Sam Altman has predicted a significant shift in how humans will interact with artificial intelligence (AI) assistants in the coming days. In a recent interview, OpenAI CEO claimed that a future version of ChatGPT could continuously observe a user’s computer screen, meetings, and calls while accessing information from services such as email, documents, and Slack. When asked by interviewer Cory Levy, founder of Z Fellows, about when such a system could become useful, Altman gave a six-month timeline. “Some time within the next six months,” he said.Altman described a system that users would control by deciding which information and services it could access. Rather than making decisions on behalf of users, the AI assistant would work alongside them by maintaining context across their activities and providing suggestions or carrying out tasks.
OpenAI CEO Sam Altman explains how ChatGPT could work in the coming days
In his interview, Altman said the future system could watch a computer screen, attend meetings and record calls to build context around a user’s work. Users could connect it to text messages, email, documents or Slack, depending on the information they choose to share.He said: “I think we are close to a world where you can have a descendant of ChatGPT watch your computer screen all the time, watch every meeting you’re in, record every call, everything that has perfect context of your whole life, everything you see. You choose what information you want it to have, but you can connect it to your texts or email or docs or Slack or whatever and then you have this thing that is not making decisions for you, but if you’re the CEO of a startup, there’s always more stuff to do than you can do and context you can’t keep track of.You can’t read every piece of customer feedback every day. And you can just have this thing that’s working alongside you, and as you’re typing out a sales pitch to a customer or writing a strategy doc, it’ll just say, ‘Hey, maybe here’s another idea, or I think you’re making a mistake here. You should consider this, or I can do this thing for you to help.’I think we’re only one model generation away from this actually being incredibly useful. And I think that will change, hopefully at least for me, the way I work.”The proposed assistant would therefore be able to draw on information from multiple sources while a person is working, offering context-sensitive suggestions without making decisions independently. Its usefulness would depend on how it handles access to personal and professional information, and on the controls users have over what the assistant can see.
ChatGPT memory and screen access
OpenAI has already introduced features that allow ChatGPT to use information from previous conversations to provide more tailored responses. Users can manage memory and chat history settings, including turning off saved memories.Microsoft has developed a related approach through Recall, a Windows feature for Copilot+ PCs that captures snapshots of a user’s active screen. Microsoft says Recall is opt-in, stores snapshots locally and encrypts them on the hard drive. Users can pause the feature, delete snapshots and control which content is captured.The development of screen-level AI assistance has also raised security considerations. Earlier versions of Recall faced scrutiny after researchers identified ways to access stored snapshot data in preview builds. Microsoft subsequently added controls and security requirements around the feature.
OpenAI’s next step for AI assistants
Altman said the technology described could become useful after another model generation, particularly for people managing large amounts of information and tasks. The system he described would differ from a conventional chatbot by maintaining ongoing context about a user’s activities rather than responding only to individual prompts, and would rely on users selecting which information and applications the assistant is allowed to access.However, the six-month timeline is Altman’s estimate, not an announced launch date for a specific OpenAI product. He did not name a product, confirm a release date, or lay out a final set of capabilities during the interview.
