The Download: Amsterdam’s welfare AI experiment, and making humanoid robots safer
1 min read
Summary
A new investigation has been produced by MIT Technology Review working in conjunction with Lighthouse Reports and Dutch newspaper Trouw to discover why Amsterdam’s welfare scheme AI project failed.
The investigation reveals that there were inadequate safeguards in place, with some reviewers working on cases they had personal experience of, while others had clear conflicts of interest as they were involved in implementing the technology.
Other issues included lack of clarity over cases, leading to queries not being answered.
There was also a lack of diversity within the review team, with over 70% being white and aged over 40.
One of the most interesting AI applications is in the military; in Ukraine, it’s being used to guide drones and in the face of the Russian payload, this has proved critical.
It has also emerged that US agencies were monitoring visits to Elon Musk by foreign nationals in 2022 and 2023.
And finally, MIT Technology Review has analysed the water consumption of ChatGPT and estimates it would require 15mx more water than is used in NYC each year.