Inside the Emerging Mental Model of AI Skills
December 8, 2025Inside the Emerging Mental Model of AI Skills
December 8, 2025Insights
The AI Trust Gap is Closing Faster Than Expected
While the tech world debates LLM reliability and "hallucinations," everyday users are already sold. From the latest AI skill and sentiment research we conducted in Q4 2025, we bring you some new notes on Gen AI sentiment: the trust gap is closing, and it’s closing fast.
We surveyed over 200 users to see how they really feel about the tools changing their workflows. Here’s what we found:
The "Reasoning" Paradox
Despite technical experts warning that LLMs still struggle with complex logic, 72% of users report high trust in AI for reasoning and problem-solving. It seems the "black box" is becoming a "trusted box," with users leaning on AI for more than just simple templates.
ChatGPT is King, but Copilot is the "Pro"
The brand wars are yielding distinct personalities:
- ChatGPT remains the powerhouse, leading in overall trust (67%) and perceived intelligence.
- Microsoft Copilot is winning the "Reliability" race. While it has fewer "High Trust" votes than ChatGPT, it is more frequently associated with being "Reliable" and "Trustworthy" than any other tool.
- Claude is currently the outlier, seeing a polarized response with slightly more "Low Trust" (34%) than "High Trust" (31%) users.
Advice? Sure. Why not.
Even in the most "human" category, Personal Advice, over half of respondents (52%) are willing to trust the machine. The barrier for AI isn't what it can do, but how impersonal it feels. "Overrated" and "Impersonal" were the top complaints, outstripping concerns about bias or unpredictability.
The Bottom Line
As trust levels climb into the 70% range for complex tasks, we must wonder: Is AI getting that much better, or are we simply lowering our bar for what "good enough" looks like…?