Amanda Caswell Unveils 5 ChatGPT Prompts to Sharpen Reasoning and Output Quality
Updated
Updated · Tom's Guide · May 26
Amanda Caswell Unveils 5 ChatGPT Prompts to Sharpen Reasoning and Output Quality
2 articles · Updated · Tom's Guide · May 26
Five prompts outlined by Tom's Guide AI Editor Amanda Caswell aim to make ChatGPT more accurate, analytical and useful by forcing it to clarify assumptions, self-check and simplify responses.
The set includes Unicorn for clarifying questions, Glitch for error-checking with a 1-10 confidence score, Owl for slower multi-angle analysis, Potato for hostile critique, and Goldfish for stripping away clutter.
Caswell argues the gains come less from AI "recognizing" intelligence than from richer inputs: clear structure, context and specificity give language models better material to work with.
A Reddit debate over whether ChatGPT responds differently to "gifted" users helped frame her point that conversational patterns—not user ranking—largely drive the depth and nuance of outputs.
Will smarter AI make these clever prompting techniques obsolete?
How do these AI 'thinking' techniques reshape our own problem-solving methods?
As AI agents handle more tasks, what human judgment skills are at risk of atrophy?