After a whirlwind of activities during OpenAI’s 12-day “shipmas” celebration, CEO Sam Altman turned to social media, specifically X, to invite a flood of suggestions by asking, “What would you like OpenAI to build or improve by 2025?” As anticipated, the post quickly garnered massive attention, with users eagerly detailing their ideas and expectations.
One user immediately suggested designing a drag-and-drop interface that connects multiple AI models to create refined outputs. This concept isn’t entirely new, having been previously discussed in OpenAI’s Developer Forum. Contributors there believe that optimizing specific models for individual tasks could yield superior results.
Altman found this proposal intriguing, admitting it was a concept he hadn’t previously considered. Another user laid out a wishlist of enhancements, including a more powerful successor to GPT-4o and seamless integration of GPT/o/Sora into ChatGPT. They also advocated for longer context inputs, more frequent updates, and a reasonably priced “middle tier” subscription plan in the $50-70 range. Altman cryptically hinted at some forthcoming developments that might meet these requests when he suggested the user would be pleased with OpenAI’s progress over the year.
Additionally, there was a request for easing the restrictions on the next-gen AI models, proposing a so-called “cleaner” version. Altman acknowledged the demand for what he described as a “grown-up mode.”
There were also calls for family accounts, enabling the creation of child-specific profiles with safeguards in place. This feature could allow children to explore within parental limits while potentially offering insights into their learning and interests.
Meanwhile, @davidlach on X asked for better agents, to which Altman enigmatically replied, “Happy 2025.” OpenAI appears to be on track to introduce a new AI agent, known as Operator, by January 2025. Late last year, Kevin Weil, Chief Product Officer, expressed optimism, sharing, “I think 2025 will be the year agentic systems finally become mainstream.”
So, what might we expect from OpenAI in 2025 and beyond? There’s a great deal of anticipation around the possibility of a stronger GPT-4o successor with enhanced capabilities, providing longer context and improved memory. Users have also expressed interest in a “grown-up mode” with fewer restrictions, though its development is mired in discussions about security and privacy. Additionally, Sora might see significant improvements, unlocking more potential and possibly supporting longer videos.
Altman remains optimistic about the possibilities in AI, mentioning common themes like AGI, agents, enhanced upgrades, better memory, and more personalized features as future goals. Curiously, some exciting updates OpenAI plans to implement weren’t mentioned in public discussions at all.
Amid all these developments, AGI remains high on the agenda, with speculation that OpenAI might have already reached this significant milestone following the rollout of its o1 reasoning model. In a related twist, reports suggest OpenAI might remove a restrictive AGI clause that could affect its alliance with Microsoft to secure ongoing funding and cloud resources.
We can’t rule out more surprises from OpenAI in 2025, as Altman teased: “Many great updates we have coming were mentioned not at all or very little!” This hints that the company is poised to unveil innovations that have yet to enter the public discourse, keeping the tech world on its toes.
