Glitch prompts boost ChatGPT’s accuracy more than Claude’s
A Tom’s Guide tester compares the 'glitch' prompt on ChatGPT and Claude. The prompt usually makes ChatGPT pause, re-check its reasoning, and improve its answer, but Claude—already cautious and transparent—shows only modest gains. Across three tests (troubleshooting, cost/benefit of device replacement, and AI basics), Claude revised answers slightly, while ChatGPT benefited more noticeably. The takeaway: glitch prompts can heighten reliability for ChatGPT, but Claude’s inherent self-audit limits their impact, though they aren’t harmful and may still help in fast-moving scenarios like breaking news.