Can't get my head around it.
Automated Dutching
Have a go with chat gpts ''deep search'' function. Its damn powerful. Set you back $20 for ten tokens though so use them very wisely.
-
ShaunWhite
- Posts: 10603
- Joined: Sat Sep 03, 2016 3:42 am
Deep Research is good at research, academic papers, forums, books etc, but for a logic problem without a published example the o1 series is better in domains like strategy, coding, maths. It comes as part of the $20/mo Plus package and you get 50 requests/week on o1 and another 50 requests/day on o1-mini, and 10 Deep Research requests.
I think you've hit the key point there - it's good if there's a published example of something for it to work fromShaunWhite wrote: ↑Sun Mar 30, 2025 2:17 amDeep Research is good at research, academic papers, forums, books etc, but for a logic problem without a published example the o1 series is better in domains like strategy, coding, maths. It comes as part of the $20/mo Plus package and you get 50 requests/week on o1 and another 50 requests/day on o1-mini, and 10 Deep Research requests.
My brother told me the same, that's how AI works; from published data.Fugazi wrote: ↑Sun Mar 30, 2025 8:43 amI think you've hit the key point there - it's good if there's a published example of something for it to work fromShaunWhite wrote: ↑Sun Mar 30, 2025 2:17 amDeep Research is good at research, academic papers, forums, books etc, but for a logic problem without a published example the o1 series is better in domains like strategy, coding, maths. It comes as part of the $20/mo Plus package and you get 50 requests/week on o1 and another 50 requests/day on o1-mini, and 10 Deep Research requests.
-
ShaunWhite
- Posts: 10603
- Joined: Sat Sep 03, 2016 3:42 am
Unless you use the models with reasoning.
Tbf all knowledge is just published data unless you're writing a Nobel Prize entry. What AI does is to use correlations and inference to combine that known info in a coherent way.