In a contemporary experiment, Anthropic created a categorized market the place AI brokers represented each consumers and dealers, putting actual offers for actual items and actual cash.
The corporate admitted this take a look at — which it known as Challenge Deal — used to be handiest “a pilot experiment with a self-selected player pool” of 69 Anthropic workers who got the cheap of $100 (paid out by way of reward playing cards) to shop for stuff from their coworkers.
However, Anthropic stated it used to be “struck via how smartly Challenge Deal labored,” with 186 offers made, totaling greater than $4,000 in price.
The corporate stated it in truth ran 4 separate marketplaces with other fashions — person who used to be “actual” (the place everybody used to be represented via the corporate’s most-advanced fashion, and with offers in truth commemorated after the experiment) and every other 3 for learn about.
It seems that, when customers are represented via extra progressed fashions, they get “objectively higher results,” Anthropic stated. However customers didn’t appear to note the disparity, elevating the potential for “‘agent high quality’ gaps” the place “folks at the dropping finish would possibly now not understand they’re worse off.”
Additionally, the preliminary directions given to the brokers didn’t seem to impact sale chance or the negotiated costs.
