Making AI algorithms show their work, an expansion to Google's AI Ethics research team, and more!

Last Week in AI #116

May 17, 2021

Mini Briefs

Making AI algorithms show their work

One of the main draws of machine learning algorithms is their ability to perform tasks without being explicitly programmed with rules to accomplish those tasks. In the process of learning to perform those tasks, these algorithms notice patterns and develop "rules" of their own, however, and the developers of those algorithms don't always know what those rules are. Cold Spring Harbor Laboratory Assistant Professor Peter Koo, who developed a neural network to look for patterns in RNA that increase a protein's ability to bind to them, wanted to know whether his algorithm was focusing on short sequences of RNA letters or secondary characteristics of RNA strands. Koo and his team developed a method called Global Importance Analysis to test which rules their algorithm generated to make its predictions.

Google Plans to Double AI Ethics Research Staff

After a number of departures and re-organizations, Google's AI Ethics research team has found itself at the center of a great deal of coverage. These same occurrences have also caused Google to lose credibility in the area and frayed its relationship with external groups. VP of Engineering Marian Croak (pictured), who was recently tapped to lead the team, has said that the team will double its size to 200 members. CEO Sundar Pichai has also said the team will receive a boost to its operating budget.

Podcast

Check out our weekly podcast covering these stories!
Website | RSS | iTunes | Spotify | YouTube

News

Advances & Business

A scientist created emotion recognition AI for animals - "A researcher at Wageningen University & Research recently published a pre-print article detailing a system by which facial recognition AI could be used to identify and measure the emotional state of farm animals."
AI analog chip startup Mythic raises $70 Million - "Mythic, a startup making artificial intelligence chips for use in gadgets such as smart home devices or security cameras, on Tuesday said it raised $70 million in its latest round of funding."
Soniox Poised To Advance AI Speech Recognition Market With Increased Accuracy - "Earlier this week, I had an opportunity to interview Klemen Simonic, the Founder and CEO of Soniox, who has built a promising new AI self-learning infrastructure and toolset to build advanced speech and audio understanding solutions to solve complex business problems."
Chef Robotics raises $7.7M to help automate kitchens - "San Francisco-based Chef Robotics today announced that it has raised a combined $7.7 million pre-seed and seed round, with the goal of helping automate certain aspects of food preparation."
Facebook’s new technique helps AI systems forget irrelevant information - "Facebook says it has developed an AI technique that enables machine learning models to only retain certain information while forgetting the rest."
Google Replaces BERT Self-Attention with Fourier Transform: 92% Accuracy, 7 Times Faster on GPUs - "A research team from Google shows that replacing transformers’ self-attention sublayers with Fourier Transform achieves 92 percent of BERT accuracy on the GLUE benchmark with training times seven times faster on GPUs and twice as fast on TPUs."
Yelp built an AI system to identify spam and inappropriate photos - "Yelp says its engineers built a custom, in-house system using machine learning algorithms to analyze hundreds of thousands of photo uploads per day — detecting inappropriate and spammy photos at scale."

Concerns & Hype

Tesla privately admits Elon Musk has been exaggerating about ‘full self-driving’ - "Tesla CEO Elon Musk has been overstating the capabilities of the company’s advanced driver assist system, the company’s director of Autopilot software told the California Department of Motor Vehicles."
The Pentagon Inches Toward Letting AI Control Weapons - "Last August, several dozen military drones and tank-like robots took to the skies and roads 40 miles south of Seattle. Their mission: Find terrorists suspected of hiding among several buildings."
Black and Queer AI Groups Say They'll Spurn Google Funding - "Three groups focused on increasing diversity in artificial intelligence say they will no longer take funding from Google."
Emotion recognition: can AI detect human feelings from a face? - "The market for the technology is growing rapidly despite questions from scientists about whether it works."
Grand Theft Auto Looks Frighteningly Photorealistic With This Machine Learning Technique - "As powerful as video game consoles have become, even the most graphically stunning video games don’t look like realistic, real-world footage, which is arguably the ultimate goal."
Cooperative AI: machines must learn to find common ground - "Artificial-intelligence assistants and recommendation algorithms interact with billions of people every day, influencing lives in myriad ways, yet they still have little understanding of humans."

Analysis & Policy

Agencies sharing AI use cases after December executive order - "The executive order on trustworthy artificial intelligence issued by President Trump in December has encouraged agencies like the Department of Veterans Affairs to share best practices."

Expert Opinions & Discussion within the field

This Researcher Says AI Is Neither Artificial nor Intelligent - "Kate Crawford, who holds positions at USC and Microsoft, says in a new book that even experts working on the technology misunderstand AI. "

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

Last Week in AI