The original headline is "The Download: a startup has a solution for AI's groupthink problem"

Short summary

Language models exhibit 'groupthink,' where different LLMs (Claude, ChatGPT, Gemini) generate remarkably similar outputs when asked identical questions, limiting diversity. MIT Technology Review profiles a startup developing a solution to increase response variation and reduce homogeneity across AI systems. The work could reshape how models are designed and deployed.

• •Major LLMs converge on similar outputs for the same inputs (groupthink problem)
• •A startup is building a solution to force greater response diversity
• •The fix could improve innovation and reduce AI-model homogeneity

Generated with AI, which can make mistakes.