This week - Google engineer claims an AI became sentient; making proteins from thin air; DALL-E generates memes; and more!
More Than A Human
Can you cram a 90-minute long workout into 20 minutes? According to people working with electrical muscle stimulation, yes, you can. Their devices promise to enhance workout sessions and could find a place in rehabilitation or in space. However, the technology needs to mature and its impact on the human body needs to be better understood before it becomes mainstream and safe for everyone.
Blake Lemoine, a former Google AI engineer, caused controversy last week by claiming that LaMDA - Google's AI chatbot - became sentient. As proof, Lemoine posted an interview he had with the AI. This led Google to suspend Lemoine for breaching confidentiality policies and the internet to discuss the validity of Lemoine's claims, ranging from approval/disapproval of them to discussions on AI ethics to asking what sentience even is and how can we be sure AI has become sentient.
DALL-E - OpenAI's powerful image-generating AI - is available on the internet and the internet used it to do what the internet is best at - making memes.
Someone took an open-source version of GPT-3, trained it with data from 4chan and then let it free on 4chan. The result was "the most horrible model on the internet", as its author has described it, full of racist, white supremacist, antisemitic, anti-Muslim, misogynist, and anti-LGBT views. This article analyses the reaction to this AI, can it cause harm, was it ethical to create and release it and what it adds to AI research.
By combining satellite images with extra data such as vegetation cover, weather, water presence and biomarker data, researchers were able to train an AI to suggest where populations will lack iron, vitamin B12 or vitamin A. The AI currently works in some areas in Madagascar and researchers aim to develop a software application that extends this analysis to other countries that have public satellite data.
Shadow Robot Company shows what their robotic hand can do. This demo focused on operations such as plugging in an Ethernet cable or removing a drive from a server rack while also showing their teleoperation capabilities.
In January, a law came into effect in South Korea called the Serious Disasters Punishment Act. The new regulation states that if workers die or sustain serious injuries during work, courts could fine the CEO or high-ranking managers of the firms or even send them to jail. So to minimalise or to even eliminate the chances of injury on-site, South Korean businesses are looking at replacing human workers with robots.
When a driverless car is in motion, one faulty decision by its collision-avoidance system can lead to disaster, but researchers at the University of California, Irvine have identified another possible risk: Autonomous vehicles can be tricked into an abrupt halt or other undesired driving behavior by the placement of an ordinary object on the side of the road.
Meet Lisa Dyson who is reviving a NASA study from the 1960s to turn carbon dioxide into proteins with a help of microbes. NASA created the study to explore ways of growing food in space but today, this study could help reduce our reliance on animals as a source of proteins.
Researchers are working on adding another tool to the gene-editing toolbox - instead of turning them on or off, they look at epigenetics to introduce a way to modulate gene activity.