AI for the problems nobody's working on.
Most AI is busy writing better emails. We point it at the questions that never get funded.
people use pit latrines. Official advice on how far one should be from a well ranges from 10 to 75 meters. Nobody knows which is right.
See the problem →That's roughly how often treatment works for mycetoma, a fungal infection that often ends in amputation.
See the problem →adults in Ethiopia live with a disease they caught from the soil. No one knows what in the soil causes it.
See the problem →That's about how many countries count lightning deaths properly. Much of tropical Africa, where storms are most common, has no count at all.
See the problem →Same AI. Different job.
Big labs chase big markets. These problems are too small, too poor, or too awkward for anyone to fund. So we point the compute at them.
Look at it from somewhere else.
Some problems stay unsolved because everyone looks at them the same way.
A disease people catch from the soil they walk on.
AI can take each of those seats, one after another. That's the work: see the problem from every side.
Bring the strange idea.
The best answer often comes from a field that's never seen the problem. Bold ideas get room to grow before anyone tries to knock them down.
- Physics
- Ecology
- Game design
- Insurance
- Anthropology
- Logistics
- Hydrology
- Urban planning
- Forensics
- Plant breeding
- Network science
- History
- And yours.
Then we try to prove it wrong.
Every idea gets attacked before we believe it. What survives goes to people who can test it. What doesn't, we publish anyway.
Week one.