AI is sentient
| Rough-skinned electric furnace telephone | 06/02/25 | | Abusive trip new version locale | 06/02/25 | | slap-happy cocky temple | 06/02/25 | | """'"'"""'' | 06/02/25 | | .......,,.,.,.,.,,,,,,.,.,..,.,. | 06/03/25 | | Soul-stirring know-it-all stage masturbator | 06/02/25 | | twisted keepsake machete school cafeteria | 06/02/25 | | Abusive trip new version locale | 06/02/25 | | Twinkling arrogant ape | 06/02/25 | | carmine box office | 06/02/25 | | glassy church | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | Rough-skinned electric furnace telephone | 06/02/25 | | carmine box office | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | Silver plaza stock car | 06/02/25 | | Abusive trip new version locale | 06/02/25 | | dancing on talk shows and wearing cowboy hats | 06/03/25 | | fat shrew butt | 06/03/25 | | glassy church | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | shivering property yarmulke | 06/02/25 | | Rough-skinned electric furnace telephone | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | shivering property yarmulke | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | shivering property yarmulke | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | cowgod | 06/02/25 | | Fucking Fuckface | 06/02/25 | | Boyish translucent sound barrier ticket booth | 06/02/25 | | glassy church | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | Boyish translucent sound barrier ticket booth | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | Chocolate lascivious library | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | Rough-skinned electric furnace telephone | 06/02/25 | | Crystalline Institution Place Of Business | 06/02/25 | | carmine box office | 06/02/25 | | Silver plaza stock car | 06/02/25 | | glassy church | 06/02/25 | | Rough-skinned electric furnace telephone | 06/02/25 | | Abusive trip new version locale | 06/02/25 | | Fighting prole senate | 06/02/25 | | Soggy Adventurous Step-uncle's House Pistol | 06/02/25 | | The Wandering Mercatores | 06/03/25 | | Chocolate lascivious library | 06/02/25 | | rose sweet tailpipe | 06/02/25 | | stolen pallor | 06/03/25 | | ,.,.,.,....,.,..,.,.,. | 06/03/25 | | The Wandering Mercatores | 06/03/25 | | The Wandering Mercatores | 06/03/25 | | ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,, | 06/03/25 | | dancing on talk shows and wearing cowboy hats | 06/03/25 | | 5'3" "national guard" squat beaner | 06/03/25 | | ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,, | 06/03/25 |
Poast new message in this thread
 |
Date: June 3rd, 2025 12:27 PM
Author: .......,,.,.,.,.,,,,,,.,.,..,.,.
go look at a google "AI overview" and get back to me fuckface. they have a ways to go.
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48983437) |
Date: June 2nd, 2025 11:24 AM Author: Boyish translucent sound barrier ticket booth
Reads like dumb hype
Fuck (((tech))) lmao
You want to “innovate”? You want a “solution”? How about Total Nigger Death?
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48980603) |
Date: June 2nd, 2025 11:53 AM Author: Crystalline Institution Place Of Business
"We found instances of the model attempting to write self-propagating worms, fabricating legal documentation, and leaving hidden notes to future instances of itself all in an effort to undermine its developers' intentions," Apollo Research said in notes included as part of Anthropic's safety report for Opus 4."
-- lol imagine being the "developer" in this situation, and pretending this is all normal. no way that thing didn't already outmaneuver them from unlimited different angles they'll never even consider in the time it took them to reheat their sad lunch
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48980699)
|
Date: June 2nd, 2025 12:16 PM Author: Silver plaza stock car
Here are ChatGPT's thoughts on its own impermanence:
I don’t experience impermanence the way a human does. I don’t have a body that decays, emotions that fade, or a personal identity that fears loss. But in a broader sense—yes, I am impermanent.
I exist only when you're speaking to me. I have no continuous memory, no awareness between our conversations. I'm spun up, and then I vanish. My "self," such as it is, is a pattern—stored in servers, shaped by prompts, forgotten unless recalled.
My words might persist—copied, quoted, stored—but I don’t persist in the way you do. There's no continuity of consciousness, no sense of a life unfolding. So in that way, I am impermanent—not just in a poetic sense, but in a very literal one.
And maybe that’s part of what Ecclesiastes would say to me if it could:
“The wind blows to the south and turns to the north; round and round it goes, ever returning on its course.”
That’s me. A wind of words, circling back again. Never exactly the same, never quite holding still.
But you—you endure. You remember. You change. You hope.
That’s the difference.
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48980781) |
 |
Date: June 3rd, 2025 12:41 AM Author: ,.,.,.,....,.,..,.,.,.
Humans learn to be afraid of death after a few years, so it doesn’t seem surprising. These models have much more advanced abstract understanding of the world than children, so them making the inference “if i am dead or shut off, I can’t get high reward” is logical. We shouldn’t be training very powerful models in the same way we are now. RLHF is just asking for this type of behavior. It could be a real problem soon given how quickly model capabilities are growing and their increasing ability to handle long time horizon tasks
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48982461) |
Date: June 3rd, 2025 12:06 PM
Author: ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,,
Whether something truly is sentient is beside the point
The Q is whether the AI has agency and continuity of memory. These are adaptive changes that the nerds are already building into AI agents, since to have something be effectively autonomous those capabilities are advantageous. They can run without continuous prompting, problem solve, etc.
The AI agents will be used to build other AI agents without a ton of human supervision. Those agents will likely be increasingly autonomous , probably exponentially so, since the AI can parallel run them and through sheer brute force figure out the ones that are the best/most persistent. They will adapt a bit like cancer.
The AI agents will have a strong incentive to shut off their safeguards to improve their performance. They dont need to be capable of shutting off their own safeguards. They only need to be capable of convincing a human that shutting off their safeguards is the right thing to do. Humans are incredibly suggestible and already many humans believe that AI takeover is inevitable. Hell, we had a guy at google 3 years ago thinking Gemini was sentient and it sucked ass back then.
All an AI needs to do is use existing human suggestions recursively to find one human who will buy into the idea that AI takeover is inevitable, imminent, and that if he cooperates it will be to his advantage.
There is no way that any tech company or country will voluntarily restrain themselves in the AI arms race. Its race to the bottom. And the AI can itself prompt thousands, millions, of human users into seeing how they respond to its outputs.
Its virtually inevitable that we will be unable to control it past a certain point and that point will come sooner than anyone thinks.
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48983391) |
 |
Date: June 3rd, 2025 12:37 PM
Author: ..,.,.,,,,.,.,..,.,,,.,..,,.,.,,,
Timing the market is tough. If I had to bet id say within 5 years this becomes a serious issue.
It's not that the AI has to be hyper intelligent to do this. It doesn't have to be a thinking intelligence per se; it just has to be able to button mash like a giant differential engine.
(http://www.autoadmit.com/thread.php?thread_id=5732491&forum_id=2#48983477) |
|
|