Last Week in AI #128: warehouse robots, drone delivery woes, AI to tackle online harassment

A new generation of AI-powered robots is taking over warehouses, The slow collapse of Amazon's drone delivery dream, This New Way to Train AI Could Curb Online Harassment, and more!

Aug 9, 2021

A new generation of AI-powered robots is taking over warehouses

A growing number of companies are working to commercialize advances in robotic perception and planning for the task of warehouse automation. Unlike already deployed robots which typically perform the same motions over and over and don't take context into account, these companies' robots have the intelligence to adaptively manipulate a wide variety of objects. These kinds of skills are useful for automating tasks such as pick-and-place, which involves moving a objects between different containers. While it will take time to retrofit existing factories and warehouses to make use of this technology, it appears likely that this adoption is only a matter of time.

For now, somewhere around 2,000 AI-powered robots have been deployed, with a typical warehouse housing one or two...But the industry has reached a new inflection point, and [it's predicted] that each warehouse will soon house upwards of 10 robots, growing the total to tens of thousands within the next few years.

The slow collapse of Amazon's drone delivery dream

The UK Amazon drone delivery team has laid off more than 100 employees as it scaled back operations over the last two years. Interviews with former employees reveal a chaotic and dysfunctional working environment, from changing product requirements to the constant churn of workers and managers. While Amazon opened the UK team in 2016 with a PR blitz, promising drone deliveries in months, the company's technical approach made the endeavor that much harder. In particular, unlike drones from other companies that drop packages meters off the ground, Amazon opted to land the drone outside customer's houses. Doing so safely and robustly in a wide variety of conditions proved a difficult challenge for AI-powered drones. As the CEO of the drone company Matternet puts it, "promises of mass market e-commerce drone delivery [...] are far from reality."

This New Way to Train AI Could Curb Online Harassment

Surgeon hands remove a toxic symbol from a computer.

Online harassment is a scourge that has persisted since the early days of the internet. In recent years, researchers have used machine learning to identify content that constitutes misogyny and other forms of harassment. But training those machine learning algorithms requires labeled data, and deciding whether an online post constitutes harassment is a tricky issue. Rather than using part-time contractors who are paid pennies per post, a group of Danish researchers has hired a group of 8 people full-time. The benefit is that this group, with their differing identities and political opinions, might be able to help determine what counts as harassment more accurately than other labeling methods would.

If you are not subscribed yet and like our weekly AI newsletters, feel free to subscribe below!

Other News

Advances

Schools Look for Help From AI Teacher's Assistants - "Researchers are developing artificial intelligence aimed at keeping students engaged and saving educators' time Resume Subscription"

Researchers From Tel Aviv University, UC Berkeley and NVIDIA Introduce "DETReg", A Novel Unsupervised AI For Object Detection - "The application of self-supervised pretraining to computer vision has been beneficial, especially for object detection. However, previous approaches were not designed with localization in mind - which poses a key obstacle when it comes to tasks like detecting objects."

Facebook AI Releases "VoxPopuli", A Large-Scale Open Multilingual Speech Corpus For AI Translations in NLP Systems - "With the wide-scale use of speech recognition and translation technologies, these AI systems can be implemented in many different languages."

DeepMind & Google Use Neural Networks to Solve Mixed Integer Programs - "Linear programming is used to maximize or minimize a linear objective function subject to one or more constraints, while mixed integer programming (MIP) adds one additional condition: that at least one of the variables can only take on integer values."

This SoftBank-Backed Korean Edtech Startup Helps Solve Your Math Problems With AI - "After raising $50 million earlier this month in a SoftBank-backed funding round, Seoul-based edtech startup Mathpresso aims to disrupt the traditional tutoring industry with AI and plans to join the growing list of Korean companies going public."

Microsoft AI Researchers Introduce A Neural Network With 135 Billion Parameters And Deployed It On Bing To Improve Search Results - "Transformer-based deep learning models like GPT-3 have been getting much attention in the machine learning world. These models excel at understanding semantic relationships, and they have contributed to large improvements in Microsoft Bing's search experience."

Mozilla Common Voice Adds 16 New Languages and 4,600 New Hours of Speech - "The Mozilla Common Voice initiative has released a new, expanded data set featuring 16 new languages - like Basaa and Kazakh - and 4,622 new hours of speech. Mozilla Common Voice is an open-source initiative to make voice technology more inclusive."

Google AI Introduces Two New Datasets, 'timeDial and Disfl-QA, For Conversational NLP (Natural Language Processing) - "Natural language processing (NLP) has made significant advancements in recent years, with applications in learning, comprehending, and generating human language content."

Here Come the Robot Nurses - "The pandemic increased demand and possibilities for automating care, but doing so may deliver racist stereotypes and unemployment for women of color."

China's lonely hearts reboot online romance with artificial intelligence - "While human companions can be elusive or fickle, chatbots are always there to listen."

Prompt-based learning can make language models more capable - "Given a range of carefully designed prompts, a language model trained in an unsupervised fashion - that is, on unlabeled data - can be used to solve a number of tasks."

Fast, Efficient Neural Networks Copy Dragonfly Brains - "In each of our brains, 86 billion neurons work in parallel, processing inputs from senses and memories to produce the many feats of human cognition."

With a nudge from AI, ketamine emerges as a potential rare disease treatment - "In the seven difficult years since their son Mateo was diagnosed with a rare disease, Victoria Malvagno and Frank Solorzano have been waiting for medicine to catch up with their lives."

Google AI Introduces translatotron 2, A Neural Direct Speech-To-Speech Translation Model Without The Deepfake Potential - "Google has been working in the arena of artificial intelligence for quite some time now, and it has managed to achieve some amazing feats in the same. One such creation is its Translatotron that was released in the year 2019."

Tably, An AI-Powered Camera App, Give You A Deeper Understanding Of Your Cat's Health And Mood - "Cats are a popular pet choice, but many cat owners are puzzled by their feline companions' emotions. This is a challenging assignment because cats are not very expressive and are good at hiding their feelings when in pain. Tably, a new AI-powered software by Sylvester."

Man defends against package thieves using machine learning AI, flour, and very loud sirens - "There's no feeling worse than watching your packages getting lifted with nothing you can do to stop the porch pirates. After recently having a package stolen, Youtuber Ryder Calm Down built himself his own anti-package pilfer alarm system using AI, flour, and a truck horn."

Business

Nvidia is tracking more than 8,500 AI startups with $60B in funding - "Those companies have raised more than $60 billion in funding and come from 90 countries, Nvidia said."

Corvus Robotics' Autonomous Drones Tackle Warehouses - "Looking to such specialized nervous systems as a model for artificial intelligence may prove just as valuable, if not more so, than studying the human brain. Consider the brains of those ants in your pantry. Each has some 250,000 neurons. Larger insects have closer to 1 million."

How DoorDash Uses ML To Solve Last Mile Connectivity - "Tony Xu used to deliver Dominos pizzas from door to door. Today, he is the CEO and co-creator of one of the top AI-driven food delivery apps in the US, DoorDash. Not to mention the poster boy of the American dream."

Concerns

AI has the worst superpower - medical racism. - "Is this the darkest timeline? Are we the baddies? Hi! It has been a while!"

Tesla engineers were quietly concerned with Elon Musk's early claims about the company's self-driving tech, according to a new book - "Elon Musk has caught heat for years over the way he exaggerates the capabilities of Tesla's driver-assistance features: Autopilot and Full Self-Driving. Autopilot doesn't pilot cars automatically, and Full Self-Driving doesn't make them autonomous."

If You Don't Trust A.I. Yet, You're Not Wrong - "Mr. Pasquale and Mr. Malgieri are law professors. Mr. Pasquale is the author of New Laws of Robotics: Defending Human Expertise in the Age of AI. Americans have good reason to be skeptical of artificial intelligence."

After DeepMind's cofounder was placed on leave for bullying, Google promoted him - "Last week, DeepMind ... made headlines for a biotech breakthrough. This week, it's making headlines for a very different reason."

Tortured phrases' give away fabricated research papers - "You have full access to this article via your institution. In April 2021, a series of strange phrases in journal articles piqued the interest of a group of computer scientists."

AI bias is prevalent but preventable - here's how to root it out - "Simply put, we must continuously detect and eliminate human biases from AI applications for the technology to reach its potential."

Hype

Pentagon Experimenting With AI To "Predict The Future" - "The Pentagon is reportedly experimenting with AI and other cutting edge technology in the hope of being able to "predict the future"."

These Algorithms Look at X-Rays'and Somehow Detect Your Race - "Millions of dollars are being spent to develop artificial intelligence software that reads x-rays and other medical scans in hopes it can spot things doctors look for but sometimes miss, such as lung cancers."

AI Wrote Better Phishing Emails Than Humans in a Recent Test - "Natural language processing continues to find its way into unexpected corners. This time, it's phishing emails."

Analysis

The "Who" in Explainable AI: New Study Explores the Creator-Consumer Gap - "With the increasing deployment of AI systems in high-stakes decision-making domains such as healthcare, finance and law, the technology's explainability has become an issue of public concern."

Why CAPTCHA Pictures Are So Unbearably Depressing - "Part of it is the sheer hassle of repeatedly identifying objects - traffic lights, staircases, palm trees and buses - just so I can finish a web search. I also don't like being forced to donate free labor to AI companies to help train their visual-recognition systems."

Policy

Bipartisan bill looks to get acquisition workforce on board with AI - "The leaders of the Senate Homeland Security and Government Affairs Committee are looking to get the federal workforce - particularly program managers and acquisition specialists - on board with artificial intelligence."

Explainers

Introduction to AI for Social Good - "New technology is not good or evil in and of itself. It's all about how people choose to use it. - David Wong, Professor of Philosophy at Duke University"

Looking for work? Here's how to write a resume that an AI will love. - "Once you optimize your resume, you may want to practice interviewing with an AI too. Through job-matching platforms and AI-powered games and interviews, companies are relying more and more on artificial intelligence to streamline the hiring process."

Podcast

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

Last Week in AI