Google fires second AI ethics researcher amid diversity dispute
What's the story
Google has fired Margaret Mitchell who is the founder and former co-lead of the company's Ethical AI group. Google claimed she was fired for violating its code of conduct and security policies. Google's statement comes after Mitchell announced the news through a tweet. Her ouster comes hot on the heels of similar firing of another Ethical AI co-leader Timnit Gebru.
He said, she said?
Mitchell and Google make wildly divergent claims about the dismissal
Axios, which originally broke the news, cites a source claiming that Mitchell was locked out of her official email account since last month. This was ostensibly done to prevent her from checking official correspondence for evidence backing up Gebru's claim of discrimination and harassment. Google, however, counters that Mitchell's violations "included the exfiltration of confidential business-sensitive documents and private data of other employees".
Ethics under scrutiny
Mitchell's firing comes after a month-long internal investigation
Google's director for AI research, Zoubin Ghahramani, informed Mitchell's team of her firing on Friday. The meeting was called in short notice and scarce explanation was provided about the decision. Meanwhile, speaking to media, Google revealed that an internal investigation against Mitchell was ongoing since January 19. The findings of a team of investigators were adjudicated by a review committee leading to the dismissal.
Contentious research
Another co-founder of Ethical AI team was also fired earlier
There was discontent brewing in Google's Ethical AI arm following the firing of Gebru, who was in the headlines for exposing bias in facial recognition systems. Gebru claimed her firing was punishment for questioning Google's decision to censor papers critical of its AI systems, one of which was a paper authored by Mitchell and Gebru underlining how AI mimicking language could hurt marginalized individuals.
PR stunt gone awry
Ethical AI arm gave Google serious case of buyer's remorse
Google's Ethical AI arm looks like a PR campaign gone awry. The company had hired scientists promising research freedom, but the honeymoon period ended when the researchers began submitting papers critical of Google's own products. Google responded by introducing "sensitive topics" review, seemingly to guard against legal and regulatory trouble stemming from such research papers. Mitchell had raised censorship concerns about the policy.