Microsoft's trajectory in the field of artificial intelligence is undergoing a significant transformation. Once ambitious in developing its own large models, the company has now shifted its focus to collaboration with OpenAI and introduced a new "calibration" feature aimed at addressing the "hallucination" issues in AI-generated content.
From Self-Development to Leveraging Partnerships: The Evolution of Microsoft's AI Strategy
Initially, Microsoft invested heavily in generative AI (GenAI) with the hope of creating AI models that could rival or even surpass OpenAI's GPT series. However, during the lengthy research and development process, Microsoft encountered bottlenecks, and the progress of its self-developed models did not meet expectations. To break the deadlock, Microsoft turned to "distillation" technology, leveraging OpenAI's models to generate large amounts of high-quality data, which was then used to train its own models. Although this strategy yielded some results, with personnel changes and project adjustments, Microsoft's AI strategy seemed to become more enigmatic.
Image Source Note: The image was generated by AI, provided by the image licensing service Midjourney
"Calibration" Feature: A New Breakthrough in AI Content Security
To enhance the reliability of AI-generated content, Microsoft introduced the "calibration" feature. This technology ensures the accuracy of output results by real-time detection and correction of "hallucination" content generated by AI. The term "hallucination" refers to text generated by AI that lacks factual basis and may even contain false information.
The working principle of the "calibration" feature is as follows: First, the system scans the AI-generated text to identify statements lacking supporting evidence; then, it marks these statements and provides explanatory notes; finally, the system automatically modifies these statements to align with real data.
Reasons Behind Microsoft's Strategic Adjustment
The reasons for Microsoft's strategic adjustment are multifaceted. On one hand, the development cycle of self-developed large models is long, costs are high, and technical bottlenecks are encountered. On the other hand, OpenAI's models have already achieved significant success in the market, and Microsoft aims to enhance its competitiveness in the AI field quickly through collaboration with OpenAI. Additionally, as AI application scenarios continue to expand, higher demands are placed on the reliability of AI-generated content, and the introduction of the "calibration" feature is intended to meet this need.
Microsoft's adjustment in its AI strategy reflects the intense competition in the AI industry and the growing concern for AI safety. Through collaboration with OpenAI and the introduction of the "calibration" feature, Microsoft has charted a new course for development in the AI field. However, the development of AI technology is still in its infancy, and there are many challenges to overcome in the future.