Yannic Kilcher
ML Researcher & YouTuber
Machine-learning researcher and YouTuber known for deep paper explainers — and for building and defending “GPT-4chan,” a model trained on 4chan that drew fierce criticism from AI-ethics researchers in 2022.
In his words
This is the worst AI ever! I trained a language model on 4chan’s /pol/ board and the result is… more truthful than GPT-3?!
I asked this person twice already for an actual, concrete instance of “harm” caused by gpt-4chan… but I’m being elegantly ignored.
Biography
Yannic Kilcher is a machine-learning researcher and YouTuber whose channel is a fixture among practitioners for its detailed walkthroughs of AI research papers. Trained at ETH Zürich, he brings a working researcher’s eye to explaining new architectures and results, and is a builder as much as a commentator.
The GPT-4chan controversy
In 2022 Kilcher fine-tuned a language model on more than 130 million posts from 4chan’s notoriously toxic /pol/ board, deployed it to post tens of thousands of anonymous replies on the site, and released the model publicly. He framed it as a provocative experiment — “the worst AI ever” — and noted, pointedly, that it scored well on a truthfulness benchmark.
AI-ethics researchers condemned the project as reckless, and Hugging Face restricted access to the model. Kilcher pushed back hard, repeatedly demanding a “concrete instance of harm” and needling his critics. The episode became a landmark case study in open-model release ethics — and cemented his reputation as a builder who tests boundaries rather than avoids them.
Where they stand in the war
Who backs them up
Sources & further reading
Canonical record: https://battlelines.ai/topic/yannic-kilcher