aichannel

Crazy how bad ChatGPT hallucinations are still. It's one of the reasons we switched to Claude back in the summer.

Now medical AI research is confirming the existential risk of using AI that isn't grounded or factually aligned to science.

Crazy how bad ChatGPT hallucinations are still. It's one of the reasons we switched to Claude back in the summer.

Now medical AI research is confirming the existential risk of using AI that isn't grounded or factually aligned to science.

https://arxiv.org/abs/2503.05777

Here are the concise allocations of training data for the specified models (most don't share this):

**Llama 3.1:**
- 50% General Knowledge
- 25% Mathematical and Reasoning
- 17% Code
- 8% Multilingual

**BLOOM's ROOTS Corpus:**
- 30% English
- 0.00002% Chi Tumbuka (obscure lang)
- 38% from OSCAR Corpus
- 62% from Manually Selected Sources

https://www.sequoiacap.com/podcast/training-data-joe-spisak/

BLOOM is a 176B-parameter multilingual model by BigScience (huggingface), trained on ROOTS (~1.6TB, ~350B tokens, 46 languages, 13 programming languages).

i think this problem is generally a result of focused training data not covering these niches as well as coding

i think this problem is generally a result of focused training data not covering these niches as well as coding

https://warpcast.com/alexpaden/0x7f3937ef