OpenAI CEO Sam Altman and other executives participated in a Reddit Ask Me Anything (AMA) session, addressing questions about AI competition, open source strategy, pricing, and the future of OpenAI’s models. This discussion comes as OpenAI deals with increasing competition from Chinese AI firms like DeepSeek, regulatory challenges in Washington, and a major data center project underway, while reportedly preparing for a significant funding round.
Altman acknowledged that DeepSeek's progress has reduced OpenAI's lead in AI. He admitted that OpenAI's closed-source approach might have been a mistake and suggested the need for a different open source strategy. However, he mentioned that not everyone at OpenAI agrees with this, and it is not a top priority currently. Kevin Weil, OpenAI’s chief product officer, indicated that the company might consider open sourcing older models that are no longer cutting-edge, though no timeline or specifics were provided.
In response to one of the questions, he said, "Yes, we are discussing. I personally think we have been on the ‘wrong side of history’ here and need to figure out a different open source strategy; not everyone at OpenAI shares this view, and it's also not our current highest priority."
A key topic was OpenAI’s “reasoning” AI models, like o3-mini, which launched recently. These models do not currently reveal their complete decision-making process to prevent competitors from accessing training data. However, Weil hinted that OpenAI might soon disclose more of its models’ reasoning, though revealing the entire thought process could make the models easier to replicate.
Regarding ChatGPT pricing, Altman and Weil denied any plans to increase costs. Altman expressed a desire to make ChatGPT cheaper over time, if feasible. This follows Altman's earlier admission that OpenAI is losing money on its $200-per-month ChatGPT Pro plan, which offers premium access to AI models.
Weil confirmed that increased compute power directly correlates with better AI performance, which is a primary reason for Stargate, OpenAI’s upcoming large-scale data center project, expected to need billions in funding.
Altman was also questioned about recursive self-improvement in AI, where systems could enhance themselves without human intervention. He noted a “fast takeoff” scenario as more likely than he previously thought, without elaborating. Concerns were also raised about OpenAI's partnership with the U.S. government for nuclear defense research. Weil dismissed worries about the models being used to develop nuclear weapons, expressing trust in the scientists involved.
The AMA provided brief updates on OpenAI’s forthcoming models, including the o3 Reasoning Model, expected to launch in more than a few weeks but less than a few months, and GPT-5, for which there is no confirmed release date. OpenAI is also working on DALL-E 4, a successor to DALL-E 3, which has lagged behind competitors.