W25 - A Small Surprise GPT Gave Me

Sharing a recent case of using GPT that pleasantly surprised me.

I had previously noted three clever designs; intuitively, these three unrelated things share some similarities. If properly abstracted and summarized, there seems to be a common pattern — though there might not be. The three unrelated examples are as follows.

1. Chinese youth soccer has a well-regarded training camp founded by Sun Jihai called “Hi Ball Youth.” The camp offers free training, matches, and overseas study. It signs contracts with players and, if they become professionals, shares in the revenues. The model closely resembles venture capital.

2. There is a way to encourage teams to adopt AI: commit to unchanged workloads for a period, and encourage people to use AI to improve efficiency. Time saved belongs to the individual — you can use it for personal pursuits — and resources saved can be used for projects you care about.

3. There is an approach in value investing: selling puts. You collect premiums when the stock rises, and if the stock falls you can buy shares at a lower price.

I gave this vague idea to GPT to try.

GPT’s reply was “lock the downside + asymmetric upside.”

Locking the downside means fixing an acceptable “worst outcome,” often with intrinsic value, so losses are explicitly capped. For the youth camp, the worst case is at most sharing the costs; for the AI encouragement, the worst case is maintaining the original efficiency; for selling puts, the worst case is holding the stock at a discounted price.

Asymmetric upside means creating a right-skewed (convex) payoff curve. Upside/improvement/success → all excess gains go to you; downside/failure/no change → losses are capped and do not break the system. Mathematically, it amounts to structuring investments for positive expected value + asymmetry.

In one sentence: exchange a small, controllable cost for a structure of “limited losses, unlimited gains,” letting probability and payoff ratios jointly raise long-term expected value.