Cool new work from @patrickbissett.bsky.social on the sharing of control between humans and machines. tl;dr: it makes us worse.
Reposted from
Patrick Bissett
New preprint (osf.io/preprints/ps...) evaluating how shared control environments, like partially automated cars, influence human cognitive control. In short, response inhibition was impaired in multiple ways when subjects were told that an AI will usually stop for them, but infrequently it doesn’t.
Comments
On the other hand, I am confident the driving of my car is more safe when I use openpilot (and supervise it) than when I drive without openpilot.
Note: I choose to use openpilot in uncomplicated highway situations, but it never has attention lapses of any magnitude.