MANILA, Philippines – Data scientist Timnit Gebru warned in 2020 that based on the data sets that fuel artificial intelligence (AI) models, AI will have the power to further deepen the dominance of the white, male, and mostly West-centric way of thinking.
The then-co-leader of Google’s small ethical AI team argued in a paper she published that AI “overrepresent hegemonic viewpoints and encode biases potentially damaging to marginalized populations.”
This then triggered a cascade of events that culminated in her eventual Google exit in 2020. According to Gebru she was fired, but the tech giant claims it was voluntary.
Gebru, along with fellow researchers Emily M. Bender, Angelina McMillan-Major, and Margaret Mitchell, co-wrote a paper titled, “On the dangers of stochastic parrots: Can language models be too big?” and which discussed issues surrounding AI. These included how large language models basically repeat statements of data and add verbal noise to make them seem like intelligent speech.
In a statement on March 23, the authors of the paper also spoke out against the use of their paper in an Elon Musk-backed campaign to pause AI research for six months, citing future risks to mankind.
They pointed out, “Such language that inflates the capabilities of automated systems and anthropomorphizes them, as we note in Stochastic Parrots, deceives people into thinking that there is a sentient being behind the synthetic media. This not only lures people into uncritically trusting the outputs of systems like ChatGPT, but also misattributes agency.”
Their statement also takes to task the tech companies that are developing AI systems without regard for how the development may make current inequities – such as racial and gender biases – worse. Gebru and company said that “we should focus on the very real and very present exploitative practices of the companies claiming to build them, who are rapidly centralizing power and increasing social inequities.”
In a live interview, Maria Ressa and Timnit Gebru will talk about the moral and political questions surrounding AI, as well as ways toward breaking down the biases and inequalities that are built into it.
Rappler+ members will be able to join the private Q&A (Live chat) after the interview at 9 am (Manila time) on Thursday, June 1. If you want to attend, join Rappler+ and gain access to other events like this.
Members get exclusive content and become part of a global community of people who believe in the power of independent journalism. – Rappler.com
How does this make you feel?
