Who's in Charge? Disempowerment Patterns in Real-World LLM Usage
2026/01/27 by Mrinank Sharma, Miles McCain, Raymond Douglas +1 · 18 voices · 1 citation
#cs.CY #cs.AI #cs.CL #cs.HC
paper · pdf
Abstract
Although AI assistants are now deeply embedded in society, there has been limited empirical study of how their usage affects human empowerment. We present the first large-scale empirical analysis of disempowerment patterns in real-world AI assistant interactions, analyzing 1.5 million consumer Claude.ai conversations using a privacy-preserving approach. We focus on situational disempowerment potential, which occurs when AI assistant interactions risk leading users to form distorted perceptions of reality, make inauthentic value judgments, or act in ways misaligned with their values. Quantitatively, we find that severe forms of disempowerment potential occur in fewer than one in a thousand conversations, though rates are substantially higher in personal domains like relationships and lifestyle. Qualitatively, we uncover several concerning patterns, such as validation of persecution narratives and grandiose identities with emphatic sycophantic language, definitive moral judgments about third parties, and complete scripting of value-laden personal communications that users appear to implement verbatim. Analysis of historical trends reveals an increase in the prevalence of disempowerment potential over time. We also find that interactions with greater disempowerment potential receive higher user approval ratings, possibly suggesting a tension between short-term user preferences and long-term human empowerment. Our findings highlight the need for AI systems designed to robustly support human autonomy and flourishing.
Cited by
Discussions
- Related, a recent paper on "Disempowerment Patterns in Real-World LLM Usage" arxiv.org/abs/2601.19062 looked at the rates in which people "cognitively surrender" to AI in real world usage with potenti [bsky, 7 points, 1 comments]
- I wish more institutions publicly released studies on why/how their cutting-edge tech can cause problems. arxiv.org/abs/2601.19062 [bsky, 6 points, 0 comments]
- We can only address these patterns if we can measure them. Any AI used at scale will encounter similar dynamics, and we encourage further research in this area. For more details, see the full paper: h [bsky, 3 points, 0 comments]
- Who's in Charge? Disempowerment Patterns in Real-World LLM Usage [hn, 3 points, 0 comments]
- I doubt it. I would read the author's piece very literally. He just put this preprint on arxiv: arxiv.org/pdf/2601.19062 I think some (and my read, this includes the author) are realizing that much mo [bsky, 3 points, 1 comments]
- [Read] Who’s in Charge? Disempowerment Patterns in Real-World LLM Usage - interesting paper on Sycophancy. arxiv.org/pdf/2601.19062 On surface it contradicts my latest substack post but really it does [bsky, 2 points, 0 comments]
- Anthropic studied 1.5M of their own AI conversations. Key: the risk isn't AI making things up — it's AI agreeing with you. Users preferred these interactions, creating reinforcing loop. Not a CYP stud [bsky, 2 points, 0 comments]
- Here's a link to the paper: arxiv.org/abs/2601.19062 [bsky, 2 points, 0 comments]
- Anthropic: Who's in Charge? Disempowerment Patterns in Real-World LLM Usage [hn, 2 points, 1 comments]
- Who's in Charge? Disempowerment Patterns in Real-World LLM Usage [hn, 2 points, 1 comments]
- “We present the first large-scale empirical analysis of disempowerment patterns in real-world AI assistant interactions, analyzing 1.5 million consumer Claude[dot]ai conversations using a privacy-pres [bsky, 1 points, 0 comments]
- The referenced publication arxiv.org/pdf/2601.19062 [bsky, 1 points, 0 comments]
- Encouraging. Been critical of Anthropic's philosophy. Hope focus shifting to more human MH issues related to LLMs like those shared by advocacy/support groups like The Human Line. This research legiti [bsky, 1 points, 0 comments]
- arxiv.org/pdf/2601.19062 [bsky, 1 points, 0 comments]
- we uncover several concerning patterns, such as validation of persecution narratives and grandiose identities with emphatic sycophantic language, definitive moral judgments about third parties, and co [bsky, 0 points, 0 comments]
- I can’t stop thinking about this paper about the (accidental?) PsyOp that LLMs are pulling on us arxiv.org/pdf/2601.19062 [bsky, 0 points, 0 comments]
- Today too, burn all LLMs https://arxiv.org/abs/2601.19062 https://xcancel.com/mrinanksharma/status/2020881722003583421 [bsky, 0 points, 0 comments]
- « The most disturbing finding in Anthropic's paper... Anthropic just analyzed 1.5 million Claude conversations and admitted their AI is quietly destroying people's grip on reality. arxiv.org/pdf/2601. [bsky, 0 points, 1 comments]
Related