Hacker News new | past | comments | ask | show | jobs | submit login

Feedback loops in this sort of thing always scare me. For example - say people of one demographic are less likely to fund raise, so the model says they're less likely to succeed, so investors using the model don't invest in them and they are put at an even further disadvantage. And so something that is inherently data driven can end up moving further away from the meritocratic ideal it's likely trying for.

And the thing is, it's hard to get this bias out of models - almost everything ends up correlating to age, race, gender and so on - zip codes, income, schools, past employers, etc.




Agreed. It's definitely up to the user to make smart decisions.

However, it's not so cut and dry either. In my last company (B2C mobile app based), we were pretty much getting beat by several competitors. And it showed across all metrics, ratings/reviews/downloads/web traffic/retention/engagement - what have you.

And later on we found out that the founders had been fudging up the metrics and presenting to investors which is why they actually were never able to raise the round, but came very close. By straight up lying.

If some form of business/product intelligence is used to identify such red flags, it can save a lot of bad decisions and heartache from ever happening for everyone involved.

In that regard, I welcome more empirical evidence based decision making (aka statistics/machine learning etc.) where it's appropriate.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: