Last Week in AI #113
New EU AI regulations, AI for spotting pollution hotspots, and more!
Researchers from Duke university recently published a new machine learning model capable of predicting levels of PM2.5 particles in cities within a resolution of 300 meters from weather and satellite imagery data. This is a big improvement over previous methods of localizing pollution, which typically give data on the resolutions of cities, not city blocks. Finer resolution data is available, but they come from costly installations of ground-level sensor networks or driving a sensor-equipped car around the city. Currently, the model only works for one city at a time, and it needs to be retrained for new cities. In the future, the researchers hope that their method can enable better policy making regarding pollutions and more accurate studies on the effects of pollution on health.
The European Commission released a draft legislation proposal on regulations of AI applications that aim to curb the potential harms of deploying these automated systems. To do so, the proposal categorizes a series of applications as "high-risk" and would tightly regulate these areas, in some cases requiring explicit approval from the EU for commercial applications. Such high-risk areas include facial recognition, engagement optimization, student evaluation, and hiring candidate evaluation.
However, the article's claim that these regulations leave "loophole for mass surveillance" by law enforcement is overstated. From the proposal: "The use of those systems for the purpose of law enforcement should therefore be prohibited, except in three exhaustively listed and narrowly defined situations," and that "each use of a ‘real-time’ remote biometric identification system in publicly accessible spaces for the purpose of law enforcement should be subject to an express and specific authorisation by a judicial authority or by an independent administrative authority of a Member State."
Advances & Business
Cerebras' New Monster AI Chip Adds 1.4 Trillion Transistors - "Shift to 7-nanometer process boosts the second-generation chip's transistor count to a mind boggling 2.6-trillion"
China's Exascale Prototype Supercomputer Tests AI Workloads - "We are starting to see more exascale and large supercomputing sites benchmark and project on deep learning capabilities of systems designed for HPC applications but only a few have run system-wide tests to see how their machines might stack up against standard CNN and other metrics."
NVIDIA, Stanford & Microsoft Propose Efficient Trillion-Parameter Language Model Training on GPU Clusters - "Large-scale transformer-based language models have produced substantial gains in the field of natural language processing (NLP)."
Nvidia announces Morpheus, an AI-powered app framework for cybersecurity - "Join GamesBeat Summit 2021 this April 28-29. Register for a free or VIP pass today."
Facebook unveils AI model to mix up cancer-curing cocktails with existing drugs - "No longer content to simply provide a platform for keeping up with long-lost cousins and spreading conspiracy theories, Facebook has branched out into developing artificial intelligence to help treat complex diseases."
Facebook AI, McGill U & Mila Promote "Translationese" to Boost NMT System Faithfulness - "This is a boom time for commercial applications of Neural Machine Translation (NMT), with the performance of multilingual systems rapidly advancing both in terms of translation quality and robustness to input perturbations such as spelling or grammatical errors."
OpenCV AI Competition 2021 Highlights and Team Profiles Part 1 - "If social media is any indication, OpenCV AI Competition 2021 participants have really hit the ground running."
Researchers From Allen Institute for AI Releases AI2-THOR 3.0, An Embodied AI Framework for Visual Object Manipulation In Robotics Testing Scenario - "The Allen Institute for AI (AI2) announces the release of AI2-THOR 3.0, which is an embodied AI framework."
Shares of TuSimple, a developer of autonomous trucks, fall sharply after I.P.O. before recovering. - "Shares in TuSimple, a developer of autonomous trucks that is backed by Volkswagen and UPS, fell sharply on Thursday after its initial public offering, suggesting that investors have doubts about the company's promise of putting its technology on the road by 2024."
An AI-powered version of Albert Einstein has joined UneeQ's growing lineup of 'digital humans' - "The latest virtual companions created by UneeQ, a New Zealand and Austin-based company specializing in "digital humans," include an Albert Einstein chatbot. The launch was timed to mark the 100th anniversary of his Nobel Prize in Physics."
Hive raises $85M for AI-based APIs to help moderate content, identify objects and more - "As content moderation continues to be a critical aspect of how social media platforms work - one that they may be pressured to get right, or at least do better in tackling - a startup that has built a set of data and image models to help with that, along with any other tasks that require automat"
Can Medical Alexas Make Us Healthier? - "America's health care system needs help. But don't expect technology to be a silver bullet."
Concerns & Hype
CR Engineers Show a Tesla Will Drive With No One in the Driver's Seat - "Consumer Reports engineers easily tricked our Tesla Model Y this week so that it could drive on Autopilot, the automaker's driver assistance feature, without anyone in the driver's seat - a scenario that would present extreme danger if it were repeated on public roads."
Walmart Is Pulling Plug on More Robots - "Retailer has been phasing out automated pickup towers as curbside services have become popular"
Stop talking about AI ethics. It's time to talk about power. - "At the turn of the 20th century, a German horse took Europe by storm. Clever Hans, as he was known, could seemingly perform all sorts of tricks previously limited to humans."
Analysis & Policy
Tesla is being scrutinized by Senate Democrats for Autopilot misuse - "Two Senate Democrats are urging federal regulators to take "corrective actions" against Tesla to prevent further misuse of the company's advanced driver assist feature."
To ensure inclusivity, the Biden administration must double down on AI development initiatives - "The National Security Commission on Artificial Intelligence (NSCAI) issued a report last month delivering an uncomfortable public message: America is not prepared to defend or compete in the AI era. It leads to two key questions that demand our immediate response: Will the U.S."
Senate bill looks to boost AI talent in government - "Students would receive scholarships to study artificial intelligence in exchange for federal service, should a Senate bill introduced Wednesday become law."
Expert Opinions & Discussion within the field
Geoffrey Hinton has a hunch about what's next for AI - "Back in November, the computer scientist and cognitive psychologist Geoffrey Hinton had a hunch. After a half-century's worth of attempts - some wildly successful - he'd arrived at another promising insight into how the brain works and how to replicate its circuitry in a computer."
Two ways AI technology is like Nuclear technology - "We often hear that AI is the new electricity and data the new oil. The former at least is kind of true. AI is already pervasive in all of our digital artefacts and interactions. And to be fair, so is the latter."
Melanie Mitchell Takes AI Research Back to Its Roots - "To build a general artificial intelligence, we may need to know more about our own minds, argues the computer scientist Melanie Mitchell."
That's all for this week! If you are not subscribed and liked this, feel free to subscribe below!
Copyright © 2021 Skynet Today, All rights reserved.