Your Subscription to AI WEEKLY is changing, here is what's new.
If we offered AI Weekly Pro - a premium newsletter for a small fee, which of these features would be most beneficial to you? Click here to take part in a brief survey.
In the News
How AI Is Being Transformed by ‘Foundation Models’
A few month ago, scientists and engineers described during a conference held at Stanford how AI research has succeeded in creating models that are generic. Those models called Foundation models use an enormous, single data set and can perform a variety of different tasks with relatively little programmer input rather than being tailored to a single task and dataset. This article reflects on the debates around these foundation models.
Sponsor
What is the easiest way to curate your training data?
Don’t miss our next tech talk where we’ll cover how to slice and dice 3D data with only a text search, speed up dataset creation pipeline, even with challenging data types like 3D and how to filter your dataset with plain old words instead of writing queries with precise syntax.
In The News
Symposium on AI and Climate Action kicks off at Tsinghua University
The healthy development of AI can better serve the community of life for man and nature said global experts at the Symposium on Artificial Intelligence and Climate Action on April 26, 2022.
How AI can help fight misinformation
Advances in artificial intelligence (AI) could increasingly help online information users sort out fact from fiction. The Global Disinformation Index (GDI) collects data on how misinformation – or disinformation, when deliberate – travels and spreads.
Building Trustworthy AI With Mozilla
In this inspiring episode, Temi Popo, Chenai Chair, Kathy Pham, and Wiebke Toussaint discuss the different ways Mozilla is building trustworthy AI on a global scale, and so much more.
Applied use cases
Meta has built a massive new language AI—and it’s giving it away for free
Meta’s AI lab has created a massive new language model that shares both the remarkable abilities and the harmful flaws of OpenAI’s pioneering neural network GPT-3. Meta is making its model, called Open Pretrained Transformer (OPT), available for non-commercial use. The company sparked controversy in...
Deep Learning in Neuroimaging
A few examples of imaging techniques that do not require surgery include macroscopic imaging techniques such as functional magnetic resonance imaging (fMRI) or approaches with a high temporal resolution such as electroencephalogy (EEG). The scales of neuroimaging techniques we discuss here range...
Ars Frontiers is next week—here’s what’s on tap at our first conference
The road to Ars Frontiers The week of the event, we’ll be kicking things off with a series of virtual streaming chats that take place from May 9 through May 11. One of our Road to Frontiers virtual streams will be about cryptography and privacy, featuring Lesley Carhart, director of incident...
Ethics
TechScape: This cutting edge AI creates art on demand – why is it so contentious?
In this week’s newsletter: Dall-E 2 can conjure vivid pictures of dogs in berets to astronauts playing basketball. The first was Google’s PaLM, a new language model (the same basic type of AI as the famous GPT series) that shows a pretty stunning ability to comprehend and parse complex statements –...
Future Superhuman: Our transhuman lives in a make-or-break century
That's the view which Elise Bohan, Senior Research Fellow at the Future of Humanity Institute, develops in her new book, Future Superhuman: Our Transhuman Lives in a Make-or-Break Century.
A Guide to the Role of Auditing in the Proposed European AI Regulation
The proposed European Artificial Intelligence Act (AIA) is likely to become an important reference point that establishes precedence in terms of how AI systems can be regulated. The two primary enforcement mechanisms proposed in the AIA, have been little studied, however.
Robotics
How robotics is changing the future of craft businesses
Automation is the megatrend in industry. But it has not yet caught on in the craft businesses. And this is despite the fact that many companies are facing problems such as a shortage of skilled workers or a lack of trainees.
This robot chef can taste salt with its arm
The Cambridge group weren’t new to the idea, having previously created a robot that could make omelettes and improve its egg-making prowess with human feedback. One British startup offers a set of robotic cooking arms, which costs over $300,000, that can make thousands of recipes—but it still needs...
TurtleBot
Built on top of iRobot’s Create 3 in close partnership with Open Robotics, the TurtleBot 4 is “the next-generation of the world’s most popular open-source robotics platform for education and research, offering superior computing power, more payload capacity, improved sensors, and a world class user...
Research
On the effect of pre-training corpora on in-context learning by large-scale language model.
Pre-training corpus There are seven sub-corpora in the HyperCLOVA corpus: Blog, Cafe, News, Comments, KiN, Modu, and Ency. The five tasks consist of binary sentiment classification, machine reading comprehension, Korean to English translation, English to Korean translation, and topic...
Galileo Launches to Give Data Scientists the Superpowers They Need for Unstructured Data Machine Learning
SAN FRANCISCO, May 03, 2022 (GLOBE NEWSWIRE) -- Today Galileo emerged from stealth with the first machine learning (ML) data intelligence platform for unstructured data that gives data scientists the ability to inspect, discover and fix critical ML data errors 10x faster across the entire ML...
Vision Transformers 1: Low Earth Orbit Satellites
At Myrtle.ai, we have been investigating whether Vision Transformers would make a good candidate algorithm for deployment on a Low Earth Orbit (LEO) satellite system. In this series of posts we describe how we applied compression techniques to transformer-based computer vision models and show how...
Sponsor
What is the easiest way to curate your training data?
Don’t miss our next tech talk where we’ll cover how to slice and dice 3D data with only a text search, speed up dataset creation pipeline, even with challenging data types like 3D and how to filter your dataset with plain old words instead of writing queries with precise syntax.