Photo by Digital Buggu on Pexels

A new report indicates that generative AI projects are struggling to deliver significant financial returns for most businesses. Research conducted by NANDA, an organization stemming from MIT Media Lab, reveals that a mere 5% of generative AI pilot projects actually reach production and generate measurable monetary value.

The study, which included interviews with enterprise leaders, analysis of public AI initiatives, and a broad survey, points to back-office workflow automation as the most fruitful area for gen AI, yielding cost savings through decreased reliance on external agencies. While individual employees report productivity gains from tools like ChatGPT (90%), these benefits haven’t scaled to create significant institutional-level impact.

One of the key roadblocks cited by organizations is the lack of contextual understanding within current gen AI models. The report emphasizes the crucial role of strategic partnerships with vendors capable of providing tailored systems that adapt to a company’s unique needs. The media and telecom sector has seen the most positive impact, while the energy and materials industries have experienced minimal adoption. Sales and marketing remain the most popular deployment areas, but complex tasks continue to be handled primarily by human workers.

The report’s overall message underscores the critical importance of vendor collaborations in ensuring successful generative AI deployments and ultimately, a return on investment.