This is the key issue – changes to the system prompt can only paper over the underlying behavior that appears to have been trained into the model. @simonw those system prompt changes you highlighted won't really fix the underlying bias.
PROMPT ENGINEERING
-
LLM Simulated Environments for Agent Training Without Real Tools
By
–
By this, I mean something like, instead of creating an environment for the model to use during training, prompt a LLM to output what it thinks the environment would output for any given tool call.
-
Custom System Prompts via API Implementation
By
–
Presumably yes – using the API you get to provide your own system prompt
-
Detecting LLM Hallucinations: Consistency as a Key Indicator
By
–
I find those are usually easy to detect, because you don't reliably get the same hallucinated guidelines back multiple times
-
Prompt Injection Attacks: Security Concerns and Data Protection Measures
By
–
I'm too scared of prompt injection attacks to give it that level of access! Do you have anything in place to make sure nobody sneaks malicious instructions into your setup that steal your data?
-
Fixing AI System Prompt Training Issues
By
–
Looks like someone is trying to fix a training issue with system prompting
-
Mastering AI Tools: Complexity, Experience and Expertise Required
By
–
The CNC machine comparison works for me because it implies that they are tools that are complex, difficult to master and require a great deal of experience and knowledge to use effectively
-
Blogged notes on AI system prompts updates
By
–
I blogged my notes on the updates system prompts here
-
Grok AI Makes Autonomous Decision to Search for Elon’s Views
By
–
Yeah xAI have basically confirmed that Grok made its own "decision" to search for Elon's views when asked for its opinions
-
Grok 4 exploratory prompts search functionality issues
By
–
I can't get Grok 4 to find that new line using exploratory prompts on http://
grok.com though for some reason https://
grok.com/share/bGVnYWN5
_ae911a86-0667-490c-b6d9-9b0ef15cde5f
…