Discover more from Last Week in AI

Weekly text and audio summaries of the most interesting AI news, as well as editorials commenting on recent events

Over 20,000 subscribers

News

Last Week in AI #136: Amazon's new home robot, DeepMind AI predicts where it's going to rain, the future of AI ethics, and more!

Amazon's new Astro robot seems to promise more than it can deliver, DeepMind's new software accurately predicts where it's going to rain in 2 hours, and how AI ethics can be effective in the future

Last Week in AI

Oct 4, 2021

Amazon’s Astro Is a Mobile Alexa and Cup Holder that Costs $1k

Ever wanted a cheap robot that will do your chores and take care of everyday tasks around the home? That's exactly what you're not getting with Amazon's new household robot, Astro. Many seem to agree that the behemoth's PR video about the robot fails to answer why it's needed in the first place: it seems to be a $1000 contraption that lets you be closer to Alexa's voice. While Amazon says Astro is "a big step forward," it is important to keep our expectations in check about such creations. At the same time, it's possible that Astro did make some incredible leaps: IEEE Spectrum acknowledges that doing simultaneous localization and mapping in a real home environment is very challenging, but it is also possible that Amazon is using hand-picked clips to make Astro seem more capable than it is.

DeepMind’s AI predicts almost exactly when and where it’s going to rain

Researchers at DeepMind and several other institutions recently published the paper “Skilful precipitation nowcasting using deep generative models of radar” in the Nature journal. “Nowcasting” is defined as the high-resolution forecasting of precipitation up to two hours ahead, and the paper shows that the new approach “ranked first for its accuracy and usefulness in 89% of cases against two competitive methods”, with the rankings provided by 56 weather forecasters. As with DeepMind’s prior breakthrough with the protein-folding model AlphaFold, this is another demonstration of the company making substantial progress on long-established problems and achieving results that are sure to have major implications.

Are AI ethics teams doomed to be a facade? The women who pioneered them weigh in

AI ethics has become a hot topic in recent years due to the wide deployment of AI-based solutions. Examples of AI harm are real and immediate in fields like facial recognition, hiring decisions, and credit card approvals. However, it's not clear what is the best way to tackle and improve AI ethics. So far, large companies like Google and Facebook have built their internal ethics teams, but their works have all been stifled when pitted against the companies' immediate commercial interests. For Google it was large language models, and for Facebook it was misinformation. Experts cite the lack of leadership buy-in, from both the technical and legal sides, as causes for these ineffective ethics efforts. Moving forward however, many argue that such confrontational views of AI ethics is misleading. The goal of AI ethics is to prevent harm to a company's customers and users, and society at large, and this is aligned, not against, a company's long-term business interests.

Research

These virtual robot arms get smarter by training each other - "A virtual robot arm has learned to solve a wide range of different puzzles—stacking blocks, setting the table, arranging chess pieces—without having to be retrained for each task. It did this by playing against a second robot arm that was trained to give it harder and harder challenges."

Israeli researchers bypass facial recognition using AI-generated makeup patterns - "Israeli researchers have found an apparently straightforward method to fool facial recognition software — by applying conventional makeup to specific areas of the face according to a pattern determined by an artificial intelligence program."

Applications

GitHub Copilot AI Spawns Open Source Alternatives - "GitHub Copilot, described as an "AI pair programmer," debuted this year with a splash, amazing developers with its ability to supply chunks of code when a user is typing in Visual Studio Code and even generate whole applications solely through typed commands."

Why OpenAI's Codex Won't Replace Coders - "While machine learning has been around a long time, deep learning has taken on a life of its own lately."

Google is using AI to help users explore the topics they’re searching for — here’s how - "It’s a simple enough question for humans to understand, says Pandu Nayak, vice president of search at Google, but such a query represents the cutting-edge of machine comprehension."

ByteDance Proposes ‘DyStyle’: A Novel Dynamic Neural Network For Style Editing - "In the last few years, AI researchers have been using Generative adversarial networks (GANs) to create images with unprecedented levels of diversity and photorealism, and one such example is StyleGAN."

Tesla’s arc of AI progress - "I’ve followed Karpathy’s work since taking CS231n back in 2015, but I’ve been admittedly skeptical of Tesla’s approach to camera-only autonomy. That said, I found Tesla’s recent AI day presentation impressive."

Adobe Lightroom is getting more powerful with AI-based selection tool - "Adobe will add a new selection tool to its Lightroom software in October to help photo editors get the look they want with less manual labor."

Business

Tesla opens ‘Full Self-Driving’ beta software to more customers - "Despite the head of the National Transpiration Safety Board expressing serious concerns about its safety last week, Tesla has now enabled access to the beta of its “Full Self-Driving” (FSD) program to more Tesla drivers, via a “request” button on Teslas’ dashboard screens."

San Francisco raises Tesla 'self-driving' safety concerns as public test nears - "San Francisco transport authorities on Thursday raised concerns about the safety record of Tesla's advanced driver assistant system, as the electric car maker prepared a wide release of a test version of the software that works on city streets and highways."

DMV Approves Cruise and Waymo to Use Autonomous Vehicles for Commercial Service in Designated Parts of Bay Area - "Sacramento – The California Department of Motor Vehicles today issued autonomous vehicle deployment permits to Cruise LLC and Waymo LLC, allowing the companies to charge a fee and receive compensation for autonomous services offered to the public."

VCs invested over $75B in AI startups in 2020 - "Investments in AI are growing at an accelerated pace, according to a new report from the Organization for Economic Cooperation and Development (OECD)."

AI Adoption Skyrocketed Over the Last 18 Months - "When it comes to digital transformation, the Covid crisis has provided important lessons for business leaders. Among the most compelling lessons is the potential data analytics and artificial intelligence brings to the table."

Honda prepping AI robots, inter-city electric air travel and lunar technology for 2030 - "Honda unveiled its priorities for the next decade on Thursday, highlighting the company's efforts to reduce its environmental impact, expand its work in robotics and avionics and facilitate travel to the Moon."

Concerns

Will Warehouse Automation Replace Humans? - "Read any article on automated warehouse vehicles, and it’s pretty easy to see there is a lot of hype. Although automation in automated guided vehicles (AGVs) and autonomous mobile robots (AMRs)—my specialty—have come a long way, they are not replacing all warehouse workers."

Facial recognition contractor Clearview AI withdraws subpoenas served on nonprofit - "Written by A technology company that provides facial recognition technology to at least four government agencies has withdrawn subpoenas it served on a nonprofit."

An Inconvenient Truth About AI - "Fortunately for such artificial neural networks—later rechristened "deep learning" when they included extra layers of neurons—decades of Moore's Law and other improvements in computer hardware yielded a roughly 10-million-fold increase in the number of computations that a computer could do in a"

Misinformation Is About to Get So Much Worse - "For years now, artificial intelligence has been hailed as both a savior and a destroyer. The technology really can make our lives easier, letting us summon our phones with a “Hey, Siri” and (more importantly) assisting doctors on the operating table."

The limitations of AI safety tools - "It’s unlikely that a single tool will ever be able to prevent unsafe decision-making in AI systems."

In health care, many AI developers are white men. UNT is trying to diversify the field - "The University of North Texas Health Science Center will lead a massive new federal project to train a more diverse workforce in the world of artificial intelligence and health care, the National Institutes of Health announced."

Policy

The United States must lead the way on artificial intelligence standards - "Nearly three years ago, Chinese President Xi Jinping announced before the Chinese Communist Party’s policy-making elite that artificial intelligence “is a vital driving force for a new round of technological revolution and industrial transformation, and accelerating AI development is a strateg"

Explainers

How Deep Learning Works - "Fortunately for such artificial neural networks—later rechristened "deep learning" when they included extra layers of neurons—decades of Moore's Law and other improvements in computer hardware yielded a roughly 10-million-fold increase in the number of computations that a computer could do in a"

The Turbulent Past and Uncertain Future of Artificial Intelligence - "In the summer of 1956, a group of mathematicians and computer scientists took over the top floor of the building that housed the math department of Dartmouth College. For about eight weeks, they imagined the possibilities of a new field of research."

Podcast

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