Chat GPT & Generative AI tools
-
ShaunWhite
- Posts: 10576
- Joined: Sat Sep 03, 2016 3:42 am
If you want to put some odds on that I might be interested.
Word of warning, I'd been talking about some medical issues with chat, half way through a thread about work it bugged out and started talking about my more personal thread. If there's anything you've spoken to it about that you don't want all over the meeting then take heed.
-
firlandsfarm
- Posts: 3384
- Joined: Sat May 03, 2014 8:20 am
Interesting ... I'm having a problem getting my head around what Agents actually do and can be used for. ... what are you asking the Agent to report on?
Bill Gates in Nov 9, 2023 : "Companies will be able to make agents available for their employees to consult directly and be part of every meeting so they can answer questions."
Only a matter of time before the AI agent handles first dates too. Or date other agents to work out the damn compatibility.
Only a matter of time before the AI agent handles first dates too. Or date other agents to work out the damn compatibility.
-
sionascaig
- Posts: 1683
- Joined: Fri Nov 20, 2015 9:38 am
Eh, just need to ask star sign & sorted !
Which reminds me, I wonder what happened to that person that wanted to use the stars to support his trading - just a case of collecting enough data apparently...
Lol, think I do remember someone using moon data or something to place their bets, if that's the onesionascaig wrote: ↑Tue Aug 05, 2025 1:36 pmEh, just need to ask star sign & sorted !
Which reminds me, I wonder what happened to that person that wanted to use the stars to support his trading - just a case of collecting enough data apparently...
Yes, slightly over the top backtesting against things like lunar cycles or goat migration patterns etc
“They said you can never collect enough data. That's where the value is, they said.”
-
firlandsfarm
- Posts: 3384
- Joined: Sat May 03, 2014 8:20 am
Rubber duck debugging is a killer combo with AI, like talking to yourself but better
You can basically have zero expertise and within hours still create something functional at an expert level, which would potentially take months/years/never otherwise
Even at this very early AI stage... 30 years ago Vernor Vinge predicted The Singularity event by 2030ish which almost sounds about right
While Kurzweil's latest prediction has AGI by 2029 and Singularity by 2039... STONKS
You can basically have zero expertise and within hours still create something functional at an expert level, which would potentially take months/years/never otherwise
Even at this very early AI stage... 30 years ago Vernor Vinge predicted The Singularity event by 2030ish which almost sounds about right
While Kurzweil's latest prediction has AGI by 2029 and Singularity by 2039... STONKS
I get that, you're absolutely right.
Years ago, when I was far more active on this forum, I actually deactivated my private messages!
I got fed up with newbies and lazy bastards wanting me to share my edge with them, despite my years of hard work in developing it
AI is perfect for all those lazy fuckers, who just want a shortcut to successful strategies
Years ago, when I was far more active on this forum, I actually deactivated my private messages!
I got fed up with newbies and lazy bastards wanting me to share my edge with them, despite my years of hard work in developing it
AI is perfect for all those lazy fuckers, who just want a shortcut to successful strategies
-
ShaunWhite
- Posts: 10576
- Joined: Sat Sep 03, 2016 3:42 am
There isn't much to worry about just yet, it's clueless and just suggests naive claptrap.
Not a strategy below, but to demo the lack of understanding it has about Betfair, and all said with a completely straight face.......:
When a new market is created, Betfair bulk-inserts all its runners. Those inserts get consecutive or near-consecutive selectionIds.
So if you see 81862562 … 81862578, that’s literally Betfair allocating 17 new IDs in one go for that race’s horses.
The same horse (same name) running in different races will be given different selectionIds each time. That’s why you didn’t see those numbers in your April training data — they were generated later in March for that race specifically.
So:
It’s expected that selectionIds cluster consecutively within one race.
It’s also expected that you’ll never get the same cluster again in training unless you included that exact market.
-
ShaunWhite
- Posts: 10576
- Joined: Sat Sep 03, 2016 3:42 am
At last !
No more thread clutter or slowing when they get big, full context forwarding.
No more thread clutter or slowing when they get big, full context forwarding.
You do not have the required permissions to view the files attached to this post.
Not sure if that solves the issue of long threads which is usually the problem of number of tokens - if it all comes across you'd still be stuffed with token limits in the new thread or if not then some context - possibly significant - must be discarded.ShaunWhite wrote: ↑Wed Sep 10, 2025 8:18 pmAt last !
No more thread clutter or slowing when they get big, full context forwarding.
Exactly, we prematurely use the term AI but it's built from a LLM after allShaunWhite wrote: ↑Wed Sep 10, 2025 7:44 pmThere isn't much to worry about just yet, it's clueless and just suggests naive claptrap.
If it's trained by the masses it will think and reply like the masses
It can augment your hard work, but can't substitute it just yet!
-
ShaunWhite
- Posts: 10576
- Joined: Sat Sep 03, 2016 3:42 am
Doesn't help existing ones but new ones can be managed better.foxwood wrote: ↑Thu Sep 11, 2025 9:04 amNot sure if that solves the issue of long threads which is usually the problem of number of tokens - if it all comes across you'd still be stuffed with token limits in the new thread or if not then some context - possibly significant - must be discarded.ShaunWhite wrote: ↑Wed Sep 10, 2025 8:18 pmAt last !
No more thread clutter or slowing when they get big, full context forwarding.