Discover more from Last Week in AI
AI to spot wildfires, human oversight of AI, and more!
Last Week in AI #121
Tower-mounted cameras have assisted California communities in spotting wildfires for years. However, it is very difficult for human operators to keep up with a wall of live video feeds 24/7 to monitor potential flareups. Since May this year, Sonoma County, where the destructive Tubbs Fire in 2017 occurred, have been experimenting with AI software to process these videos in real time and look for hotspots. Early results are mixed - sometimes the AI system can raise alerts faster than human operators, sometimes slower. Beyond Sonoma, AI-assisted wildfire monitoring is slowly being adopted across the western U.S. These systems will get better over time as more data is collected per location - the algorithm needs to be finetuned to specific weather patterns on site. In the future, many hope that AI systems can also predict likely spots for fires before they occur by accessing live sensor data beyond cameras, such as air temperature and soil moisture.
Recently, policy makers have called for greater human oversight on AI applications in industry to reduce harm. However, effective implementation of such oversight is challenging, and there are several common pitfalls. The presence of a human in a larger decision making process does not automatically make the decisions produce less harm. For example, in many cases it was found that humans overestimate the credibility of AI systems and would simply "rubber stamp" automated decisions. The presence of human operators also make it easier for institutions to both legitimize automated decisions and also shift responsibility onto the humans, instead of the AI system, when harms occur. This all underpins a seeming paradox with AI decision making, that they are supposed to "improve upon the cognitive limits and biases of humans" and yet humans are being presented as "the essential backstop overseeing algorithmic limits and biases."
Advances & Business
Cruise secures $5B credit line to buy electric, autonomous Cruise Origin vehicles from GM - "Cruise, the self-driving subsidiary of GM, has tapped a $5 billion line of credit from the automaker's financial arm to pay for hundreds of purpose-built electric and autonomous Origin vehicles as they start to roll off the assembly line."
Enterprise AI budgets are up 55% over 2020, Appen says - "Enterprises are accelerating their AI strategies as a result of the COVID-19 pandemic and ballooning their budgets accordingly"
Preferred Networks Becoming Japan's Industrial AI Powerhouse - "There's an AI unicorn startup you might not have even heard of. That is, if you live in the Western hemisphere."
Google's Next AI Move: Teaching Foreign Languages - "Google is taking a more incremental step in conversational AI by preparing to teach foreign languages through Google Search"
Facebook scientists say they can now tell where deepfakes have come from - "Facebook researchers claim that their software can identify the AI that was used to create a deepfake."
Facebook is testing AI to get you to stop fighting in its groups - "Conversations can quickly spiral out of control online, so Facebook is hoping artificial intelligence can help keep things civil."
Drones, robots, and AI are changing the face of solar and wind farm inspections, literally - "Robots and drones with deep learning computer vision and advanced imaging cameras are changing the landscape of inspections for renewable energy sites and associated equipment."
Harnessing the Wild Power of AI Image Generation - "A layout- and style-based architecture shows how to control AI capabilities to generate complex images"
Concerns & Hype
China's tech workers pushed to limits by surveillance software - "A vicious cycle of monitoring and overwork is fueling productivity -- and backlash"
McDonald's AI drive-thru bot accused of breaking biometrics privacy law - "McDonald's has been accused of illegally collecting and processing customers' voice recordings without their consent in the US state of Illinois. Like so many giant corporations, McDonald's has turned to AI technology to use computers in place of people."
Tech Companies Are Training AI to Read Your Lips - "First came facial recognition. Now, an early form of lip-reading AI is being deployed in hospitals, power plants, public transportation, and more. A patient sits in a hospital bed, a bandage covering his neck with a small opening for the tracheostomy tube that supplies him with oxygen."
Inside the fight to reclaim AI from Big Tech's control - "In 2020, as the co-lead of Google's ethical AI team, Gebru had reached out to Emily Bender, a linguistics professor at the University of Washington, and asked to collaborate on research about the troubling direction of artificial intelligence."
14 Ways AI Could Become A Detriment To Society - "Completely dependent on the humans who build and train it, AI can be infused with the biases of its creators. And other negative consequences can come with the increasing dominance of AI that aren't obvious to the average layperson."
Study shows physicians' reluctance to use machine-learning for prostate cancer treatment planning - "A study shows that a machine-learning generated treatment plan for patients with prostate cancer, while accurate, was less likely to be used by physicians in practice. Advancements in machine-learning (ML) algorithms in medicine have demonstrated that such systems can be as accurate as humans."
Bias isn't the only problem with credit scores -and no, AI can't help - "The biggest-ever study of real people's mortgage data shows that predictive tools used to approve or reject loans are less accurate for minorities."
The Efforts to Make Text-Based AI Less Racist and Terrible - "Language models like GPT-3 can write poetry, but they often amplify negative stereotypes. Researchers are trying different approaches to address the problem."
South Korea to Deploy Rail-Mounted Robot and AI Surveillance on Border - "South Korea will install a rail-mounted robot and artificial intelligence (AI)-based surveillance system later this year in 'no man's land' on the Korean peninsula."
Analysis & Policy
Taming AI's Can/Should Problem - "Bias in artificial intelligence seems to be an unending and endemic problem."
Regulators know teleoperation is key for self-driving vehicles to succeed - "The United Kingdom, Sweden, Japan, California, Michigan, and Texas are just some of the places where teleoperation is mandatory for any operation of autonomous vehicles for testing purposes or pilot programs"
Expert Opinions & Discussion within the field
Amazon Alexa head scientist on developing trustworthy AI systems - "Rohit Prasad, head scientist at Amazon's Alexa division, believes that the industry is at an inflection point. Moving forward, it must ensure that AI learns about users in the same ways users learn so that a level of trust is maintained"
There's More To Machine Learning Than CNNs - "Neural networks - and convolutional neural networks (CNNs) in particular - have received an abundance of attention over the last few years, but they're not the only useful machine-learning structures."
Fun and Dystopia With AI-Based Code Generation Using GPT-J-6B - "Since OpenAI will not open-source the 175 billion parameter GPT-3 text generation model, others such as EleutherAI are developing their own, by training not-quite-as-large Transformer-based models but still getting impressive results."
Copyright © 2021 Skynet Today, All rights reserved.