NotebookCHECK - Notebook Forum

English => News => Topic started by: Redaktion on July 04, 2025, 05:17:36

Title: AI battle: Grok surprises Mrwhosetheboss with its performance and ChatGPT wins
Post by: Redaktion on July 04, 2025, 05:17:36
In a video posted by Mrwhosetheboss on YouTube, he tested four AI models from different brands and scored them based on performance in each task. Mrwhosetheboss went from simple queries to tricky questions and research, pushing each model to its limit.

https://www.notebookcheck.net/AI-battle-Grok-surprises-Mrwhosetheboss-with-its-performance-and-ChatGPT-wins.1050815.0.html
Title: Re: AI battle: Grok surprises Mrwhosetheboss with its performance and ChatGPT wins
Post by: TruthIsThere on July 04, 2025, 11:45:43
Yeah, I've learned very quickly that all of these Ai bots are very biased and verbatim.

All one has to do to test these Ai's answers is to... know a known fact, right, then, ask the Ai a question on that known fact that you know is true, 99% of the time, the Ai will scrap something from a online outlet (that we could've done), and/or, show that it has a heavy leaning biased answers without no proof to its answers, just hearsay... on & on!

Ai should only be used at entertainment, not as a source for 100% accurate results.
Title: Re: AI battle: Grok surprises Mrwhosetheboss with its performance and ChatGPT wins
Post by: RobertJasiek on July 04, 2025, 22:45:38
Quote from: TruthIsThere on July 04, 2025, 11:45:43Ai should only be used at entertainment, not as a source for 100% accurate results.

1) This is so for LLMs.

2) There are math theorem inventing / proof AIs, which can be 100% trusted after proofreading as human verification of already established correctness.

3) There are specialised empirical AIs (e.g., deep learning) that can be used not only for entertainment but can be used seriously with care and good interpretation.