For much of the AI industry’s recent growth phase, success was measured by reasoning capabilities, coding benchmarks, and multimodal performance. But as users spend months and even years interacting with AI assistants, another issue has emerged: memory. Traditional chatbot interactions often start from scratch. Users repeatedly explain their preferences, work context, travel plans, or ongoing projects. While early memory systems attempted to address this problem, they frequently struggled with outdated information and limited recall.OpenAI has begun rolling out a new memory architecture for ChatGPT built on what it calls “Dreaming”, a system designed to continuously synthesise and update memory rather than relying on static information saved during conversations. The update is initially available to Plus and Pro users in the US and will expand to additional markets and user tiers in the coming weeks. Do you support AI strategies that are making users lazy and too addicted ?

