In the wake of the controversial firing of two highly respected AI researchers who authored a paper that Google disliked, Google has now been accused of modifying at least three papers on AI in order to prevent them ‘casting Google technology in a negative light’.

Timnit Gebru, formerly co-lead of Google’s AI Ethics team, was forced out of the company in December after co-authoring a paper that highlighted the potential biases (racial, gender, Western etc) in the AI models that Google uses to power its search engine and other core products. The paper was also critical of the carbon footprint involved in computing the models. Gebru’s co-author Margaret Mitchell was fired more recently, after being asked to retract her name from the published paper. Now another Google researcher, Nicolas Carlini, has accused Google of ‘deeply insidious’ edits to AI research that are akin to ‘Big Brother stepping in’.

So What

This raises really worrying questions about the reshaping and censoring of AI research for a commercial agenda – and points to the dark side of the concentration of so much AI research in a private, unaccountable company. Google is a major source of AI research, given its outsize funding and resources in the space – that dwarfs that of academia. It is also particularly concerning that research investigating ethical  and environmental issues with AI has been targeted. This adds further weight to calls for stronger regulation and oversight of Big Tech.