Sources
Loading...
Additional media
Loading...
DeepNewz, mobile.
People-sourced. AI-powered. Unbiased News.
Researchers are exploring ways to mitigate hallucinations in Large Language Models (LLMs) by gaining control over them rather than eliminating them. Hallucinations in LLMs are compared to human experiences of hallucinations and schizophrenia. The debate revolves around whether LLMs can be considered Artificial General Intelligence (AGI) despite these issues.
People who hallucinate words and entire sentences when speaking are diagnosed with schizophrenia. If you say that LLMs are AGI despite hallucinations (humans hallucinate too!) you are saying that being a schizophrenic is not a medical condition.
People who hallucinate words and entire sentences when speaking are diagnosed with schizophrenia. If you think that LLMs are AGI despite hallucinations (humans hallucinate too!) you are saying that being a schizophrenic is not a medical condition.
Fixing the problem of hallucinations in LLMs isn’t about eliminating them, but gaining control over them. Avoiding hallucinations is essential for some cognitive tasks, and intentionally dipping into them is essential for others which are just as important to creating knowledge. https://t.co/3AQGYvuQdS
