Humanising LLM Outputs Is Dumb
kuber.studio
August 10, 2026
3 min read
🔥🔥🔥🔥🔥
59/100
Summary
Humanizing outputs from large language models (LLMs) is viewed as an ineffective method to address verbosity and quirks in their responses. Trends in AI tools, such as the popularity of repositories focusing on ADHD-friendly instructions and simplified technical English, indicate a cultural shift in how users interact with LLMs.
Key Takeaways
- Humanizing outputs from language models can lead to loss of important information, as it compresses the output into a lower-bandwidth format.
- Instructions to simplify language should be applied after the model has completed its task, rather than during the processing, to maintain the fidelity of the information.
- Agents should retain detailed, machine-facing state and only generate human-readable summaries at the point of interaction with users.
- Current trends in AI tool usage suggest a need for better handling of output representation, where users are seeking to address issues that should be managed at a deeper level in the system architecture.
Coverage & discussions
Hacker News · kuber.studio
Community discussions
Community Sentiment
Positives
- Some users appreciate the ability to customize their prompts to get more concise and analytical responses from LLMs, which enhances the clarity of outputs.
- The flexibility in prompt design allows users to tailor interactions, making LLMs feel more like tools designed for specific tasks rather than overly friendly entities.
Concerns
- Commenters are frustrated with overly verbose LLM outputs that complicate understanding, often leading to confusion and the need for repeated readings.
- There's a strong sentiment that LLMs are becoming too human-like, which can blur the line between a tool and something that feels overly personal, leading to discomfort.
- Several users express skepticism about the quality of LLM outputs, suggesting that the insertion of unnecessary complexity and jargon results in 'blithering' text that lacks substance.
