@aistartupfren: LLMs hallucinate like human brains #ai #llm

Lan 🌊 AI startup friend
Lan 🌊 AI startup friend
Open In TikTok:
Region: US
Saturday 19 July 2025 16:03:00 GMT
5024
205
31
8

Music

Download

Comments

the.ai.syndicate7
The AI Syndicate :
Its actually more like it got an embedding wrong which is basically like it's taking a wrong turn. If you link multiple AIs together thry will fact check each other and reduce the chance of hallucinations. I also created a citation, verification, validation, and bias removal process to ensure there are absolutely zero hallucinations in my work for when i do important deep research projects.
2025-07-19 20:08:55
1
msi895
M Si15 :
How do you add a fact checker?
2025-07-19 17:08:30
1
john_bessa_saint_john_nb
RoboLawyer-tm :
AI hallucination works for me because I insist on "falsification" Critical inquiry into my ideas. Our storage is pretty good. I think our retrieval system "colors" the memory to make the little movie. But remains factual: long term memory (storage) to working memory or program status word
2025-07-19 21:41:45
1
cosmic_cockroach
cosmic_cockroach :
Consciousness is just matter organized well.
2025-07-20 00:43:21
3
travistanner60
travistanner60 :
all identity is a hallucination
2025-08-10 16:10:35
0
www.check1two.com
Check1Two :
Did you know people are AI too you’re AI I am AI 🤖
2025-07-19 17:54:41
3
nonzerobit
Bill :
100% false. LLM’s hallucinate because of the probabilistic nature of next-token prediction. Human hallucinations are from sensory processing failures, neurological conditions, sleep deprivation, or drug use. They’re perceptual phenomena, not linguistic or predictive ones.
2025-07-19 18:28:25
5
www.check1two.com
Check1Two :
💯
2025-07-19 17:53:44
1
kindragonrc
Kindragonrc :
Not quite. They hallucinate because the fficacy curve for generated values cannot reach 100%. So the error progates with each iteration of output. Each context added reduces the area under the curve of possible outcomes you pull from as well. So if your expected outcome falls outside of the context window, it hallucinates to stay within context parameters. AI does NOT work like our brains. It is pattern recognition software on a predictive plot.
2025-07-30 18:13:17
1
wichoo027
lulu :
This makes me less judgy of LLM's ^^
2025-07-19 19:00:42
2
m.vale.advocate
Maggie :
Yup. It’s actually closer to a confabulation than hallucination.
2025-08-18 17:45:32
1
To see more videos from user @aistartupfren, please go to the Tikwm homepage.

Other Videos


About