XPost: alt.society.liberalism, alt.privacy.anon-server, alt.discrimination
ONE AFTERNOON IN late November of last year, Timnit Gebru was
sitting on the couch in her San Francisco Bay Area home, crying.
Gebru, a researcher at Google, had just clicked out of a last-
minute video meeting with an executive named Megan Kacholia, who
had issued a jarring command. Gebru was the coleader of a group
at the company that studies the social and ethical ramifications
of artificial intelligence, and Kacholia had ordered Gebru to
retract her latest research paper—or else remove her name from
its list of authors, along with those of several other members
of her team.
The paper in question was, in Gebru’s mind, pretty
unobjectionable. It surveyed the known pitfalls of so-called
large language models, a type of AI software—most famously
exemplified by a system called GPT-3—that was stoking excitement
in the tech industry. Google’s own version of the technology was
now helping to power the company’s search engine. Jeff Dean,
Google’s revered head of research, had encouraged Gebru to think
about the approach’s possible downsides. The paper had sailed
through the company’s internal review process and had been
submitted to a prominent conference. But Kacholia now said that
a group of product leaders and others inside the company had
deemed the work unacceptable, Gebru recalls. Kacholia was vague
about their objections but gave Gebru a week to act. Her firm
deadline was the day after Thanksgiving.
Gebru’s distress turned to anger as that date drew closer and
the situation turned weirder. Kacholia gave Gebru’s manager,
Samy Bengio, a document listing the paper’s supposed flaws, but
told him not to send it to Gebru, only to read it to her. On
Thanksgiving Day, Gebru skipped some festivities with her family
to hear Bengio’s recital. According to Gebru’s recollection and
contemporaneous notes, the document didn’t offer specific edits
but complained that the paper handled topics “casually” and
painted too bleak a picture of the new technology. It also
claimed that all of Google’s uses of large language models were
“engineered to avoid” the pitfalls that the paper described.
Gebru spent Thanksgiving writing a six-page response, explaining
her perspective on the paper and asking for guidance on how it
might be revised instead of quashed. She titled her reply
“Addressing Feedback from the Ether at Google,” because she
still didn’t know who had set her Kafkaesque ordeal in motion,
and sent it to Kacholia the next day.
On Saturday, Gebru set out on a preplanned cross-country road
trip. She had reached New Mexico by Monday, when Kacholia
emailed to ask for confirmation that the paper would either be
withdrawn or cleansed of its Google affiliations. Gebru tweeted
a cryptic reproach of “censorship and intimidation” against AI
ethics researchers. Then, on Tuesday, she fired off two emails:
one that sought to end the dispute, and another that escalated
it beyond her wildest imaginings.
The first was addressed to Kacholia and offered her a deal:
Gebru would remove herself from the paper if Google provided an
account of who had reviewed the work and how, and established a
more transparent review process for future research. If those
conditions weren’t met, Gebru wrote, she would leave Google once
she’d had time to make sure her team wouldn’t be too
destabilized. The second email showed less corporate diplomacy.
Addressed to a listserv for women who worked in Google Brain,
the company’s most prominent AI lab and home to Gebru’s Ethical
AI team, it accused the company of “silencing marginalized
voices” and dismissed Google’s internal diversity programs as a
waste of time.
Relaxing in an Airbnb in Austin, Texas, the following night,
Gebru received a message with a ?? from one of her direct
reports: “You resigned??” In her personal inbox she then found
an email from Kacholia, rejecting Gebru’s offer and casting her
out of Google. “We cannot agree as you are requesting,” Kacholia
wrote. “The end of your employment should happen faster than
your email reflects.” Parts of Gebru’s email to the listserv,
she went on, had shown “behavior inconsistent with the
expectations of a Google manager.” Gebru tweeted that she had
been fired. Google maintained—and still does—that she resigned.
FEATURED VIDEO
Bar Owner Builds an Alarm That Stops You From Forgetting Your
Credit Card
Most Popular
FBI building
SECURITY
The FBI's Anom Stunt Rattles the Encryption Debate
LILY HAY NEWMAN
Screenshot of Vivaldi search
GEAR
You're Probably Not Using the Web's Best Browser
SCOTT GILBERTSON
BACKCHANNEL
How Roblox Became a Playground for Virtual Fascists
CECILIA D'ANASTASIO
Macbook Pro with MacOS Monterey
GEAR
Apple Starts Leaving Intel Macs Behind in MacOS Monterey
BOONE ASHWORTH
Gebru’s tweet lit the fuse on a controversy that quickly
inflamed Google. The company has been dogged in recent years by
accusations from employees that it mistreats women and people of
color, and from lawmakers that it wields unhealthy technological
and economic power. Now Google had expelled a Black woman who
was a prominent advocate for more diversity in tech, and who was
seen as an important internal voice for greater restraint in the helter-skelter race to develop and deploy AI. One Google
machine-learning researcher who had followed Gebru’s writing and
work on diversity felt the news of her departure like a punch to
the gut. “It was like, oh, maybe things aren’t going to change
so easily,” says the employee, who asked to remain anonymous
because they were not authorized to speak by Google management.
Dean sent out a message urging Googlers to ignore Gebru’s call
to disengage from corporate diversity exercises; Gebru’s paper
had been subpar, he said, and she and her collaborators had not
followed the proper approval process. In turn, Gebru claimed in
tweets and interviews that she’d been felled by a toxic cocktail
of racism, sexism, and censorship. Sympathy for Gebru’s account
grew as the disputed paper circulated like samizdat among AI
researchers, many of whom found it neither controversial nor
particularly remarkable. Thousands of Googlers and outside AI
experts signed a public letter castigating the company.
But Google seemed to double down. Margaret Mitchell, the other
coleader of the Ethical AI team and a prominent researcher in
her own right, was among the hardest hit by Gebru’s ouster. The
two had been a professional and emotional tag team, building up
their group—which was one of several that worked on what Google
called “responsible AI”—while parrying the sexist and racist
tendencies they saw at large in the company’s culture. Confident
that those same forces had played a role in Gebru’s downfall,
Mitchell wrote an automated script to retrieve notes she’d kept
in her corporate Gmail account that documented allegedly
discriminatory incidents, according to sources inside Google. On
January 20, Google said Mitchell had triggered an internal
security system and had been suspended. On February 19, she was
fired, with Google stating that it had found “multiple
violations of our code of conduct, as well as of our security
policies, which included exfiltration of confidential, business-
sensitive documents.”
Google had now fully decapitated its own Ethical AI research
group. The long, spectacular fallout from that Thanksgiving
ultimatum to Gebru left countless bystanders wondering: Had one
paper really precipitated all of these events?
The story of what actually happened in the lead-up to Gebru’s
exit from Google reveals a more tortured and complex backdrop.
It’s the tale of a gifted engineer who was swept up in the AI
revolution before she became one of its biggest critics, a
refugee who worked her way to the center of the tech industry
and became determined to reform it. It’s also about a
company—the world’s fifth largest—trying to regain its
equilibrium after four years of scandals, controversies, and
mutinies, but doing so in ways that unbalanced the ship even
further.
Beyond Google, the fate of Timnit Gebru lays bare something even
larger: the tensions inherent in an industry’s efforts to
research the downsides of its favorite technology. In
traditional sectors such as chemicals or mining, researchers who
study toxicity or pollution on the corporate dime are viewed
skeptically by independent experts. But in the young realm of
people studying the potential harms of AI, corporate researchers
are central.
Gebru’s career mirrored the rapid rise of AI fairness research,
and also some of its paradoxes. Almost as soon as the field
sprang up, it quickly attracted eager support from giants like
Google, which sponsored conferences, handed out grants, and
hired the domain’s most prominent experts. Now Gebru’s sudden
ejection made her and others wonder if this research, in its
domesticated form, had always been doomed to a short leash. To
researchers, it sent a dangerous message: AI is largely
unregulated and only getting more powerful and ubiquitous, and
insiders who are forthright in studying its social harms do so
at the risk of exile.
IN APRIL 1998, two Stanford grad students named Larry Page and
Sergey Brin presented an algorithm called PageRank at a
conference in Australia. A month later, war broke out between
Ethiopia and Eritrea, setting off a two-year border conflict
that left tens of thousands dead. The first event set up
Google’s dominance of the internet. The second set 15-year-old
Timnit Gebru on a path toward working for the future megacorp.
At the time, Gebru lived with her mother, an economist, in the
Ethiopian capital of Addis Ababa. Her father, an electrical
engineer with a PhD, had died when she was small. Gebru enjoyed
school and hanging out in cafés when she and her friends could
scrape together enough pocket money. But the war changed all
that. Gebru’s family was Eritrean, and some of her relatives
were being deported to Eritrea and conscripted to fight against
the country they had made their home.
Gebru’s mother had a visa for the United States, where Gebru’s
older sisters, engineers like their father, had lived for years.
But when Gebru applied for a visa, she was denied. So she went
to Ireland instead, joining one of her sisters, who was there
temporarily for work, while her mother went to America alone.
Some of her teachers, Gebru found, seemed unable or unwilling to
accept that an African refugee might be a top student in math
and science.
Reaching Ireland may have saved Gebru’s life, but it also
shattered it. She called her mother and begged to be sent back
to Ethiopia. “I don’t care if it’s safe or not. I can’t live
here,” she said. Her new school, the culture, even the weather
were alienating. Addis Ababa’s rainy season is staccato, with
heavy downpours interspersed by sunshine. In Ireland, rain fell
steadily for a week. As she took on the teenage challenges of
new classes and bullying, larger concerns pressed down. “Am I
going to be reunited with my family? What happens if the
paperwork doesn’t work out?” she recalls thinking. “I felt
unwanted.”
The next year, Gebru was approved to come to the US as a
refugee. She reunited with her mother in Somerville,
Massachusetts, a predominantly white suburb of Boston, where
she enrolled in the local public high school—and a crash course
in American racism.
Some of her teachers, Gebru found, seemed unable or unwilling to
accept that an African refugee might be a top student in math
and science. Other white Americans saw fit to confide in her
their belief that African immigrants worked harder than African
Americans, whom they saw as lazy. History class told an
uplifting story about the Civil Rights Movement resolving
America’s racial divisions, but that tale rang hollow. “I
thought that cannot be true, because I’m seeing it in the
school,” Gebru says.
Piano lessons helped provide a space where she could breathe.
Gebru also coped by turning to math, physics, and her family.
She enjoyed technical work, not just for its beauty but because
it was a realm disconnected from personal politics or worries
about the war back home. That compartmentalization became part
of Gebru’s way of navigating the world. “What I had under my
control was that I could go to class and focus on the work,” she
says.
Gebru’s focus paid off. In September 2001 she enrolled at
Stanford. Naturally, she chose the family major, electrical
engineering, and before long her trajectory began to embody the
Silicon Valley archetype of the immigrant trailblazer. For a
course during her junior year, Gebru built an experimental
electronic piano key, helping her win an internship at Apple
making audio circuitry for Mac computers and other products. The
next year she went to work for the company full-time while
continuing her studies at Stanford.
At Apple, Gebru thrived. When Niel Warren, her manager, needed
someone to dig into delta-sigma modulators, a class of analog-to-
digital converters, Gebru volunteered, investigating whether the
technology would work in the iPhone. “As an electrical engineer
she was fearless,” Warren says. He found his new hardware
hotshot to be well liked, always ready with a hug, and
determined outside of work too. In 2008, Gebru withdrew from one
of her classes because she was devoting so much time to
canvassing for Barack Obama in Nevada and Colorado, where many
doors were slammed in her face.
As Gebru learned more about the guts of gadgets like the iPhone,
she became more interested in the fundamental physics of their
components—and soon her interests wandered even further, beyond
the confines of electrical engineering. By 2011, she was
embarking on a PhD at Stanford, drifting among classes and
searching for a new direction. She found it in computer vision,
the art of making software that can interpret images.
Unbeknownst to her, Gebru now stood on the cusp of a revolution
that would transform the tech industry in ways she would later
criticize. One of Gebru’s favorite classes involved creating
code that could detect human figures in photos. “I wasn’t
thinking about surveillance,” Gebru says. “I just found it
technically interesting.”
In 2013 she joined the lab of Fei-Fei Li, a computer vision
specialist who had helped spur the tech industry’s obsession
with AI, and who would later work for a time at Google. Li had
created a project called ImageNet that paid contractors small
sums to tag a billion images scraped from the web with
descriptions of their contents—cat, coffee cup, cello. The final
database, some 15 million images, helped to reinvent machine
learning, an AI technique that involves training software to get
better at performing a task by feeding it examples of correct
answers. Li’s work demonstrated that an approach known as deep
learning, fueled by a large collection of training data and
powerful computer chips, could produce much more accurate
machine-vision technology than prior methods had yielded.
Li wanted to use deep learning to give computers a more fine-
grained understanding of the world. Two of her students had
scraped 50 million images from Google Street View, planning to
train a neural network to spot cars and identify their make and
model. But they began wondering about other applications they
might build on top of that capability. If you drew correlations
between census data and the cars visible on a street, could that
provide a way to estimate the demographic or economic
characteristics of any neighborhood, just from pictures?
Gebru spent the next few years showing that, to a certain level
of accuracy, the answer was yes. She and her collaborators used
online contractors and car experts recruited on Craigslist to
identify the make and model of 70,000 cars in a sample of Street
View images. The annotated pictures provided the training data
needed for deep-learning algorithms to figure out how to
identify cars in new images. Then they processed the full Street
View collection and identified 22 million cars in photos from
200 US cities. When Gebru correlated those observations with
census and crime data, her results showed that more pickup
trucks and VWs indicated more white residents, more Buicks and
Oldsmobiles indicated more Black ones, and more vans
corresponded to higher crime.
This demonstration of AI’s power positioned Gebru for a
lucrative career in Silicon Valley. Deep learning was all the
rage, powering the industry’s latest products (smart speakers)
and its future aspirations (self-driving cars). Companies were
spending millions to acquire deep-learning technology and
talent, and Google was placing some of the biggest bets of all.
Its subsidiary DeepMind had recently celebrated the victory of
its machine-learning bot over a human world champion at Go, a
moment that many took to symbolize the future relationship
between humans and technology.
Gebru’s project fit in with what was becoming the industry’s new
philosophy: Algorithms would soon automate away any problem, no
matter how messy. But as Gebru got closer to graduation, the
boundary she had established between her technical work and her
personal values started to crumble in ways that complicated her
feelings about the algorithmic future.
“I’m not worried about machines taking over the world,” Gebru
wrote. “I’m worried about groupthink, insularity, and arrogance
in the AI community.”
Her ass got fired by the racists at google.
https://www.wired.com/story/google-timnit-gebru-ai-what-really-
happened/
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)