Comment on Why AI Needs a “Genie Coefficient” by Weather
Why AI Needs a “Genie Coefficient” This essay was written with Barath Raghavan, and originally appeared in IEEE Spectrum. None measure whether it does what you mean: the distance between what you ask an AI to do and the unspoken assumptions about how you want the AI to do it. We propose a new metric: the Genie coefficient.
- ▪Why AI Needs a “Genie Coefficient” This essay was written with Barath Raghavan, and originally appeared in IEEE Spectrum.
- ▪None measure whether it does what you mean: the distance between what you ask an AI to do and the unspoken assumptions about how you want the AI to do it.
- ▪We propose a new metric: the Genie coefficient.
Schneier on Security » Comments Feed files mainly under news. We currently carry 3 of its stories.
Opening excerpt (first ~120 words) tap to expand
Why AI Needs a “Genie Coefficient” This essay was written with Barath Raghavan, and originally appeared in IEEE Spectrum. Major benchmarks measure what AI can do. None measure whether it does what you mean: the distance between what you ask an AI to do and the unspoken assumptions about how you want the AI to do it. We propose a new metric: the Genie coefficient. There’s often a gap between one person’s request and another’s understanding. Most of the time, we bridge it using general knowledge. For example, if you ask a friend to get you coffee, they’ll pour a cup from the pot or buy one from a coffee shop. They won’t bring you a bag of raw beans or snatch a cup from a stranger and hand it to you. You never specified any of this. You never had to.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Schneier on Security » Comments Feed.
Discussion
0 comments
