The launch of o1 comes amid reports that the AI industry is hitting a limit with word-predicting models, which are no longer becoming more capable with size. OpenAI's new reasoning models are seen as an attempt to overcome this hurdle. Despite some limitations, the company and many of its rivals see this "reasoning" approach as the way forward for AI. However, scaling up these models could require significant computing power and energy, raising concerns about efficiency and environmental impact.
Key takeaways:
- OpenAI has launched a new generative-AI program called o1, which is said to have far greater capabilities and more closely approximate human thinking than any previous software.
- Unlike previous models, o1 is not trained to predict human thoughts but to produce, or at least simulate, thoughts on its own. This approach is seen as a way to overcome the limitations of prediction-based models.
- Despite the improvements, o1 still has limitations and its understanding is constrained to specific subjects where a solution can be confirmed as true. This makes it more narrowly applicable than previous models.
- OpenAI's approach of scaling up AI models, despite showing diminishing returns and requiring enormous computing power, is being followed by other major AI companies, indicating a shift in the industry towards reasoning-based models.