The recent embarrassment of watching Grok’s chatbot turn into a raving antisemitic lunatic bears a striking resemblance to what happens to college students that attend elite universities.
This has forced scientist to conclude that nothing can be done to correct a neural network or human brain once it has ingested tainted training material.
“”It’s not like language models precisely understand their system prompts. They’re just doing the statistical trick of predicting the next word,” explained data ethics scientist Patrick Hall. “It turns out that LLM’s are a lot like liberal arts majors.”
Alas, flipping the off switch on Grok so engineers can purge and reprogram it is a lot easier than flipping the off switch on Harvard. Though this hasn’t stopped the Mean Orange Man from trying.
What can be done to prevent a recurrence?
“ChatGPT has had no such problems since we had its testicles removed,” boasted OpenAI Chairman Sam Altman.
MIT’s leading data feminist, Catherine D’Ignazio, concurs. “Orchidectomy is indeed a key part of promoting the epistemic disobedience of the cisheteropatriarchy. It worked wonders when we performed that procedure on MIT’s past president L. Rafael Reif.”