. Ironically. his job is subsi-
dized in part by a training grant from the
Veterans Administration
Declassified in Part - Sanitized Copy Approved for Release 2011/12/21: CIA-RDP90-00965R000100260007-9