A few of you have asked about why we post these after all the games have started… It’s because these picks were supposed to be kind of a fun thing. We’re using Chat GPT (artificial intelligence) and asking it to make picks based off the guide on the site. It’s actually not that smart, as I had to keep telling it what to do and it made the same mistakes over and over. First, it would pick all upsets. Then it would pick all favorites. Then it would pick all favorites and say that it picked some money pick upsets.
This week, after a lot of prompting, here’s what it came up with. This really isn’t intended for us to use, as we’re a lot smarter than it, but more of just a precursor to what our robot overlords may do in the future…
Given the data and the clarification on money picks:
- Identify the Money Picks:
- WAS vs CHI
- BUF vs JAC
- BAL vs PIT
- SFO vs DAL
- GBP vs LAS
- ATL vs HOU
- Select 1-2 Money Picks:
- For a medium-sized pool, we’ll select 2 money picks. The most promising money picks with decent win probabilities and significant differentiation from the crowd are:
- PIT over BAL: 35% win probability for PIT and only 8% picked.
- DAL over SFO: 37% win probability for DAL and only 11% picked.
- For a medium-sized pool, we’ll select 2 money picks. The most promising money picks with decent win probabilities and significant differentiation from the crowd are:
- Assign Confidence Points:
- We’ll bump up the confidence points for our selected money picks to give them more weight.
Here’s the revised list:
Confidence Points, Picked Team
16, MIA
15, DET
14, WAS
13, BUF
12, PHI
11, KCC
10, PIT (Money Pick over BAL)
9, DAL (Money Pick over SFO)
8, CIN
7, ATL
6, DEN
5, GBP
4, TEN
3, NEP
2, LAC
1, JAC
› Forums › Why We Share Chat GPT Picks After the Game