OpenAI CEO Sam Altman made a striking admission during a Reddit "Ask Me Anything" session, acknowledging that his company has been "on the wrong side of history" regarding its closed-source approach to artificial intelligence development.
The revelation comes as Chinese AI firm DeepSeek gains momentum with its open-source R1 model, which reportedly achieves comparable performance to OpenAI's systems at substantially lower costs.
"I personally think we need to figure out a different open source strategy," Altman stated, though he noted that not everyone at OpenAI shares this perspective and it's not currently the company's highest priority.
This potential shift marks a notable departure from OpenAI's recent proprietary development approach. The company, which began as a non-profit in 2015, has faced criticism for moving away from its original open-source mission.
Kevin Weil, OpenAI's chief product officer, indicated the company might begin by open-sourcing older models that are no longer cutting-edge. The company is also considering revealing more about its models' reasoning processes, following DeepSeek's example of showing complete thought chains.
Altman acknowledged that DeepSeek's emergence has impacted OpenAI's market position, stating, "We will produce better models, but we will maintain less of a lead than we did in previous years."
The discussion also touched on other developments, including:
- Plans to make ChatGPT more affordable
- Ongoing work on successor to DALL-E 3
- Development of the next reasoning model, o3
- Continued investment in compute power for model improvements
This potential strategic shift could reshape the AI industry landscape, potentially accelerating innovation through increased transparency and collaboration in AI development.