An article from Harvard Business Review (HBR) analyzes the issue of low efficiency in AI initiatives. According to an MIT report, up to 95% of generative AI projects fail, and an NBER study of 6,000 executives showed that 90% have not noticed productivity growth over the last three years.
What Happened
The research identified a critical problem in the approach to technology implementation: leaders often fall into the "urgency trap." Instead of developing a long-term strategy, companies attempt to use generative AI as a quick fix for current operational problems, leading to widespread initiative failures.
Context
The problem is exacerbated by the gap between demonstrating technological capabilities (demos) and achieving real return on investment (ROI). The current market is characterized by a rise in "thin wrappers" that do not solve fundamental business problems and quickly lose relevance.
Why It Matters for the Industry
For the industry, this means a risk of growing misdirected investments and overloading engineering teams with tasks that do not bring measurable value. In the long term, market consolidation is expected around companies capable of providing systemic AI integration and creating a sustainable technological foundation, including reliable data pipelines and evaluation systems (evals).
Why It Matters for Users
Readers and specialists must distinguish between implementing hyped tools for minor tasks and building a sustainable technological foundation. There is a risk of increased skepticism from businesses and increased pressure on technical teams to prove measurable ROI from current implementations.
Sources
Author
Look at AI, Editorial Staff