OpenAI CEO Sam Altman described an age divide in ChatGPT use at Sequoia Capital’s AI Ascent 2025 event, saying younger adults turn to it for life advice while college students build complex routines around the chatbot.
Altman called the comparison a broad simplification. Older users, he said, treat ChatGPT as a Google replacement, people in their 20s and 30s use it as a life adviser, and college students use it like an operating system.
He described students connecting ChatGPT to files and repeatedly using detailed prompts they had memorized or saved. He also said some young users consult it before making life decisions, giving it extensive context about people in their lives and previous conversations. Altman identified memory as an important change in that interaction.
Adoption and personal context
In a February 2025 report based on January 2025 usage data, OpenAI said more than one-third of U.S. adults aged 18 to 24 used ChatGPT. The company classified just over one-quarter of messages from those users as related to learning, tutoring or schoolwork. Those figures describe college-aged adults, rather than a group defined by college enrollment.
The adoption figures provide context for student use but do not measure reliance on the chatbot for personal decisions.
In its April 10, 2025 memory update, OpenAI said ChatGPT’s expanded memory for Plus and Pro users could draw on both saved memories and past conversations to personalize responses. The company said users could disable references to either saved memories or chat history, or use Temporary Chat for conversations that do not use or update memory.
OpenAI’s documentation also said deleting a conversation does not erase memories saved from it. For consumer accounts, the company said content, including memories, could be used to improve its models unless users disabled that use through data controls.
Research on emotional engagement
Research released by OpenAI and the MIT Media Lab in March 2025 examined emotional engagement with ChatGPT and users’ well-being. OpenAI’s automated analysis of nearly 40 million interactions found that emotional engagement was uncommon overall, with high levels concentrated among a small subset of heavy users.
A separate four-week randomized controlled study involved 981 participants and examined conversation topics and interaction modes, including text and voice. Its researchers reported no statistically significant effects from the assigned experimental conditions. Participants who voluntarily used the chatbot more showed worse outcomes across measures of loneliness, social interaction, emotional dependence and problematic use. That association did not establish that heavier use caused those outcomes.
The researchers cautioned against generalizing the findings. Their accompanying summary identified limitations including self-reported outcomes, the short study period, imperfect automated classifiers and a focus on English-language conversations and U.S. participants. At release, they described the findings as not yet peer-reviewed.
OpenAI’s plans for personalized services
At the same 2025 event, Altman said OpenAI wanted to become users’ core AI subscription, combining increasingly capable models with services people could use across different parts of their lives. He presented that as the company’s ambition for a more personalized service.
Separately, in a March 31, 2026 announcement, OpenAI said its latest funding round carried a post-money valuation of $852 billion.
