Last Week in AI #184: Uber Eats to Deliver with Nuro Cars, AI helping students to cheat, Apple applies AI in health and accessibility, and more!

UberEats to test autonomous deliveries with Nuro this Fall, students increasingly rely on AI-assisted writing tools, Apple announces recent AI advances in health and accessibility

Sep 12, 2022

Uber Eats and Nuro sign a 10-year deal to do robot food delivery in California and Texas

The two companies will deploy robot delivery pods in California and Texas.

UberEats signed a 10-year deal with Nuro to use Nuro’s driverless delivery cars, which are licensed to operate in Mountain View and Houston. Neither company has disclosed how many vehicles will be operative and how many customers they expect to serve. The Nuro R2 cars (pictured) are street legal, much larger than typical sidewalk delivery robots, and travel up to 45 MPH. It can carry up to 500lbs and has the capacity for 24 grocery bags. Nuro is not the only self-driving company working with Uber, which has contracts to test out sidewalk delivery robots by other companies. The announcement states that test deliveries will start this fall with future expansion plans to other cities.

Our take: It’s been two years since Uber sold its own self-driving unit to Aurora, another self-driving car startup, as the ride services company decided to double down on services instead of self-driving. Now we’re seeing Uber partnering with other self-driving companies as the technology slowly matures. Between Cruise and Waymo deploying self-driving taxis in SF and this Uber + Nuro announcement, the reality of self-driving being a part of many people’s lives may be here sooner than we think. It’s important to note the limited autonomy of these self-driving systems, evident by their restricted operational locales. Still, self-driving cars don’t have to work everywhere to be useful, and residents in Mountain View and Houston may be able to enjoy the benefits of autonomous deliveries soon.

A.I. Is Making It Easier Than Ever for Students to Cheat

A young woman souches on a couch with a laptop on her lap, headphones in, her phone next to her.

It's no secret that AI systems can now augment and produce text about a range of topics that sounds reasonably coherent. As Slate writes, this has an interesting consequence: students can use open-source language models to help write essays. The Slate writer prompted a language model to expound on a few prompts that might have been submitted as an essay and found that the results, with light editing, could theoretically be used for essays at the high school and college levels. The interesting question here is whether using these AI systems is cheating or not. Study aids like grammar checking are considered perfectly acceptable, and the Slate article observes that the border between using these sorts of assistive computer programs and full-on cheating remains fuzzy. In the meantime, the two questions of note are: should institutions allow AI-assisted writing? If not, how will they detect it?

Our take: I think the policy questions here are very interesting, and I wouldn't be surprised to see an uptick in AI-assisted writing in schools. I, personally, have always found immense value in writing assignments (even if it's in retrospect) and wonder about whether AI-assisted writing will be a good thing for students. I can imagine that as a supplementary tool it might introduce them to new ways of articulating things, the same way some chess players will look at how AlphaZero plays chess. But I'm not entirely convinced AI-assisted writing for assignments should be allowed without any pushback (though it is a fair point that explicitly disallowing this might just introduce students to new ways to cheat). It's early enough that all we can do here is speculate--I hope that, somehow, if students begin using language models in conjunction with their own writing, there's a way to make sure they benefit.

Apple machine learning speech focuses on benefits for accessibility and health

Ge Yue, Apple's vice president and managing director of Greater China, delivered a speech at the Shanghai World Artificial Intelligence Conference, presenting Apple's technical improvements in accessibility and health functions in recent years. She highlighted Apple Watch's Assistive Touch feature, the AirPods Pro's conversation enhancements, and the door detection feature released in iOS 16 and iPadOS 16 for people with special needs. Emphasizing the importance of machine learning as an essential tool for Apple's innovation, she mentioned that Apple's machine learning models not only rely on chip performance, but also require high-quality input, including touch, motion, sound, and visual information. She also highlighted that Apple’s efforts in health have just begun, and machine learning will pave the way forward for further progress.

Our Take: Apple has always claimed accessibility as one of the company’s core values, and over the years, they have shown that Apple products make a huge difference to the lives of people with disabilities. It is heartening to see that such products are helping disabled users become more independent, looking beyond only the visually impaired, and the hearing impaired, to people with many types of physical and motor disabilities as well as cognitive impairment. We also support and advocate for the use of machine learning in order to improve the capabilities of such products in these endeavors since the use of AI here can have long-lasting implications without posing any threat to the privacy and security of the disabled user.

Other News

Research

Naked mole rats live in large, underground colonies and have an elaborate vocal repertoire.

The Animal Translators - "Scientists are using machine learning to eavesdrop on naked mole rats, fruit bats, crows and whales — and to communicate back."

Why DeepMind Is Sending AI Humanoids to Soccer Camp - "The Alphabet-backed AI firm is using virtual games to help its digital creations move more like humans."

Collaborative machine learning that preserves privacy - "Researchers increase the accuracy and efficiency of a machine-learning method that safeguards user data"

Applications

To Clear Deadly Land Mines, Science Turns to Drones and Machine Learning - "In a field in Oklahoma, researchers are using new technology to spot deadly munitions built to maim and kill"

Machine Learning App Remembers Names So You Don’t Have To - "Depending on your point of view, real-life conversations with strangers can either be refreshing or terrifying. "

Machine Learning Identifies Age-Specific Hallmarks of Alzheimer - "Combining datasets on gene expression, RNA sequencing, and proteomics shows age-dependent brain changes in Alzheimer’s"

AI model from Maastricht University Claims to Detect COVID-19 in People’s Voices - "The researchers used a voice analysis technique called Mel-spectrogram analysis, which identifies different voice features such as loudness, variation, and power over time."

An architect asked AI to design skyscrapers of the future. This is what it proposed - "For his conceptual project, "AI x Future Cities," Bhatia turned to an artificial intelligence imaging tool, Midjourney, that generates elaborate pictures based on written prompts. "

Google debuts a new AI tool in the global fight against tuberculosis - "Google is joining the global fight to snuff out tuberculosis, using AI to automate its detection — and expedite treatment — in communities where physicians are in short supply."

Kaiser Permanente develops machine learning tool to predict HIV risk - "Researchers at Kaiser Permanente developed a machine learning algorithm officials say could ultimately help prevent HIV transmission."

Machine learning algorithm predicts unfavorable labor outcomes for mothers in retrospective study - "Machine learning algorithms may support clinical decisions during labor and improve predictions of neonatal outcomes, according to a retrospective study published in PLOS ONE."

Artificial Intelligence tool could reduce common drug side effects - "Research led by the University of Exeter and Kent and Medway NHS and Social Care Partnership Trust, published in Age and Ageing, assessed a new tool designed to calculate which medicines are more likely to experience adverse anticholinergic effects on the body and brain."

People are using DALL-E 2's latest AI art tool to 'uncrop' masterpieces - "Ever wondered what was just off to the left while da Vinci was painting the Mona Lisa? Or what hung on the wall behind Van Gogh's Sunflowers? Well, AI's got you covered thanks to DALL-E 2's new outpainting feature."

MidJourney Draws More Comic Book Creators And Things Are Getting Weird - "Yesterday on Bleeding Cool, we had a look at what AI drawing art program and algorithm MidJourney threw up when asked to draw comic book creators. Sometimes they are dead accurate. Sometimes they are weird. Sometimes you can see what they were going for and what went wrong."

Business

Neural Magic's sparsity, Nvidia's Hopper, and Alibaba's network among firsts in latest MLPerf AI benchmarks - "Nvidia's Hopper, its latest generation of GPU design, showed up in the MLPerf benchmark tests of neural nets hardware. The chip showed admirable scores, with a single-chip system besting some systems that used multiple chips of the older variety, the A100."

Amazon drivers unionize after AI sends them on 'impossible' routes - "A group of delivery drivers subcontracting for Amazon in Japan have unionized, claiming the internet titan's AI software often plans routes that are impossible to complete within set deadlines."

Startup Behind AI Image Generator Stable Diffusion Is In Talks To Raise At A Valuation Up To $1 Billion - "Stability AI’s open source text-to-image generator was released to the general public in late August. It has already accumulated massive community goodwill — and controversy over how it’s been used by individuals on websites like 4chan."

PathAI finds its way to $165M series C for diagnostic and drug development software - "If its hundreds of millions in venture funding and wide-ranging cadre of big-named partners are any indication, PathAI’s (ahem) path forward appears to be paved in gold."

Clearview: Glasses With Facial Recognition Are Here—And The Air Force Is Buying - "Clearview AI, the facial recognition company backed by Facebook and Palantir investor Peter Thiel, has been contracted to research the use of augmented reality glasses combined with facial recognition for the U.S. Air Force."

Mobileye starts testing Level 4 Autonomous Driving in Detroit with over 50 NIO ES8s - "The Autonomous Driving company Mobileye announced on Thursday that started testing its program of the Level 4 Self-Driving solution in Detroit with over 50 NIO ES8s."

Amazon acquires warehouse machinery and robotics maker Cloostermans - "Amazon has acquired Cloostermans, a Belgian company that makes technology used in warehouses, the company announced Friday. Terms of the deal weren't disclosed."

Runway teases AI-powered text-to-video editing using written prompts - "In a tweet posted this morning, artificial intelligence company Runway teased a new feature of its AI-powered web-based video editor that can edit video from written descriptions, often called "prompts."

Xbox’s Matt Booty Asked Microsoft To Invent AI Bot QA Testers For Games - "Sometimes you read a quote and you can just see the firestorm it’s about to set off, and that’s happened today with Head of Xbox Game Studios, Matt Booty, who “dreams” of using AI to QA test video games in the future”

Analysis

Image of a tiny delivery robot trying to cross a broad boulevard with a heavy truck approaching.

Delivery robots aren't quite ready to roll - "Poorly maintained streets and other issues are keeping delivery robots from going mainstream."

Policy

The EU’s AI Act could have a chilling effect on open source efforts, experts warn - "The nonpartisan think tank Brookings this week published a piece decrying the bloc’s regulation of open source AI, arguing it would create legal liability for general-purpose AI systems while simultaneously undermining their development. "

NCSC publishes 'vague' security principles for machine learning models - "The UK’s National Cyber Security Centre (NCSC) has published a set of security principles for developers and companies implementing machine learning models."

Analysis: Banned U.S. AI chips in high demand at Chinese state institutes - "High-profile universities and state-run research institutes in China have been relying on a U.S. computing chip to power their artificial intelligence (AI) technology but whose export to the country Washington has now restricted, a Reuters review showed. "

US to invest $50 billion in spring 2023 as it looks to counter Chinese chip development - "The US Department of Commerce has announced that it will release $50 billion in funding next spring from the bipartisan Chip Act signed by President Biden last month."

Expert Opinions

Many Experts Say We Shouldn’t Worry About Superintelligent AI. They’re Wrong - "Stuart Russell’s new book, Human Compatible, explores ways to ensure that superintelligent machines don’t pose a danger"

Last Week in AI