Infrahuman, par-human, superhuman, efficient, optimal

by Eliezer Yudkowsky Jan 29 2017 updated Mar 8 2017

A categorization of AI ability levels relative to human, with some gotchas in the ordering. E.g., in simple domains where humans can play optimally, optimal play is not superhuman.


These thresholds aren't always ordered as above. For example, logical Tic-Tac-Toe is simple enough that humans and AIs can both play optimally; so, in the Tic-Tac-Toe domain, optimal play isn't superhuman.]

Some thresholds in 'sufficiently advanced' machine intelligence are not absolute ability levels within a domain, but abilities relative to the human programmers or operators of the AI. When this is true, it's useful to think about relative ability levels within a domain; and one generic set of distinguished thresholds in relative ability is:

The ordering of these thresholds isn't always as above. For example, in the extremely simple domain of logical Tic-Tac-Toe, humans can play optimally after a small amount of training. Optimal play in Tic-Tac-Toe is therefore not superhuman. Similarly, if an AI is playing in a rich domain but still has strange weak spots, the AI might be strongly superhuman (its play is much better and shocks human masters) but not efficient (the AI still sometimes plays wrong moves that human masters can see are wrong).

The term "human-equivalent" is deprecated because it confusingly implies a roughly human-style balance of capabilities, e.g., an AI that is roughly as good at conversation as a human and also roughly as good at arithmetic as a human. This seems pragmatically unlikely.

The other Wiki lists the categories "optimal, super-human, high-human, par-human, sub-human".

Relevant thresholds for AI alignment problems

Considering these categories as thresholds of advancement relevant to the point at which AI alignment problems first materialize:

'Human-level AI' confused with 'general intelligence'

The term "human-level AI" is sometimes used in the literature to denote Artificial General Intelligence. This should probably be avoided, because:

Arguably, the term 'human-level' should just be avoided entirely, because it's been pragmatically observed to function as a gotcha button that derails the conversation some fraction of the time; with the interrupt being "Gotcha! AIs won't have a humanlike balance of abilities!"


Ryan Carey

subhuman should contrast with superhuman, or infrahuman with suprahuman.