A recent report from MIT’s Media Lab Project NANDA found that despite enterprises spending $30–40 billion on generative artificial intelligence (AI), 95% of organizations see no business payback. According to the authors, “just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable [financial] impact.”*

That’s sobering, but it also gives us a lot to think about as we consider applying AI in engineering and design processes. One of the main takeaways from the NANDA study is that custom AI projects rarely succeed. Only 5% of in-house-developed AI tools reach full-scale rollout, which means 95% fail to progress beyond the pilot phase.

The hype around AI is sky-high, so failure is especially frustrating for organizations expecting fast, dramatic results. According to the study, most AI applications are currently used in business areas such as legal, customer service, or HR departments. There, AI must handle a variety of unpredictable scenarios; as soon as it can’t, it’s seen as a failure.

This article appeared in the October 2025 Issue of BENCHMARK.