Nature Podcast

Researchers have developed an AI tool that can calculate a person’s risk of developing over 1,000 different diseases, sometimes years in advance. The system, called Delphi-2M, was trained to identify patterns of disease progression using 400,000 people's health records from data repository the UK Biobank. This training allowed it to predict someone’s future disease risks, based on their current medical record. While AI health prediction systems do exist, they typically only estimate risks for a single disease — the authors hope that their system could one day save healthcare professionals time and be used to calculate disease burdens at a population level.

Research Article: Shmatko et al.

News: What diseases will you have in 20 years? This AI makes predictions

Evidence that refugees hosted by local families integrate better into their adoptive country — plus, the squidgy shirt that can keep wearers cool.

Research Highlight: How to help refugees thrive: have local families host them

Research Highlight: Jelly-filled garment keeps wearers cool when heat and humidity soar

Using AI tools may make you more likely to cheat at tasks like tax reporting, according to a new study. Using a well-studied test of honesty, researchers looked to see if people were more likely to engage in unethical behaviour if given the option of delegating it to an AI. Including AIs seemed to increase the chance that someone would be dishonest, which raises concerns about the impacts of these tools on ethics.

Research Article: Köbis et al

News and Views: People are more likely to cheat when they delegate tasks to AI

Europe has a new supercomputer, JUPITER, that could boost its AI ambitions, and a catalogue of octopus movement.

Nature: World's most energy-efficient AI supercomputer comes online

New York Times: Building an Octopus Dictionary, One Arm Movement at a Time

Subscribe to Nature Briefing, an unmissable daily round-up of science news, opinion and analysis free in your inbox every weekday.