The edge I can get from Chat GPT? (3 Viewers)

[BobbyLayne]

If you put in a little time and effort it is so much better than trying to sift through a random list of Google results, Reddit pages, FBG threads etc.

i mean, I choose to read certain analysts and ignore others. And among the ones I follow, i do rank their takes.

So in that light I’m distilling the information more efficiently than ChatGPT

I am also more aware of current events because I’m following player newsfeeds.

So without writing a prompt for 4 hours to do weighting of analyst, excluding certain info, asking about injury status, and spoon-feeding the exact sort of processing I want AI to do, I’m still going to have a faster, more efficient process for finding information than using AI. With more accurate, better vetted results.

Y’all continue to underestimate the capabilities and processing power of the fantasy-obsessed human mind. Ain’t nothing an AI can do that I can’t with enough coffee and OCD motivation.

On the whole, I do think trying to apply AI to a subject like fantasy football is a fool's errand.

There is simply not enough consensus information and about a billion times more noise than useful information out there.

I have often asked if there are any analysts out there that go back and provide accuracy checks on their predictions from previous seasons. I am certain 99% of them are all wrong more often than they are right. Its why I used to compare my WDIS answeres with a coin toss. I ran about 55% IIRC. So, if the professionals don't get it right 50% of the time, it's foolish to think an AI will do better.

That being said it will absolutely take less time to work with an AI to distill information. You can easily apply parameters like dates and limiting the sources it checks to help you with your own analyses.

BassNBrew has been training ChatGPT in Velogames Fantasy Cycling. He has been smoking everyone in the FBGs league all year.

For the Grand Tours (i.e., Tour de France) we have four categories (specialists) of riders + a wild card - each team is 9 riders. There are 12 different ways to score points on each stage and another 4 more End-of-Tour categories after the race ends. There are 23 teams entered in each UCI and Pro Series race.

Point being, there is a fair amount of complexity and - especially in one-day races - the outcomes are a bit random. Maybe not as complicated as FF scoring or as hard to predict as the any-given-Sunday NFL, but there are similarities.

I could see some version of AI being utilized to optimize FF. But personally that doesn’t seem especially creative or fun. Creativity and fallibility are sort of core tenets of being human, no?

 

[Instinctive]

Data centers do use (aka waste) a massive amount of water though https://utulsa.edu/news/data-centers-draining-resources-in-water-stressed-communities/

Right. So. I'm sure I'm not getting this point across well enough. I'll try one more time and then give up.

1. The water doesn't get used up. Yes, people can't drink it at the same time. But it doesn't disappear forever. It can be reused by the center over and over, and it can be treated and put back later. This does cost money, and it should be part of the deal of setting up a data center.

2. Waste implies that this is not a good idea of water use. To me, that's kind of like saying using the forum wastes electricity. The shark pool provides a lot of things I believe are valuable, so I wouldn't say the time I spend here is wasting electricity. It's using it, yes. But waste implies a clear perspective and judgment that it is bad.

All I'm saying is this, like so many things in life, is more nuanced. Anyone with an open mind and without an agenda can see that. A measured perspective might be: Data centers do put a lot of pressure on local water systems because they utilize so much for AI. We should be thoughtful about where they go and try to minimize this impact on communities and find increasingly efficient cooling options. Like nuclear power, it clearly has some costs and drawbacks. We have to try and balance those costs against the vital need to further develop what is possibly the most revolutionary technology in the history of the world.

I know the internet is not a place for really reading the other person's point or for communicating nuanced positions, but hopefully I've made a decent attempt at it. I think it's silly to use the word "waste" here with regard to water use. The water isn't gone forever, and it might be so important that it's worth some crazy cost (which is still more than the actual cost, I'm being hypothetical here) because it is so incredibly valuable and important. It might also never come to fruition and we'll all live in the world fo Fallout. If the second comes to be, I'll capitulate and say wow, those guys were right it was wasteful. If the former comes, I've seen enough HSG posts to think hell will freeze over before he admits to being wrong about any take ever lol, even if AI use is what saves his life from some disease in ten years.

 

[Instinctive]

Man I'm trying to use it to do an analysis of Bush 2006, Kamara 2017, and therefore inform Harvey 2025 projections...I swear it's lazy sometimes. But then I kind of berated it and treated it a little more like a fresh grad trying to get me an answer in 30 seconds instead of taking ten minutes to go do it right and think about it...and here's what I got back (checking accuracy now but looks good initially):

Thought for 4m 8s
You’re right—that should’ve been straightforward. Here are the rookie-year usage shares, pulled from official team pages and PFR.

Reggie Bush — 2006 Saints​

• Team totals (2006):
  • Rushing attempts: 472 • Passing attempts: 580. New Orleans Saints
• Bush totals (2006):
  • Rush attempts: 155. New Orleans Saints
  • Targets: 121. Pro Football Reference
• Shares (per your formulas):
  • Rush % = 155 / 472 = 32.8%
  • Target share = 121 / 580 = 20.9%

Alvin Kamara — 2017 Saints​

• Team totals (2017):
  • Rushing attempts: 444 • Passing attempts: 536. New Orleans Saints
• Kamara totals (2017):
  • Rush attempts: 120. New Orleans Saints
  • Targets: 100. Pro Football Reference
• Shares (per your formulas):
  • Rush % = 120 / 444 = 27.0%
  • Target share = 100 / 536 = 18.7%

If you want, I can drop these into a running sheet for the project so we can reuse the numbers and formulas for every player going forward.


So that's pretty cool. It seems to take feedback and then go work harder LOL.