A recent preliminary report from MIT NANDA reveals a significant gap between the anticipated benefits of generative artificial intelligence (AI) and its actual performance in the business sector. The study, titled "The GenAI Divide: State of AI in Business 2025," found that an overwhelming 95% of the organizations surveyed did not experience measurable returns from their investments in generative AI. While a select few pilot projects—about 5%—managed to generate millions of dollars, the majority of companies are left grappling with unmet expectations amidst a backdrop of corporate spending on generative AI estimated at $30 to $40 billion.
According to the report, the disconnect is not attributed to the quality of AI models or regulatory issues but rather to challenges in implementation. Key factors hindering success include unstable workflows, limited system capabilities to maintain context and incorporate feedback, and weak alignment between AI solutions and the daily tasks of employees. Despite over 80% of organizations exploring or launching pilots with general tools like ChatGPT and Copilot, only 5% of specialized corporate systems for specific tasks have progressed to full industrial deployment.
The report also highlights differences in development approaches. Solutions created with external partners achieved implementation in approximately 67% of cases, whereas internal developments reached this milestone in only about 33% of cases. However, the authors caution that the correlation between external collaboration and superior outcomes does not imply causation, as companies define the success of their projects variably.
The findings suggest a pressing need for businesses to reassess their strategies regarding generative AI to bridge the gap between expectations and reality. This revelation may prompt competitors to rethink their AI integration approaches, focusing on more effective implementation strategies to realize the full potential of their investments.
Informational material. 18+.
