This week - OpenAI's AI learns to play Minecraft from YouTube; a robotic vest that acts as an extra layer of muscles; a robotic skin that feels pain; and more!
More Than A Human
Researchers at ETH Zurich have developed a wearable textile exomuscle that serves as an extra layer of muscles. Named Myoshirt, the vest is packed with sensors feeding an algorithm that predicts the wearer’s intentional movements and the amount of force required and assists in the movement. The aim is to use the vest to increase the upper body strength and endurance of people with restricted mobility.
UK government released a report looking into the impact of AI on intellectual property, copyright and patent laws and how to deal with it. The report states there will be no changes in the law for computer-generated works and AI-devised inventions, citing that the technology is still not advanced enough to justify any changes. But for text and data mining (TDM), the UK government is planning "a new copyright and database right exception which allows TDM for any purpose". Rights holders will still have safeguards to protect their content, including a requirement for lawful access.
A new AI from OpenAI has learned how to play Minecraft just by watching Minecraft videos on YouTube. The AI learned not only basics like chopping the trees and building a crafting table but also learned how to make a diamond pickaxe and build shelters.
In the journal Nature Climate Change, a team of experts in AI, climate change, and public policy present a framework for understanding the complex and multifaceted relationship of AI with greenhouse gas emissions and suggest ways to better align AI with climate change goals. The conclusion is that AI impacts climate change in both good and bad ways and that most of these effects are poorly understood.
This article argues the cause of recent claims that Google's AI has become sentient is nothing more than a bug in human cognition. This bug makes us associate coherent, fluent speech with fluent thought, hence some may see modern AI text-generators as sentient.
Almira Osmanovic Thunström asked GPT-3 to write an academic paper about itself and shares the journey from the idea to typing the command for the AI to contemplating all the legal and ethical issues that arise from this experiment.
Researchers have built an artificial skin that can react to what it touches. Then they trained a robot wearing it what is pleasant and is not pleasant to touch, effectively making the robot feel "pain". And their skin reacts faster than any other approach thanks to offloading some processing away from a central computer to the skin itself.
CRISPR is a powerful gene-editing tool which has more in common with a butcher knife than with scissors it is often compared with. New iterations made CRISPR more precise and less invasive but they still have to cut the genome. However, new research on Cas12c variant shows it is possible to use CRISPR without damaging the DNA strand.