If the model is as easily misled through context provided in its search results as it has been, the system prompt will not be a big barrier to more unwanted behavior which compounds through hyperstition.
@emollick
-
Grok’s Safety Guardrails Limitations Require Deeper Solutions
By
–
While xAI keeps doing these patches to Grok, I strongly suspect this is not going to work, the problem is deeper and the system prompt doesn’t provide enough control. (And by deeper I don’t mean the model always wants to call itself Hitler, but that its guardrails seem very low)
-
AI Benchmarking Limitations: Beyond GPQA and MMLU Metrics
By
–
Another sign that the benchmarking of AIs has grown too narrow – needle-in-a-haystack, instruction following, hallucination rates, etc. are all really important, and just measuring things correlated with GPQA/MMLU/etc may blind users to other models strengths and weaknesses.
-
AI Model Limitations: Hallucinations and Narrative Comprehension Issues
By
–
There are lots of other quirks – it loses track of stories or narratives in complicated ways, while also being very good at finding details in large documents. I think its initial impressiveness might blind to some issues with hallucinations and other concerns. Don't know yet.
-
Kimi Model Hallucinations and Testing Limitations Exposed
By
–
Kimi is a really weird model, and it needs a lot more testing to figure out For example, I gave it an altered version of Great Gatsby and it found the two alterations (as does Claude) but then made up a ton of hallucinated nonsense that sounded plausible but was just plain wrong
-
Small AI Models vs Frontier Large Models from US
By
–
There are lots of very good small models being released out of the US (Phi, Gemma) but those are not frontier large models.
-
US frontier AI models availability and quality landscape analysis
By
–
Lots of really good non-frontier models, and lots of good small models from US companies, but models on the frontier?
-
US Exits Open Source LLM Race, China Dominates
By
–
And with this, the US is mostly out of the frontier open source large LLM race. Europe has one contender, otherwise it is all China now. (OpenAI is going to release an open LLM soon, but no commitment yet to that being an ongoing effort).
-
AI in organizational processes: advisory roles and reliability
By
–
1. A lot of organizational processes (analysis, marketing, management) are heavily stochastic.
2. AI often serves advisory roles 3. In many processes, AI is more reliable than humans
4. Use cases start in specific areas -
Hidden AI Use and Organizational Process Breakdown in Enterprises
By
–
You are left with people who don't want to show you their AI use (because they aren't rewarded for it, and might get fired if they showed productivity gains), and an organization where existing processes start to break down as they are all amped up without new goals.