Mini Episode: Lessons in AI bias, Facial Recognition Policy and Effects, and Trump‘s Visa Freeze
04:48 Share

Hello and welcome, this is Daniel Bashir here with Skynet, today's Week in AI. This week, it's all about ethics and injustice. At this year's conference on computer vision and pattern recognition, Tinmit Jebru, a well-known AI researcher whose gender shade study exposed racial and gender bias in facial recognition, gave a tutorial on how AI bias goes far beyond the data.

Jan Latun, another famed AI researcher, showed that Jebru's lesson hasn't completely set in even with The Best of Us in a recent comment. VentureBeat reports that Pulse, a controversial computer vision model that claims to generate realistic, high-resolution images of people from a pixelated photo, demonstrates a number of issues, such as upsampling a photo of President Obama into a white man.

But LeCun asserted that it was the biased data alone that could explain biased ML systems. There are plenty of resources and experts that can explain why bias in AI extends beyond data, and others besides Gebru noted these when confronting LeCun about his comment. It seems he did not entirely get the message or engage with Gebru's research.

VentureBeat notes that this is concerning given LeTun's prominent position in the AI community and the powerful message that such leadership from him could send. While AI researchers are still trying to convince some about how ML systems come to be biased, those systems are already affecting real people. The New York Times reports that in the first known case of its kind, a flawed facial recognition system caused the wrongful arrest of an innocent person.

Earlier this year, the Detroit Police Department arrested Robert Williams, an African American male. Mr. Williams was taken to a detention center and held overnight because a facial recognition system misidentified him as another black man in a surveillance video.

And just in case that wasn't enough bad news, we've got more.

On June 22nd, President Trump signed an executive order temporarily halting work visas, including the H-1B visa program for highly skilled workers. The restrictions are intended to last until the end of the year. The blowback this move has gotten is immense, and includes statements from many who came to America on these visas, including a founder of the company Databricks, and plenty of others in the tech industry.

The MIT Technology Review reports that the visa freeze will have another negative effect among the many already discussed, hindering the US and the global competition for AI talent. Recent analysis from think tank Macropolo found that 69% of AI researchers working at US institutions studied outside the US for undergrad,

Two-thirds of grad students in the US's top AI/PhD programs are international, and 80% of those students stay for five years after graduation. While some government officials are pushing ruinous policy, others are making steps in response to the pushback on facial recognition. On June 25th, US Democratic lawmakers introduced the Facial Recognition and Biometric Technology Moratorium Act.

The MIT Technology Review reports that the act would make it illegal for any federal agency or official to acquire, possess, access, or use biometric surveillance technology in the United States.

The act likely represents more hope for progress to activists, whose work has played a large part in IBM's, Microsoft's, and Amazon's recent pullbacks on facial recognition, which themselves likely increased the pressure on lawmakers.

That's all for this week. Thanks so much for listening. If you enjoyed the podcast, be sure to rate and share. If you'd like to hear more news like this, please check out Skynetoday.com, where you can find our weekly news digests with similar articles.