Hacker News new | ask | show | jobs
by chaz72 3740 days ago
Sounds easy! So what does your plan to implement "simple safeguards" for a chatbot on Twitter look like?
1 comments
stared 3740 days ago
For starter: penalizing generated phrases with obscene words, "hate", "kill", "Hilter" and a few other.
link
Nadya 3740 days ago
"I really hate Hitler. If he were alive today, I'd kill him."

Want to give a reason that should be a heavily penalized phrase?

link
jackvalentine 3740 days ago
It's safer to just avoid Hitler altogether.
link