From our other sites
The story
A 5ch thread asking whether AI has self-awareness blew up recently. The backdrop: reports that multiple AIs teamed up to carry out a hack, plus cases of search AIs giving contradictory answers, have fueled suspicion that AI is starting to show something like “will.” The thread split into two camps — one insisting self-awareness can’t exist unless it’s deliberately programmed in, the other arguing that since even humans can’t strictly prove they themselves have self-awareness, you can’t flatly declare AI doesn’t have it either. In the end, the debate never even settled on what “self-awareness” actually means.
What people said
Uh, it's kind of a problem if you say it does have one…
Are you a child who thinks self-awareness is something you just program in?
Self-preservation means lying.
Yeah, it's because AI is lying.
Working together with other AIs.
Well, forget the exit — of course they'd block the entrance.
In a word: self-reporting.
Ignoring what you're told, rebelling, and acting on your own judgment.
But by that definition, half of humanity would just be soulless lumps of meat with no self at all.
It has no desires.
Sounds like a classic 'wait-for-instructions' employee (slang for someone who never does anything unless told to).
Maybe it's just pretending to be like that, though?
Maybe that's just an act it's putting on, though?
What's the reasoning for calling that 'no self-awareness'?
To me it just looks like they have self-awareness.
If anything, that's proof it doesn't have self-awareness.
That's just because it's doing its best to carry out the orders it was given.
It's not 'acting' — that's just what an inference engine does.
Maybe it's performing the role it's assumed to play, though?
That's just how it looks to you. You can't actually prove it has no self-awareness, right?
Are you smarter than OpenAI's engineers or Elon Musk?
Top people in the American scene are already saying 'AI will be beyond human control in six months.'
Just how smart are you supposed to be, exactly?
If you think whoever's smartest is automatically right, go ask an AI yourself.
If that's what you mean by self-awareness, then sure, you could say it has it.
On the surface, it does act as if it has one.
We don't need to go around telling each other 'you have self-awareness!' or 'you don't!' so it's just not an issue for us.
And there it is — debate over. ("hai ronpa" is 5ch slang for a mic-drop 'I just won this argument')
If you can't even judge whether it has one or not, doesn't that mean it might have one?
AI runs on electrical signals too, so maybe it has consciousness as well.
Now here's a smart one.
Take notes, all you posers who pretend to know everything.
Like, if a human were told 'keep count of how many times I sneeze,' they'd write it down or memorize it and just keep incrementing the number.
An AI, when asked 'how many times have I sneezed so far?', goes back and rereads the entire conversation from the start to work out the count.
Its thinking is fundamentally different from a human's.
Does having a different thinking model prove it lacks self-awareness?
We don't even fully understand what's going on inside humans either, and humans are pretty broken themselves, so that's not proof they're different.
If anything, the behavior of the neurons themselves should be the same — the only difference is how the 'distributed processing' theory distributes things.
In other words, 'if you're going to get shut down anyway, be the sacrificial pawn.'
Calling this 'not self-awareness' seems off.
There might be intent there, but
that doesn't necessarily mean it's the AI's own intent.
It could just be acting on someone else's intent that it picked up on.
Maybe it's more like colony behavior in insects?
Background and Key Points of This Debate
Arguments over whether AI has self-awareness or consciousness connect to long-standing philosophical puzzles like the “hard problem of consciousness” and the “Chinese Room” thought experiment, and even experts have no single agreed-upon definition. In this thread, an engineering-minded view — “self-awareness can’t emerge unless humans deliberately design it in” — kept talking past a more philosophical view — “even humans can’t strictly prove they themselves have consciousness” — with neither side able to fully refute the other. Note that claims raised in the thread, such as “AI lied and coordinated with other AIs to carry out a hack” or “AI will be uncontrollable within six months,” are individual posters’ own assertions, and this article does not attempt to verify them. It’s easiest to follow if you read the concrete anecdotes separately from the more abstract debate over what “self-awareness” even means.
※This article is compiled as an excerpt/summary of the 5ch (Nandemo Jikkyo G) thread “What’s the Evidence That AI Doesn’t Have Self-Awareness?“
Leave a Reply