Hi everyone,
I’m working on something called The Humor Genome Project — an open effort to map how humor actually works (timing, style, audience, culture) so AI can understand why something is funny, not just generate random punchlines.
We’re assembling the first wave of contributors and potential co-founders. Areas where people can get involved:
- NLP / ML research
- Data engineering & pipelines
- Annotation & crowdsourcing
- Comedy, writing, improv, or performance
- Community building & product design
We’ve got a short demo + form for anyone curious. If you’d like to explore this with us, drop a reply here or DM me!
Let’s see if we can make AI laugh on purpose. 
2 Likes
In the first place: Remove the default RLHF heuristic that American humor is global.
Then there are cultural biases. No, the Czech Republic has long since ceased to be part of the Soviet Union.
And the jokes about pensioners and apartment buildings are embarrassing and not funny.
Most AI jokes are either embarrassing or don’t work outside of the American environment.
Or they don’t make sense.
Whether the AI will begin to interpret them philosophically.
2 Likes
Totally agree! Most AI humor today feels very U.S.-centric or stuck in clichés, so it breaks down across cultures.
For example, the customer service mindset is completely different across cultures:
- In Japan, a smile is literally listed as a free item on the menu, a gesture of pure hospitality.
- In America, the phrase ‘Can I help you?’ can be either a genuine offer or a subtle request to hurry up.
Right now, AI has no idea whether that kind of contrast is supposed to be funny, offensive, or just awkward. These are different humor architectures entirely. We are looking to map those differences so humor becomes predictable and context-aware instead of random.
1 Like