Gary Marcus

Rank 16 of 47

Score 71

Reposted by Gary Marcus

Steven Pinker

@sapinker

Why hallucinations are baked into the architecture of Large Language Models (a point also made by @GaryMarcus)

Wendy

@wendyweeww

28d

Hallucination is baked into LLMs. Can't be eliminated, it's how they work. @DarioAmodei says LLMs hallucinate less than humans. But it's not about less or more. It's the differing & dangerous nature of the hallucination, making it unlikely LLMs will cause mass unemployment
(1/n)

7/26/2025, 6:22:25 PM

The statement and accompanying conversation discuss the inherent nature of hallucinations in Large Language Models (LLMs) and their implications. It highlights the architectural reasons for hallucinations and compares them to human cognition, emphasizing the challenges in eliminating such behavior. The conversation aims to inform and clarify misconceptions about LLMs, contributing to public understanding of AI technology.

Principle 1:
I will strive to do no harm with my words and actions.
The statement aims to inform and clarify the nature of LLM hallucinations, striving to do no harm by providing accurate information. [+1]
Principle 3:
I will use my words and actions to promote understanding, empathy, and compassion.
The conversation promotes understanding by explaining complex AI concepts in an accessible manner, fostering empathy towards the limitations of LLMs. [+2]
Principle 4:
I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
The discussion engages in constructive dialogue by addressing potential misconceptions and providing a balanced view of LLM capabilities and limitations. [+1]