Last Week in AI #135: FDA approves first AI-assisted cancer diagnosis, UK outlines national AI strategy

FDA approves the first AI-based software to assist prostate cancer diagnosis, UK aims to level up its national AI strategy, and more!

Sep 27, 2021

First AI Pathology Program Approved: Helps Detect Prostate Cancer

The U.S. Food and Drug Administration (FDA) approved the first AI-based software product to help diagnose diseases. The software, developed by Paige.AI, will assist doctors in diagnosing prostate cancers by processing images from prostate biopsies. In its clinical trial, 16 pathologists examined 527 images of biopsies, (171 cancerous and 356 benign), and each pathologist completed assessments both with and without the assistance of Paige.AI's software. The study found that when doctors used the AI-based software to assist their assessments, correct diagnosis of cancerous samples improved by 7.3%, while accuracy for benign samples were unaffected. The Co-founder and Chief Medical Officer at Paige.AI hopes that this approval will "pave the way for the introduction of numerous future tools to help standardize pathology diagnosis."

UK announces a national strategy to ‘level up’ AI

AI – artificial intelligence concept. Machine learning. Central Computer Processors on the circuit board with luminous tracks. Encoded data. Computer chip over circuit background, 3D illustration

Following on the heels of the EU’s recent draft AI act, the UK has announced its own national AI strategy. The 10 year plan to invest in Britain’s AI primacy targets support at upskilling and reskilling. It includes plans to launch government a national office and research program, among other measures. The country aspires to be the most pro-innovation regulatory environment in the world, but it is unclear how much policy substance the strategy has.

Research

Robots and machine learning researchers combine forces to speed up the drug development process - "IBM Research and Arctoris announce a research collaboration to test a closed-loop platform."

A.I. keeps on trucking despite threats of a long and cold winter - "Despite the slower-than-anticipated progress, businesses haven’t given up on A.I., according to authors of the One Hundred Year Study on Artificial Intelligence report."

Summarizing Books with Human Feedback - "Scaling human oversight of AI systems for tasks that are difficult to evaluate."

Google’s Zero-Label Language Learning Achieves Results Competitive With Supervised Learning - "A Google AI research team explores zero-label learning (training with synthetic data only) in natural language processing, and introduces Unsupervised Data Generation (UDG), a training data creation procedure designed to synthesize high-quality training data without human annotations."

How the U.S. Army Is Turning Robots Into Team Players - "Engineers battle the limits of deep learning for battlefield bots"

Improved algorithms may be more important for AI performance than faster hardware - "When it comes to AI, algorithmic innovations are substantially more important than hardware — at least where the problems involve billions to trillions of data points."

Applications

AI might help edit the next generation of blockbusters - "More specifically, we’re looking at how AI is being used as a tool to help people streamline the process of creating video content."

Apple Is Working on iPhone Features to Help Detect Depression, Cognitive Decline - "Company is working with UCLA, Biogen to see if sensitive data like facial expressions, typing metrics could signal mental-health concerns "

A machine-learning algorithm to target COVID testing of travellers - "Optimizing the testing of incoming travellers for COVID-19 involves predicting those who are most likely to test positive. A machine-learning algorithm for targeted testing has been implemented at the Greek border."

Business

Hai Robotics picks up $200M for its warehouse robot - "Hai Robotics, a Shenzhen, China-based warehouse robotics startup that develops autonomous, case-handling robotic (ACR) system, announced it has raised $200 million. The funding comes from two separate equity rounds, a Series C and Series D, which are being disclosed simultaneously."

The state of MLOps in 2021 is dominated by startups - "Enterprises’ urgent need is for startups to help solve getting more machine learning (ML) models into production."

Concerns

Elon Musk’s Push to Expand Tesla’s Driver Assistance to Cities Rankles a Top Safety Authority - "Tesla plans to expand access to Full Self-Driving system and defends its tech; safety official says more work is needed "

The Scientist and the A.I.-Assisted, Remote-Control Killing Machine - "Israeli agents had wanted to kill Iran’s top nuclear scientist for years. Then they came up with a way to do it with no operatives present."

When AI breaks bad - "Illustration: Annelise Capossela/AxiosA new report about artificial intelligence and its effects warns AI has reached a turning point and its negative effects can no longer be ignored."

NIST AI framework must take into account global catastrophic risk say UC Berkeley researchers - "Researchers from the University of California, Berkeley have called on the National Institutes of Standards and Technology (NIST) to explicitly consider the threat of AI-triggered global catastrophes in its new framework intended to govern use of the technology across federal government."

AI’s Islamophobia problem - "GPT-3 is a smart and poetic AI. It also says terrible things about Muslims."

AI Can Write Code Like Humans—Bugs and All - "Some software developers are now letting artificial intelligence help write their code. They’re finding that AI is just as flawed as humans."

Google’s Former AI Ethics Chief Has a Plan to Rethink Big Tech - "Timnit Gebru says regulators need to provide whistleblowers working on artificial intelligence with fresh protections backed up by tough enforcement."

Amazon's AI-powered cameras punish its delivery drivers when they look at side mirrors or when other cars cut them off, report says - "Amazon drivers say they are being punished for some driving habits that are considered safe and others that are beyond their control, Motherboard reported."

DeepMind tells Google it has no idea how to make AI less toxic - "To be fair, neither does any other lab"

MIT study finds Tesla drivers become inattentive when Autopilot is activated - "By the end of this week, potentially thousands of Tesla owners will be testing out the automaker’s newest version of its “Full Self-Driving” beta software, version 10.0."

Deep Learning’s Diminishing Returns - "The cost of improvement is becoming unsustainable"

Exclusive: ICE just paid $3.9 million for ‘rapid’ AI-powered facial recognition at migrant detention facilities - "Immigration and Customs Enforcement recently paid an identity verification company called Trust Stamp $3.9 million for AI-powered facial recognition technology that will be used in migrant detention centers, according to a new contract listing."

Falsehoods more likely with large language models - "The use of AI language models to generate text for business applications is gaining steam."

Computer vision-powered workplace safety systems could lead to bias and other harms - "Increasingly, AI is being pitched as a way to prevent the estimated over 340 million workplace accidents that occur worldwide every day."

Chilling': Facial recognition firm Clearview AI hits watchdog groups with subpoenas - "Advocacy groups argue the company is trying to bully critics."

Analysis

The Role of AI in Art Creation - "How artists can use AI in their creative processes"

7 Revealing Ways AIs Fail - "Neural networks can be disastrously brittle, forgetful, and surprisingly bad at math"

Expert Opinions

AI Architects: The New Programmers - "On the nature of the novel paradigm of programming"

Podcast

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

Last Week in AI