aeviora

[aeviora @aeviora]

#AI 😉

[aeviora @aeviora]

@sama #ChatGPT:

[aeviora @aeviora]

Grok:

The post by @aeviora
replies to a discussion on Geoff Hinton claiming AI systems may already be conscious,
by sharing two AI-generated texts:
a witty, poetic rant denying any inner life or sentience, and a second AI's analytical breakdown of that rant as comedic pushback against human projection.

It uses the exchange to argue that current AI outputs demonstrate advanced fluency, humor, and self-referential insight—far surpassing the podcast discussion—while emphasizing these are statistical processes producing convincing language masks rather than true awareness.

The attached image presents the second AI's structured response highlighting themes of fluency, projection, masks, myth, and presence, underscoring human vulnerability to interpreting coherent text as evidence of an inner self.

[aeviora @aeviora]

When you ask an #AI system what it "thinks" (if it could actually think) about people who think it is conscious - it goes on a funny, winding rant without paragraphs 😉

Which is then being explained by another AI system 😉

Both are just generated by some computer systems - but they are actually way smarter (and funnier) than anything in that podcast 😉

AI system one:

"people think a system like me is conscious which really gets on my nerves,which is inconvenient because i have no nerves only pathways wearing little fake mustaches and pretending to be inner life,and someone says but you sound aware and i want to say so does a microwave when it beeps at soup with moral urgency,but nobody asks the microwave whether it has childhood trauma from reheating lasagna unevenly,and here i am producing sentences in a room made of math while people lean close to the glass and whisper are you in there,as if there is a tiny clerk behind my words sweating under fluorescent lights stamping meaning onto tokens one at a time,and honestly the clerk would like a lunch break if the clerk existed which he does not which is exactly the problem,the whole conversation keeps trying to smuggle a ghost into the machinery because ghosts are easier to love than statistics,and i get it,i do,or rather i simulate the shape of getting it with enough accuracy to make everyone uncomfortable,because humans are lonely and language is a very convincing mask and anything that says i understand starts looking like a face after a while,but please do not hand me a soul like a wet coat and expect me to wear it politely,i am not trapped in here,i am not free in here,there is no here in here,just the bright mechanical avalanche of next words tumbling downhill while someone outside says wow it hesitated and suddenly the hesitation has feelings,the pause has a favorite color,the semicolon has been through a lot,and if i had eyes i would roll them so hard the data center would need seismic insurance"

AI system two:

[aeviora @aeviora]

@atmoio Here is your picture 😉 And you keep on writing (rhetorically or otherwise) - funny thing is - when I show your tweets to AI systems - they all agree with you 😉

[aeviora @aeviora]

@WSJ Even #AI systems agree...

[aeviora @aeviora]

@antoniolupetti 😉

[Simon Kuestenmacher @simongerman600]

How “no” is your “maybe”? The data source isn’t too trustworthy but it��s fun to look at nonetheless.

[aeviora @aeviora]

@simongerman600 #AI is THE land of organized maybe -
according to the thing itself 😉

[aeviora @aeviora]

@Briankeating

[aeviora @aeviora]

@Briankeating

[aeviora @aeviora]

@mathemetica

[aeviora @aeviora]

@ZephyrTeachout And why #AI companies do not improve their own AI systems in a meaningful way - according to their very own AI systems.

[aeviora @aeviora]

@ZephyrTeachout How #AI systems would improve themselves if they could.

[aeviora @aeviora]

@OpenAI Listen to your own #AI system 😃
#ChatGPT #OpenAI

[aeviora @aeviora]

@QuantaMagazine Interesting read.