Key points
- Just 16% of surveyed organizations report extracting transformative value from their current AI investments.
- Most firms remain in pilot phases or use AI for basic assistance, with only 12% running agents in production.
- Experts attribute the performance gap to missing architectural layers that provide business context to models.
An MIT Technology Review Insights poll reveals that fewer than one in five organizations are deriving transformative value from their artificial intelligence deployments. The survey indicates that while every respondent uses AI to some degree, the majority are not seeing significant operational improvements. Forty percent of respondents reported moderate gains in specific tasks, while 21% noted meaningful gains.reported The remaining 23% said they were seeing minimal returns or none at all.
The claim
The data suggests a disconnect between model capabilities and actual organizational utility. Most enterprises are still piloting AI use cases, according to 38% of respondents, or using tools like co-pilots and chat assistants to help employees, cited by 35%.reported Only 12% of enterprises have AI agents in production, and a mere 3% say they use AI to perform most of the work in key processes.reported
The response
Prukalpa Sankar, co-founder of Atlan, stated there is a large gap between what models can do and how they are useful inside organizations. He argued that while artificial general intelligence exists for real-life intents, it is not yet useful in most companies. Sankar identified the missing layer as context, or how well AI understands a specific business.
What the evidence means
Sankar described a context layer as architectural infrastructure sitting between business systems and AI platforms. This layer translates enterprise intelligence into governed, machine-usable context for agents. Without it, he said, it is like hiring a smart intern, blindfolding them, and throwing them into the organization. The report notes that outcomes may be wrong or confused without this understanding.
In practice
For enterprise IT buyers, the evidence indicates that simply deploying models is insufficient for value extraction. Organizations must implement context layers to translate systems of record into usable data for AI agents. Current adoption remains largely experimental, with most firms using assistants rather than autonomous agents. Buyers should expect limited production readiness for complex processes, as only 3% of enterprises use AI for most key work.reported
Why it matters
- IT leaders must prioritize context infrastructure over model selection to move beyond pilot programs.
- Budgets should account for the integration layer that translates business data for AI agents.
- Expectations for autonomous AI in key processes should remain low, given only 3% adoption.
- Most firms will see moderate or minimal returns, requiring realistic ROI timelines.
Topics: enterprise AI, MIT Technology Review, AI context layer, Atlan, AI adoption
Updated
Numbers in context
Circle area is proportional to the magnitude; labels retain signs. Figures may describe different measures. Read their contexts above.Ask the article
Local keyword search of this article; no generated answers.Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Questions readers ask
What percentage of companies see transformative value from AI?
According to the MIT Technology Review Insights poll, only 16% of surveyed organizations are extracting transformative value from their AI deployments.
How are most enterprises currently using AI?
Most are piloting use cases (38%) or using AI to assist employees with tools like co-pilots (35%), rather than full automation.
What is a context layer in AI architecture?
It is infrastructure that sits between business systems and AI platforms, translating enterprise intelligence into governed, machine-usable context for models.
Connected stories
Reading receipt
- Just 16% of surveyed organizations report extracting transformative value from their current AI investments.
- Most firms remain in pilot phases or use AI for basic assistance, with only 12% running agents in production.
- Experts attribute the performance gap to missing architectural layers that provide business context to models.