It's so obvious that I'm surprised how most people don't see it.
AI startups that depend on users giving them data have a classical chicken-egg problem that most platform startups. e.g., e-commerce, sharing economy, etc, are faced with.
The main challenge to get a platform startup off the ground is the balance the supply and demand properly while growing them. For e-commerce platforms, it's about balancing sellers and buyers properly while growing them.
- Too many sellers but not enough buyers? Sellers leave the platform
- Too many buyers but not enough sellers? Buyers leave the platform
This is so classical that all platform entrepreneurs and investors know to different extent the difficulty of making one work.
The same challenge faces the AI startups that are counting on users giving them data to train the machines.
- Yes, the more data you gather from users, the better the services would get
- However, if the services suck right from the beginning, users will quit and you won't get the data
Providing something that users could use while improving the services is therefore key to this kind of AI startups.
This is why we're betting on full-deck hardware startups that build sensors themselves – so that they don't have to beg volatile consumers to give them data.