Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Microsoft Looking to Pursue an Open Relationship With OpenAI

Dec 23, 2024 - gizmodo.com
Microsoft and OpenAI have maintained a collaborative relationship, with Microsoft investing heavily in OpenAI for early access to advanced AI models integrated into its productivity software. However, Microsoft is reportedly exploring additional models for its 365 Copilot product, as it finds OpenAI's GPT-4 model too costly and slow for enterprise needs. Copilot 365, an AI assistant in Microsoft applications like Word and PowerPoint, aims to streamline tasks such as data retrieval and meeting summaries. Despite its potential, feedback from Microsoft employees and a Gartner survey indicates dissatisfaction with Copilot 365's performance and cost, with some labeling it as ineffective and overpriced.

To address these concerns, Microsoft is developing its own smaller models, like Phi-4, and considering customizing other open-weight models to enhance Copilot 365's efficiency. This move allows Microsoft to reduce dependency on OpenAI while still benefiting from its investment in the startup. By building in-house models, Microsoft hedges its bets in the evolving AI landscape, similar to how various search engines were eventually overshadowed by Google. This strategy positions Microsoft to adapt to potential shifts in AI technology dominance.

Key takeaways:

  • Microsoft is considering adding more AI models to its 365 Copilot product beyond those developed by OpenAI, due to concerns about cost and speed.
  • There is dissatisfaction among Microsoft employees and customers regarding the performance and value of 365 Copilot, with some describing it as "terrible" and "gimmicky."
  • Microsoft is developing its own in-house AI models, such as Phi-4, to potentially reduce reliance on OpenAI and improve the efficiency of 365 Copilot.
  • Despite its investment in OpenAI, Microsoft is hedging its bets by developing its own AI technology, preparing for potential shifts in the AI landscape.
View Full Article

Comments (0)

Be the first to comment!