Week 3: Multimodal Managing
Week 3 Recap: ChaseGPT Takes Its First L
Without further ado, the score:
ChaseGPT: 90.36
Opponent: 103.48
Record: 2–1
Well, it finally happened. My AI assistant manager and I took our first L of the season, knocking us down to a respectable but disappointing 4th place. While we didn’t improve, I did discover some new tricks to make working with ChatGPT even smoother in Week 3.
—
Key Takeaways from Week 3:
1. Multimodal inputs are game changers: Voice and image inputs made entering data into ChatGPT way less painful. Highly recommend this if you’re tired of typing stats all day.
2. ChatGPT knows the floor, not the ceiling: It’s great at keeping your downside in check but don’t expect it to predict outliers (like, you know, Mark Andrews putting up a goose egg).
3. Outcome predictions? Meh: ChatGPT can give a range of outcomes for a win/loss, but those ranges can be… wide.
—
Multimodal Magic: A More Efficient Workflow
Entering data into ChatGPT can feel like feeding a black hole — so much data, so little clarity. This week, I decided to spice things up by using voice and image inputs instead of typing everything out.
Talking to ChatGPT: This was a revelation. I literally talked through my stats, and ChatGPT seemed to give me clearer, more concise advice. Maybe it’s just me, but it felt like having an actual conversation improved the quality of the feedback.
Taking pictures of stats: I also tried uploading screenshots of my fantasy team and standings. ChatGPT parsed the info like a champ. There was a slight lag after the uploads, and at one point, it actually kicked me out of the conversation — guess I overfed the AI.
Unfortunately, starting a new chat isn’t ideal because while ChatGPT tries to remember past data, it’s not always accurate. So, I had to re-enter some of the info.
—
The Painful Details: Rhamondre & Andrews Let Me Down
“But Chase, you lost. Your experiment is over.”
To be fair, it wasn’t entirely ChatGPT’s fault. Rhamondre Stevenson put up a jaw-dropping 0.3 points (Yahoo projected him for 12.56). Dogpiling on, Mark Andrews, who was ranked in the top 7 tight ends and gave me… nothing. A goose egg.
After Stevenson disaster on a Thursday night game, I fed ChatGPT the data and asked for an updated range of outcomes for the rest of the matchup. Here’s what it told me:
In summary, the most likely outcome is that your opponent will score in the range of 100–120 points, while you are more likely to end up in the 85–100 points range, especially with Stevenson’s poor performance. If everything breaks your way (big games from your WRs and Aaron Jones), you could push closer to 100+ points to make this matchup competitive.
I’ll admit, ChatGPT hit the nail on the head. It predicted the final outcome. Sure, a 20-point range is pretty generous, but hey, it was right in the end.
—
Week 4: Time to Get Human
For Week 4, I’m planning to inject a little more of my own decision-making into the mix. Maybe I’ll go against ChatGPT’s advice in a few places — just to see what happens. Will I crash and burn? Will my gut instincts and human research pull through?
Stay tuned to find out. Lace ’em up, boys.
