This move comes after Microsoft's significant investment in OpenAI and the integration of OpenAI's technology into its products. However, the company's recent developments suggest a shift towards independence from OpenAI. While Microsoft's new models do not directly replicate the capabilities of OpenAI's GPT-4, they demonstrate that the company's AI strategy extends beyond its partnership with OpenAI.
Key takeaways:
- Microsoft has unveiled its own small-scale AI models, Phi-2 and Orca 2, which are said to perform as well as or better than larger models, including OpenAI's ChatGPT.
- Phi-2 is a smaller large language model with 2.7 billion parameters, trained in 14 days, and is cheaper to run due to its reduced need for intensive computing power.
- Orca 2 is designed to enhance reasoning abilities in smaller AI models, with performance levels similar to or better than models 5-10 times larger.
- These developments suggest that Microsoft is not solely reliant on OpenAI for AI technology, despite its $10 billion investment into the company.