Mikel Artetxe
@artetxem.bsky.social
Co-founder of Reka & Honorary Researcher @hitz-zentroa.bsky.social (University of the Basque Country) | Past: Research Scientist at FAIR (Meta AI)
Don't LLMs from different labs feel suspiciously similar sometimes? 🧐
I asked 10 models to tell me a joke 100 times each. 90.3% of the answers boiled down to just 3 jokes, each tied to a specific GPT version 😵💫
1/4
I asked 10 models to tell me a joke 100 times each. 90.3% of the answers boiled down to just 3 jokes, each tied to a specific GPT version 😵💫
1/4
September 3, 2025 at 5:00 PM
Don't LLMs from different labs feel suspiciously similar sometimes? 🧐
I asked 10 models to tell me a joke 100 times each. 90.3% of the answers boiled down to just 3 jokes, each tied to a specific GPT version 😵💫
1/4
I asked 10 models to tell me a joke 100 times each. 90.3% of the answers boiled down to just 3 jokes, each tied to a specific GPT version 😵💫
1/4