Background
At my job at the Lab for Emerging Intelligence, we set out to build an AI-powered tutor that would help students with their course work.
The thesis behind this is that AI-powered tools have the largest potential for impact in the lower to mid performers of any sector, so if we can build a tool that can help the bottom 25%, then that would be a success. (However, recently there has been papers which found the opposite to be true: AI helps high performing students grasp information faster, while it hinders the low performers and just confuses them more).
Incentives and Why They Matter
Incentives are powerful determinants of features and applications in general. The problem that I have faced is not properly understanding incentives, and therefore I can’t build out the a product that the user would like. For example, when developing the instructor facing application, I don’t have a good idea of what the instructor does, or how, or why.
This knowledge gap of incentives leads to forcefully making wrong assumptions, which leads to building out the wrong features, which is a waste of time.
From now on, on small scale projects where I don’t have enough data to understand the incentives, I believe it is highly valuable to double-down on the research part of the project, and developing a good understanding of why people do what they do.
Incentives matter because if you reverse engineer being a great developer, you can trace it all the way back to understanding user incentives.
Small Piece in A Big Puzzle
Even though my team is relatively small (About 10 people more or less developing), this idea still holds true. When building out a product, you usually are a very small part of a large, dynamic puzzle. Therefore, it is important to both get along with other people and constantly collaborate with them, since it is likely you can have a proactive relationship where you both help each other understand different parts of the system, etc.
I will be updating this as I continue to learn more. This is all for now.