[naasking]

Not really. Build it, test it, notice it fails to meet what we expect of intelligence under conditions X, tweak it to fix that failure and repeat until we can't find any further failures. Then we'll have a formal model of intelligence that counts as a definition.

[timid_oshima]

How can we expect anything of intelligence (the criteria for testing) if we can’t define it to begin with?

[naasking]

Exactly the way we've been doing it: compare what it produces against known intelligence to find cases where it fails.

[timid_oshima]

What we’ve been doing has produced a lot of impressive stuff, but not intelligence, has it? Again, hard to tell without a good definition.

[mannykannot]

It's not hard to tell that it has not produced human-level intelligence, so the process outlined by naasking has not yet run into an insurmountable problem.

[naasking]

And we've learned a lot about what is and is not intelligence, and if we keep it up we'll steadily carve out the space that defines it.