Discover more from Last Week in AI

Weekly text and audio summaries of the most interesting AI news, as well as editorials commenting on recent events

Over 20,000 subscribers

AI for job interviews, development of autonomous weapons, and more!

Last Week in AI #104

Last Week in AI

Feb 22, 2021

Image Credits: THERE IS STUDIO via The Washington Post

Mini Briefs

OBJECTIVE OR BIASED

A study conducted by a Bayerischer Rundfunk (a German broadcasting company) found that software that automatically assess job candidates via recorded videos are easily manipulated by factors unrelated to the interviewee. To establish a baseline, real actors were hired to record job interview videos with as little emotion as possible. Then, the experimenters changed the videos in several ways that should not affect assement outcome, including adding framed pictures or bookshelves to the background, letting the interviewee wear glasses and headscarves, and changing the brightness of the video. However, the software gave significantly different results in these modified videos. For example, adding a bookshelf to the background significantly increased the candidates's supposed "conscientiousness" score. In addition, the software doesn't seem to pay attention to the audio in the video at all, as the scores were unchanged when the audio track was swapped with that of a different actor, and when the audio was removed completely. All these suggest that AI-based HR, in addition to raising pointed ethical and privacy concerns, are also not very useful.

Can Computer Algorithms Learn to Fight Wars Ethically?

This report gives an overview of what people in the U.S. military think about developing and deploying autonomous lethal weapons. The military, from commanders to cadets, seem to be broadly aware of the limitations and opaqueness of modern AI software, which is very difficult to test and certify. Still, they feel pressure to push ahead with lethal autonomous weapons due to a fear of missing out. This is a cause of real worry, as AI evangelists within the Pentagon use FOMO as an excuse to deploy immature technology, disregarding serious ethical concerns in the process. One defense official remarks “I fear our lack of keeping up [...] I don’t fear us losing our ethical standards, our moral standards.” It is difficult to see how ethical standards are not being lost when the same official says "it doesn’t make sense to study anything in the era of AI” and misleadingly describes AI as "a living, breathing system."

A poll from December 2020 found that the majority of citizens in 26 of the 28 surveyed countries, including U.S., Russia, and China, opposed the development of AI weapons. But public sentiment doesn't necessarily track with the military's decision making.

The military is going to put AI into its weapons despite debates about morality, [Pentagon’s Joint Artificial Intelligence Center] told me: “We are going to do it. We’re going to do it deliberately. We’re going to follow policy.”

For more on this topic, see our overivew on The Rise of “Killer Robots” and the Race to Restrain Them.

News

Advances & Business

This AI-powered gadget could completely disrupt the ridiculous hearing aid market - How a company called Whisper is using AI to build better hearing aids
I let an AI analyze my Spotify to find unsigned bands I might like - "Andrson is a company with a music-matching algorithm designed to find songs that sound similar to ones you've already heard. Its product is, for now, aimed at making it easier for music labels to find would-be stars, but that's not the only thing it's useful for."
The AI Research Paper Was Real. The "Coauthor" Wasn't - "An MIT professor found his name on two papers with which he had no connection. A different paper listed a fictitious author by the name of "Bill Franks.""
A New Artificial Intelligence Makes Mistakes - on Purpose - "A chess program that learns from human error might be better at working with people or negotiating with them."
How NSF and Amazon Are Collectively Tackling Artificial Intelligence-Based Bias - "The National Science Foundation and Amazon teamed up to fund a second round of research projects aimed at promoting trustworthy artificial intelligence and mitigating bias in systems."
Artificial Neural Nets Finally Yield Clues to How Brains Learn - “The learning algorithm that enables the runaway success of deep neural networks doesn’t work in biological brains, but researchers are finding alternatives that could.”

Concerns & Hype

Who Should Stop Unethical A.I.? - "At artificial-intelligence conferences, researchers are increasingly alarmed by what they see."
Google reshuffles AI team leadership after researcher's controversial departure - "Google is making changes to the leadership of its responsible artificial intelligence efforts months after the contentious departure of prominent Black artificial intelligence researcher Timnit Gebru rattled the company."
The coming conflict over facial recognition - "The arrests and charges in the aftermath of the Jan. 6 Capitol Hill insurrection made clear the power of facial recognition, even as efforts to restrict the technology are growing."
Clearview AI's plan for invasive facial recognition is worse than you think - "The company's lust for dystopia continues."
This AI reads children's emotions as they learn - Worryingly, another startup is working on identifying emotions from images of faces of school children

Analysis & Policy

The Cambrian AI Landscape: NVIDIA, The 800Lb Gorilla - Overview of Nvidia's role in AI hardware.

That's all for this week! If you are not subscribed and liked this, feel free to subscribe below!