Focus: AI

01 Random nanowire networks – a type of computer made from crisscrossed microscopic wires – work best when stimulated “at the edge of chaos”: they produce their most interesting results when just slightly more stimulation would make them totally unpredictable.

02 Wind farms can now use an algorithm to track tiny changes in the weather and predict how much energy they’ll be producing in the next five minutes. This helps them save money because they can tell more accurately how much electricity they’ll be able to supply to the grid.

03 An AI program called a “convolutional neural network” classified galaxy shapes with an accuracy of 80% – similar to human accuracy, because it was trained on human-generated data. But it was much faster, labelling 14,000 galaxies in three seconds.

04 With the help of two electrodes in his brain connected to an AI, a man who’d been paralysed from the neck down was able to “handwrite” on a computer screen.

05 Australian researchers built a machine-learning program that, on a laptop, does the job of a supercomputer. They’re using it to find new candidate materials for solar panels.

06 Vertigo affects 40% of people at some point in their lives, but it’s very difficult to diagnose which medical condition is causing it. Researchers are working on an AI program that will use data from eye movements and other symptoms to help with diagnosis.

07 Anaemic people have different colouring on the insides of their eyelids: so different, in fact, that an algorithm can spot anaemia from smartphone photos of eyelids.

www.cosmosmagazine.com/ 01 technology/ai/artificial-intelligence-at-the-edge-of-chaos/ 02 technology/ai/renewables-make-it-into-thegrid-better-with-ai/ 03 technology/ai/an-algorithm-that-labels-galaxies/ 04 technology/ai/paralysed-man-handwrites-with-brain-chip/ 05 technology/ai/machine-learning-makes-solar-energy-efficient-cheap/ 06 health/medicine/addressing-vertigo-with-ai/ 07 technology/ai/ai-can-spot-anaemia-from-photos-of-eyelids/