OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI

OpenAI's new artificial intelligence model, internally known as Orion, did not meet the company's performance expectations, according to anonymous sources. Despite completing an initial training round in September, Orion struggled with answering coding questions it hadn't been trained on and overall, did not significantly surpass previous versions of the technology behind ChatGPT.

The sources also stated that Orion is not considered a major improvement from OpenAI's existing models. Specifically, it is not seen as a big leap from GPT-4, the system that originally powered the company's flagship chatbot, compared to the progress made from GPT-3.5 to GPT-4.

Key takeaways

OpenAI has been training a new artificial intelligence model, internally known as Orion, which it hoped would significantly surpass prior versions of the technology behind ChatGPT.
Orion did not meet the company's desired performance levels, according to two anonymous sources.
Orion struggled with answering coding questions that it hadn't been trained on.
Overall, Orion is not considered to be as big a step up from OpenAI’s existing models as GPT-4 was from GPT-3.5.

OpenAI, Google and Anthropic Are Struggling to Build More Advanced AI

Key takeaways

Discussion (0)