PERSON
Timnit Gebru
The computer scientist who measured AI’s failures with the discipline of a natural scientist, named the hidden labor and concentrated power behind the systems, and refused to call any of it inevitable—at the cost of her position inside the most powerful AI company on earth.
There is a particular kind of thinker who is most useful precisely when she is most inconvenient, and Timnit Gebru has spent her career being inconvenient to the people who least wanted to hear her. Born in Addis Ababa to Eritrean parents and educated as a refugee who navigated Ireland before obtaining political asylum in America, she came to artificial intelligence through electrical engineering and computer vision, trained at Stanford under Fei-Fei Li. What she brought from that path was not outrage but method: the discipline of measurement, the insistence that a claim about
large language models be documented rather than asserted, and the conviction that the first step toward a fair technology is simply to count what it does to whom. Her 2018 study with Joy Buolamwini, “Gender Shades,” disaggregated the performance of commercial face-recognition systems by the intersection of skin tone and gender and found error rates of under