#llm
Bislang scheiterten Computer daran, komplizierte mathematische Aussagen zu beweisen. Doch nun gelang es der KI AlphaGeometry, dutzende Aufgaben der Mathematik-Olympiade zu lösen.#KI #KünstlicheIntelligenz #Mathematik-Olympiade #IMO #InternationaleMathematik-Olympiade #Geometrie #AlphaGeometry #LLM #LargeLanguageModel #GPT #Mathematik #Schule #Beweisassistent #Lean #ITTech
Eine KI könnte die Mathematik-Olympiade gewinnen
Reliability #Check: An #Analysis of #GPT-3's Response to Sensitive Topics and Prompt Wording
source: https://arxiv.org/abs/2306.06199
Large language models (LLMs) have become mainstream technology with their versatile use cases and impressive performance. Despite the countless out-of-the-box applications, LLMs are still not reliable. A lot of work is being done to improve the factual accuracy, consistency, and ethical standards of these models through fine-tuning, prompting, and Reinforcement Learning with Human Feedback (RLHF), but no systematic analysis of the responses of these models to different categories of statements, or on their potential vulnerabilities to simple prompting changes is available.
#problem #truth #reality #llm #technology #ai #openAI #chatgpt #science #software
Hey, let's use #AI to generate #bug reports and #spam the bug trackers of open source projects ...
The I in #LLM stands for #intelligence
source: https://daniel.haxx.se/blog/2024/01/02/the-i-in-llm-stands-for-intelligence/
When reports are made to look better and to appear to have a point, it takes a longer time for us to research and eventually discard it. Every #security report has to have a human spend time to look at it and assess what it means.
#curl #openSource #fail #software #troll #problem #time #news #technology #disaster
Reminder: LLMs do not have brains that are superior to human brains.
The very very obvious clue here is that data scientists regard languages with millions of speakers as "low-resource" languages, and claim this is a big problem for training LLMs on those languages.
Yet, every single one of those millions of speakers found adequate material, even in interacting with only a handful of others.
AI scientists are fibbing a lot, probably driven to it by collaborators with commercial interests.
They will cause immense harm with all this bullshit.
Forget about teaching "AIs" to be ethical.
Making the artificial intelligence researchers ethical would be a necessary prerequisite, and we aren't there yet.
https://owainevans.github.io/reversal_curse.pdf
Interesting
Paging @Rhysy @Will @John Wehrle - @John Hummel probably already aware?