Google firing Timnit Gebru is a step back for ethical AI - one you should not ignore
Two years after it stopped selling facial recognition tech, the company has seemingly decided to bite the hand that fed it
It was only a few months ago that tech giants stood side by side to condemn in-built racial bias in the field of artificial intelligence (AI), particularly facial recognition. Microsoft, IBM, and Amazon were lauded for their decision to halt or refuse sales of controversial facial recognition software to US police departments, a move considered timely and indicative of a wider movement to clamp down on racial discrimination following the death of George Floyd.
So it’s shocking to hear that Google, a company that once championed the idea of reining in facial recognition technology, has reportedly decided to fire the very person who helped shape the company’s AI ethics.
Timnit Gebru, a Stanford Artificial Intelligence Laboratory alumna, is among the most acclaimed AI researchers in the world and, as the technical co-lead of its Ethical Artificial Intelligence Team, has contributed enormously to Google’s stance on the ethical use of the technology. After all, she is the co-author, along with Joy Buolamwini, of a revolutionary paper that highlighted the discriminatory nature of machine learning algorithms. It was this very research that contributed to Google’s 2018 decision to stop selling general-purpose AI-driven facial recognition technology.
Almost two years to the day later, Google has now decided to unceremoniously cut ties with Gebru.
According to Gebru, the company made the decision to dismiss her after she sent an email to her colleagues about the company’s decision to retract one of her research papers. This came just days after the tech giant was accused of illegally blocking staff from sharing work grievances across a number of its internal communication services.
What’s most shocking is that, ultimately, the disagreement between Gebru and her superiors seems to stem from her actions of “advocating for underrepresented people” – the very thing she was hired to do. Instead of taking the opportunity to listen and negotiate with one of their leading employees, Google reportedly decided to cut Gebru off her corporate account – a crystal clear message that she had been terminated with immediate effect.
This treatment would be seen as appalling in any situation, regardless of the employee’s position or standing. However, it’s hideous that a black female employee who has contributed so much to Google’s social and racial awareness efforts is the target. On top of that, the company has had no qualms in the past about capitalising on Gebru’s expertise, as well as her race. As one fellow Google AI employee put it, the company was “quick to get rid of @timnitGebru after constantly parading her around as an example of how much they care about ethical AI”.
In an email to Google’s employees, AI lead Jeff Dean said the recent paper Gebru submitted “was only shared with a day’s notice before its deadline — we require two weeks for this sort of review — and then instead of awaiting reviewer feedback, it was approved for submission and submitted”.
What might have started as a communication error has now turned into a massive error of judgement and, although the company might not see it this way just yet, it’s obvious that Google needs Gebru more than Gebru needs Google. It needs to feed on her research, ideas, and excellence.
The firing of Gebru comes at the end of a long year during which plenty of corporations made a lot of promises to address racism, restructure their internal policies – anything that would make them as blameless as possible in a climate where there’s growing social awareness of systemic racism.
Google needs to realise its employees are not pawns to be exchanged for “wokeness” points. If one tweet can sum up this situation best, it’s this:
The main image in this article is licensed from TechCrunch under Creative Commons license 2.0.
IT Pro contacted Google for comment for this story, but the company declined to provide a statement.
Managing security risk and compliance in a challenging landscape
How key technology partners grow with your organisationDownload now
Evaluate your order-to-cash process
15 recommended metrics to benchmark your O2C operationsDownload now
AI 360: Hold, fold, or double down?
How AI can benefit your businessDownload now
Getting started with Azure Red Hat OpenShift
A developer’s guide to improving application building and deployment capabilitiesDownload now