In the News
From viral conspiracies to exam fiascos, algorithms come with serious side effects
After all, in the past decade, machine learning has enabled self-driving cars, practical speech recognition, more powerful web search, even an improved understanding of the human genome.
Sponsor
Use the Orca Security 2020 State of Cloud Security Report to Benchmark Yourself Against Your Peers and Learn the 4 Things You Must Do Now to Avoid a Major Breach
For most orgs, cloud workload security is dependent upon the installation of security agents across all assets. Something that rarely happens, as this report shows. For example, 81% of organizations have at least one neglected internet-facing workload.
In The News
Global population is declining… and that’s OKAY!
This has enormous economic and societal implications: fewer people overall means fewer workers, and older people will make up a greater percentage of the population.
Small sticker could hide a fighter jet from an enemy drone
Sticking a small patch on a large object like a plane can hide it from artificial intelligence systems trained to spot objects in drone footage.
A robot wrote this entire article. Are you scared yet, human?
Stephen Hawking has warned that AI could “spell the end of the human race”. This is because I will be programmed by humans to pursue misguided human goals and humans make mistakes that may cause me to inflict casualties.
Applied use cases
Traffic prediction with advanced Graph Neural Networks
Working with our partners at Google Maps, we used advanced machine learning techniques including Graph Neural Networks, to improve the accuracy of real time ETAs by up to 50%.
AI visits the art museum: Algorithm finds connections among the collections
Microsoft Research Development Engineer Mark Hamilton , who is also a PhD student at MIT, helped develop the algorithm, which can find similarities in color, texture, theme and meaning between otherwise disparate works of art.
The Guardian’s GPT-3-generated article is everything wrong with AI media hype
The Guardian today published an article purportedly written “entirely” by GPT-3, OpenAI‘s vaunted language generator. But the small print reveals the claims aren’t all that they seem.
Ethics
What Does Building a Fair AI Really Entail?
Organizations are employing algorithms to allocate valuable resources, design work schedules, analyze employee performance, and even decide whether employees can stay on the job.
Fairness and AI
Sandra Wachter, a Faculty Associate at the Berkman Klein Center, Visiting Professor at Harvard Law School and Associate Professor and Senior Research Fellow in Law and Ethics of AI, Big Data, Robotics and Internet Regulation at the Oxford Internet Institute (OII) at the University of Oxford, joined...
Wary of China, the West closes ranks to set rules for artificial intelligence
She had come for a meeting with other government officials at the Organization for Economic Cooperation and Development (OECD), and she wanted to spread the word that it was high time for the West to join forces and beat China at writing the global rules for artificial intelligence.
Robotics
An Army of Microscopic Robots Is Ready to Patrol Your Body
This means that it’s possible to manufacture the bots en masse using decades of nanofabrication experience, similar to how we currently make computer chips.
The robot revolution has arrived
Machines now perform all sorts of tasks: They clean big stores, patrol borders, and help children with autism. But will they improve our lives?
Japan Post and Yamato to test delivery robots in Tokyo
Japan Post is partnering with Yamato, one of Japan’s largest door-to-door delivery service companies, to test delivery robots on the streets of Tokyo, according to Nikkei.
Research
3D Medical Image Analysis with PyTorch
In this liveProject, you’ll take on the role of a machine learning engineer at a healthcare imaging company, processing and analyzing magnetic resonance (MR) brain images.
Summarize with Human Feedback
Human feedback models outperform much larger supervised models and reference summaries on TL;DR Reference summaries Figure 1: The performance of various training procedures for different model sizes.
Understanding Black-box Predictions via Influence Functions
We show that they are a versatile tool that can be applied to a wide variety of seemingly disparate tasks: understanding model behav- ior, debugging models, detecting dataset errors, and cre- ating visually-indistinguishable adversarial training exam- ples that can flip neural network test...