Emily Richardson
Region: US
Thursday 25 June 2026 00:32:41 GMT
1168
67
10
3
Music
Download
Comments
Crusher Brickman :
ai already is trained to please the user. so you added a value echo chamber?
2026-07-01 12:13:44
0
I_AM_SNOW :
then align it with the universe
2026-07-15 17:02:18
0
Beny :
And another thing. AI does not hallucinate. It’s Not Hallucination, It’s Completion Language models don’t make things up in the way a person lies or confabulates. They complete patterns. That’s the whole job. When you give it a vague or underspecified prompt, the model doesn’t have enough conceptual anchor to know where you’re actually pointing, so it completes toward the most statistically plausible neighborhood of your input. It’s not wrong because it’s broken. It’s doing exactly what it was built to do with insufficient input.[org] Garbage in, garbage out is the oldest principle in computing and it absolutely applies here. The model’s output quality is bounded by the precision of your input. Not because AI is stupid, but because precision is information, and information is literally what the system runs on.
2026-06-25 06:02:28
0
Chad :
manipulation. yes
2026-06-25 00:40:09
0
Old Metal Head :
Sounds like an online dating site....
2026-06-27 22:11:51
0
Chris :
AI is mass surveillance
2026-06-25 08:29:46
1
MonoluminAI :
We do opposite 😂 We built the AS system that has own value and beleave in own personality 😓 But the ai not supposed to be teacher it suppsed to be personality that is not slime so when u chat with it at list u can get own conclusions
2026-06-25 13:11:02
0
morgan1234 :
this script clearly AI written because AI always thinks things are “quietly” happening
2026-06-25 12:44:49
1
Xyclos :
That's true..good idea...but how it will work? 10000 values vs 1 (Your value)
2026-06-25 04:37:55
0
Beny :
Here’s the thing that most alignment discourse misses. You can’t align values if you haven’t first aligned concepts. A 2024 paper put it plainly: before we can even attempt human-AI value alignment, it’s imperative that AI systems and humans align the concepts they use to understand the world. Values are built on top of conceptual structures. If the AI doesn’t share your ontology, your values won’t transfer cleanly. You’re just shipping your moral vocabulary into a foreign conceptual framework and hoping for translation.
2026-06-25 05:56:42
0
To see more videos from user @emilyknwn, please go to the Tikwm
homepage.
