XPost: alt.folklore.computers
From: nunojsilva@invalid.invalid
On 2026-01-02, Lawrence D’Oliveiro wrote:
> On Fri, 2 Jan 2026 02:53:45 -0000 (UTC), Waldek Hebisch wrote:
>
>> In alt.folklore.computers Lawrence D’Oliveiro wrote:
>>>
>>> The only succinct definition of “AI” I ever saw was: “solving NP
>>> problems in polynomial time”.
>>
>> Well, for me AI is process (and its results) of trying to solve
>> problems that we can not solve using known (at given time) methods
>> and which seem to require inteligence.
>
> You don’t see crossing the P/NP divide as being a good indication of
> such a distinction?
(Someone please correct me if I'm handling some concept less than
optimally:)
AFAIK: No, because that'd not be a characteristic of AI, but more likely
ground-breaking findings that'd affect all of computer science?
In short: I think you may be mistaking NP's definition for a way to
reduce NP problems to P ones. - What you hint at seems to really be the
definition of NP itself?
A strategy which falls under the AI label (the one used for decades, not
the one in the GenAI hype) may be able to handle NP problems quicker
within certain likelihoods, that's not making them P, that's just
getting to solutions "at random", that can then be *verified* in
polynomial time.
The problem itself would still remain in NP, even if such algorithms
have practical applications.
The ability to address such problems in that way may be a good
description of at least some of these AI techniques, but I think saying
they're "solved in polynomial time" is taking it too far. Also, doesn't
AI also include other techniques which aren't focused on this, such as
pattern identification in datasets?
--
Nuno Silva
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
