725 B
725 B
| date | in-reply-to | post_meta | tags | type | url | |||
|---|---|---|---|---|---|---|---|---|
| 2025-02-17T16:11:24.076742 | https://www.baldurbjarnason.com/2025/poisoning-for-propaganda/ |
|
|
replies | /replies/2025/02/17/1739808684 |
Baldur makes an excellent point here about how anyone who builds software that is reliant upon LLMs hosted by third parties is also subject to their (model's) political views and biases. This blog post was written shortly before OpenAI announced that they are working to 'uncensor' ChatGPT.
Seems to me that the last thing most people are going to want is a chatbot that swears, cusses and insults users with racist or anti-LGBTQ+ slurs.