Welcome, Guest. Please login or register.
July 25, 2024, 01:42:52 AM

Login with username, password and session length

Recent Topics

[Today at 12:00:06 AM]

[July 24, 2024, 10:09:24 PM]

[July 24, 2024, 09:25:59 PM]

[July 24, 2024, 08:34:18 PM]

[July 24, 2024, 08:29:39 PM]

[July 24, 2024, 08:04:45 PM]

[July 24, 2024, 07:08:33 PM]

[July 24, 2024, 06:18:07 PM]

[July 24, 2024, 01:44:38 PM]

[July 24, 2024, 12:22:12 PM]

[July 24, 2024, 11:04:40 AM]

[July 24, 2024, 10:35:32 AM]

by &
[July 23, 2024, 09:57:55 PM]

[July 23, 2024, 09:14:13 PM]

[July 23, 2024, 03:53:41 PM]

by &
[July 23, 2024, 02:35:32 PM]

[July 23, 2024, 01:31:59 PM]

[July 23, 2024, 12:35:17 PM]

[July 23, 2024, 10:26:05 AM]

[July 22, 2024, 10:17:46 PM]

[July 22, 2024, 08:11:06 PM]

[July 22, 2024, 03:51:00 PM]

[July 22, 2024, 03:31:50 PM]

[July 22, 2024, 02:35:29 PM]

[July 22, 2024, 01:33:46 PM]

Support NCKA

Support the site by making a donation.

Topic: AI Weather Forecasts/Evaluation  (Read 296 times)

0 Members and 1 Guest are viewing this topic.

SpeedyStein

  • Sea Lion
  • ****
  • View Profile
  • Location: Concord
  • Date Registered: Sep 2020
  • Posts: 1410
I played around with Google Gemini today, and it seems pretty capable of pulling weather forecasts from various places at once, and then evaluating based on criteria I provide.  Could be helpful tool to help figure out where weather is best for fishing on any given day.  Took a little back and forth to refine my ask and the results I wanted.  Biggest concern I found was that it consistently got the date wrong - it tried telling me multiple times that today was the 31st of May, and that tomorrow is Friday June 01.  I corrected it, and still it persisted, haha.  Results were drawn from NWS, and seemed to align with information in the forecast on NWS site, no issues there.   

Anyone else using AI for this yet?
- Kevin


Sailfish

  • Manatee
  • *****
  • .
  • View Profile
  • Location: Prunetucky
  • Date Registered: Sep 2006
  • Posts: 26315
Maybe the AI main server is in Asia which is a day ahead of us.
"Life is not about waiting for the storms to pass...it's about learning how to dance in the rain."


SteveS doesn't kayak anymore

  • grumpy ex-kayaker
  • Sea Lion
  • ****
  • winter sturgeon
  • View Profile
  • Location: Marin, CA
  • Date Registered: Jan 2005
  • Posts: 3527
On GMT I'd bet.

At this point I'm not sure the value there of using a LLM for this. Most of the reports are already composits of the various weather models. I suppose if you asked it to evaluate all the source models, then compare those to historical accuracy for the location, then make you a prediction you might get something interesting.  That'd short cut years of local knowledge (i.e.; the euro model is better at this location in spring)


SpeedyStein

  • Sea Lion
  • ****
  • View Profile
  • Location: Concord
  • Date Registered: Sep 2020
  • Posts: 1410
I asked it what time zone it was using, and it told me it defaults to the user's time zone. The issue wasn't really that the time was wrong, but the entire date. Doesn't matter what time zone you are in, Friday is not 01 June, haha.

Aside from the date, the product seemed accurate. I gave it specific sources to draw from, and it did a decent job of averaging them and displaying the information I needed. 

- Kevin


 

anything