Meredith Whittaker’s CV is remarkable: she worked at Google for 13 years, today she is one of the company’s harshest critics – especially when it comes to AI. When she started at Google in 2006, she simply needed the money. She had just completed her bachelor’s degree in literature and rhetoric. “I was broke, I needed work, and Google was the first company that made me an offer,” she said in an interview with the magazine “Republik”.
At Google, she launched several initiatives, including the Google Open Research Group, which aims to solve complex problems together with external scientists and open source activists, or the M-Lab, a globally distributed network measurement system that is the world’s largest source of open data to internet performance. As early as 2013, she criticized the practice of training AI systems with data that discriminates because it comes from the Internet. In 2017 she founded the research institute AI Now with the AI researcher Kate Crawford. It is considered the first academic institute to research the social implications of artificial intelligence.
At the time, Whittaker was still employed at Google. The company supported the founding of the institute. At that time, it simply fitted very well with the image of a cosmopolitan and fair company that also took care of people’s needs. However, when she found out about the Maven project, she wrote a letter to Google CEO Sundar Pichai demanding that they get out of the project. Maven was a military project for the US Department of Defense. 3,000 employees signed the letter. A few weeks later, Google announced that it would not renew its contract with the Department of Defense. In 2018, Whittaker was also one of the initiators of the Google Walkouts, where 20,000 Google employees protested against sexism and racism in the workplace.
Finally, in 2019, she resigned from Google. Since then, she has focused on her academic and consulting work in the context of the AI Now Institute. Since November 2021 she has been an advisory member of the US Trade Commission (FTC). She has been President of Messengers Signal since September 6, 2022.
Ms. Whittaker, you keep emphasizing that machine learning and so-called artificial intelligence are a power issue and that we have to look at and judge them from that perspective. But haven’t we been talking about AI exacerbating inequalities for years – and that without anything changing?
I’ve already seen changes in our debates since 2015. That’s the year the idea for AI Now came about because I was increasingly concerned about what this so-called AI would entail. At that time, AI was a hype that was mainly fueled by the tech industry. The slowly growing critical community was not talking about power at the time. The criticism revolves around big data, i.e. the collection of large amounts of data. Later the bias came into focus, racial biases for example, which you thought you could remove from the AI and then have a nice “pure” technology. We learned that we can’t talk about bias without talking about structural racism in society.
But at the latest now you can no longer avoid the topic of power dynamics. What are you missing from the current discussion?
I miss the fact that we don’t just talk about power in the abstract, but make it clear who benefits here. We must name the companies that develop and use systems, the names of those who use them and benefit from them. It must be made clear that these “accidents,” like the Michigan algorithm that misclassified over 40,000 unemployment benefit claimants as fraudsters, are not tragic isolated incidents. There is a power system behind this.
To home page