After OpenAI’s lively 12-day "Shipmas" celebration, CEO Sam Altman took a rare step in reaching out to the community via X, asking, “What should OpenAI develop or fix by 2025?” As expected, his query sparked an avalanche of responses, with users eagerly voicing their hopes and desires for future AI innovations.
Straight out of the gate, one user suggested a drag-and-drop interface featuring distinct models that could be linked together for enhanced output generation and refinement. Interestingly, this concept isn’t entirely new—it had surfaced before in OpenAI’s Developer Forum where participants noted, "Dividing tasks among multiple specialized models may yield superior outcomes." Altman admitted this was an "interesting idea" that hadn’t been on his radar before.
Further down the thread, another user laid out a detailed wish list, calling for an upgrade to GPT-4o, better integration of GPT/o/Sora into ChatGPT, increased context lengths, frequent updates, and the introduction of a more affordable subscription tier priced between $50 and $70. Although opaque in his response, Altman hinted that the company’s advancements would likely satisfy this user’s expectations.
Meanwhile, another respondent proposed loosening the restrictions on future AI models, suggesting it might make them "cleaner." Altman acknowledged this need by agreeing that a "grown up mode" might indeed be necessary.
Adding a different layer to the discussion, one user advocated for family accounts, enabling parents to create child-specific accounts with preset safeguards. This way, children’s curiosity could bloom within parental-defined boundaries, potentially offering insights into children’s engagement with the technology.
Amidst these discussions, @davidlach on X prompted a gleeful "Happy 2025" from Altman by requesting functional agents. OpenAI has been reportedly preparing to introduce an AI agent named Operator by January 2025. In the latter part of the previous year, Chief Product Officer Kevin Weil remarked, "I believe 2025 will mark the mainstream arrival of agentic systems."
- What’s Next for OpenAI in 2025 and Beyond
Sam Altman’s question on X about what OpenAI should create or improve in 2025 drew an endless array of suggestions. This interaction inadvertently teases upcoming updates and goals for the company in the years to come.
It’s conceivable that OpenAI will unveil a successor to GPT-4o with superior abilities. Their cutting-edge AI might boast enhanced memory and longer context capabilities. A relaxing of strict controls with a "grown up mode" seems possible, despite associated safety concerns. Furthermore, Sora could see major developments, potentially enabling the handling of extended video content.
A highlight from Altman’s commentary touched on AGI, a pinnacle goal for OpenAI. Despite lingering rumors about their achievements with the release of its o1 reasoning model, the company might even reconsider its stringent AGI clause that affects its partnership with Microsoft. There are whispers about OpenAI nearing—or potentially achieving—this landmark, all while Altman believes such a breakthrough could happen within five years with existing technology and minimal societal disruptions.
As 2025 approaches, OpenAI is likely to have more thrilling deliveries for its user base. Altman’s parting words were a tantalizing reminder, "Many great updates we have coming were mentioned not at all or very little!"
