Thunderbolt in Silicon Valley? Timnit Gebru, a black artificial intelligence researcher at Google, has just been fired after working on a study demonstrating the weaknesses of an AI-based technology that powers Google’s search engine in particular – Mountain View .
What happened ?
Timnit Gebru was born and raised in Ethiopia. He is 37 years old and until recently co-led the ethical artificial intelligence team at Google. She stood out in 2018 when she worked as a researcher at Stanford University to develop a paper highlighting the racist and sexist prejudices of artificial intelligence. In the same year she joined Mountain View with the aim of developing an ethical AI. The researcher often points out the lack of diversity in Silicon Valley and expresses her fears about AI and its deviations, especially in the area of facial recognition.
In the same category
Microsoft introduces AI that can describe images “as well as humans”
Recently, Timnit Gebru and four other Google employees were instructed by their supervisors not to present a study at a conference or to remove their names from it. In an email to several of the company’s employees, Gebru said that its executives wanted to cover up his research and ignored his comments on ethical issues such as the shortage of women hired to join Google. “Your life starts to get worse when you stand up for underrepresented people. You start to anger other leaders. There is no way that other documents or conversations can get anything, ”she wrote. She also states that if Google does not deign to make her statements regarding the above facts, she will step down at some point.
I got fired by @JeffDean for my email to Brain Women and Allies. My company account has been blocked. So I was fired immediately 🙂
– Timnit Gebru (@timnitGebru) December 3, 2020
Shortly after sending it, Timnit Gebru wrote on Twitter that her company email address had been truncated and she was fired by Jeff Dean, head of AI at Google. The latter, however, denies the researcher’s claims and confirms that she made the decision to resign herself and that her work beyond that is not valid. “It was dehumanizing. You may have reasons to stop our research. Most annoying, however, is that they refuse to discuss the why, ”says Dr. Gebru.
Google is more and more in conflict with its employees
Timnit Gebru is a highly respected and committed researcher in the field of AI. Your dismissal is a blemish for Mountain View, which is committed to a culture in which everyone is equal and can express themselves freely. Since the last year, Google employees have been increasingly taking action against their companies. For example, some of them demanded that it be excluded from gay pride because its values are inconsistent with the congregation. Two workers involved in union work were also laid off by Google after they were spied on by the company in a labor law violation, the National Labor Relations Board recalled.
In addition, Google has introduced new procedures to prevent employees from organizing and mobilizing with one another while they are spying on their online activities. The company has also hired business moderators whose job it is to reduce conflict. Timnit Gebru’s layoff is unlikely to improve relationships between the company’s employees and their managers.
The problems identified by the researcher are well known in the technology world. In addition, several studies have shown the racial prejudices of artificial intelligence, which are mainly conceived by white men and also trained on white subjects. While Google has reiterated its desire to address such discrepancies in its systems, the problem lies in the fact that, according to Timnit Gebru, the final decisions are often made by men: “They not only prioritize” hire, but suppress the “minority people their voices, “she said.
Sundar Pichai, CEO of Alphabet, is betting on artificial intelligence and is betting on better regulation of this technology earlier this year. However, Timnit Gebru’s dismissal does not appear to really be in line with this policy.