Action items for AI decision makers in 2026

Artificial intelligence has dominated economic and business attention for the past several years. But the hype cycle is slowing as organizations confront the challenges of enterprise AI deployment and the need to drive tangible business value.

In an MIT Sloan Management Review article on their AI predictions for 2026 and a related video, Thomas Davenport and Randy Bean say that they expect a level-set year for AI. 

Davenport, a Babson professor and a fellow at the MIT Initiative on the Digital Economy, and Bean, an adviser to Fortune 1000 companies, advise business and technology leaders to focus on the organizational structure, tools, and strategies required for deploying the technology at enterprise scale in the year ahead.

Here are their five AI insights for 2026, along with action items for decision makers:

1. Agentic AI isn’t ready for prime time — yet 

Despite its meteoric rise predicted last year, Davenport and Bean are dialing back expectations for agentic AI, a class of AI systems that can perceive, reason, and complete tasks independently or with minimal human supervision. 

Ongoing hallucinations and mistakes, coupled with the ease with which hackers can hijack an agentic AI system using prompt injection and other methods, has been a wakeup call that has slowed adoption. “Companies will continue to have some human in the loop” to create guardrails for agentic AI, Davenport said, but that undermines its promised productivity advantage. 

While some industry watchers expect that it will be a decade or more before such issues are ironed out, Davenport and Bean are more bullish, predicting that AI agents will handle most transactions in many large-scale business processes within five years.

Action item: Companies should begin envisioning how AI agents can facilitate new ways of working, starting with use cases that can be reused across the organization. It’s also important to start fostering internal capabilities to create and test agents.

2. The AI bubble will deflate, with economic ramifications 

AI has monopolized boardroom discussions and inflated the stock market. This year, Davenport and Bean expect a reckoning, likely sooner rather than later. The emphasis on user growth over profits is reminiscent of the dot-com bubble, Davenport and Bean write. “Often technologies are overestimated in the short term, but their transformational impact is very much underestimated in the long term,” Bean said. 

Action item: Companies should take full advantage of the AI technologies they already have while also exploring the impact that investments can have on future business strategies. 

AI Executive Academy

In person at MIT Sloan

Register Now

3. Generative AI should become an enterprise rather than individual resource

Organizations have mostly taken an individual-level approach to generative AI, with employees using the technology to boost their own productivity. It’s less common for companies to apply generative AI to enterprise workflows and processes. Until they do, it will be difficult to aggregate results and quantify business value.

Action item: Companies should move beyond individual productivity and consider enterprise-oriented generative AI use cases, such as facilitating new-product development or enriching the customer experience, in order to drive value.

4. The optimal reporting structure for AI has yet to be determined

Support for data and AI leadership roles is at record highs in large enterprises, but it remains unclear who owns responsibility for AI or what constitutes the optimal reporting structure. In the 2026 AI & Data Leadership Executive Benchmark Survey, 38% of responding companies said that they have appointed a chief AI officer or an equivalent role, but there was little consensus on to whom that job reports. (Currently it’s split among business, technology, and transformation leadership.) “It’s likely that the diverse reporting relationships are contributing to the widespread problem of AI (particularly generative AI) not delivering sufficient business value,” Davenport and Bean write.

Action item: Companies should consider appointing an individual to unify data, analytics, and AI and report to business leadership. “As AI gains in prominence, it’s being elevated into a C-suite role,” Bean said. He cited JPMorgan as a prime example: There, a new AI-focused executive sits on a 14-person operating committee reporting directly to chairman and CEO Jamie Dimon.

5. “AI factories” can help organizations accelerate value

Davenport and Bean define “AI factories” as “combinations of technology platforms, methods, data, and previously developed algorithms that make it fast and easy to build AI systems.” 

“It’s not just putting up a big data center and filling it full of GPU chips. It’s a capability within an organization,” Davenport said.

Instead of requiring data scientists and businesspeople to replicate work or figure out what data is available, these AI factories establish a tool and business process foundation that allows the firm to efficiently and cost-effectively build out AI at scale.

Action item: Forward-thinking firms should use this year to ramp up AI factories to expand the number of use cases internally and drive more economic value from their AI investments.

Read the Article: “Five Trends in AI and Data Science for 2026”

Watch the Video

Thomas H. Davenport is a professor of information technology and management and faculty director of the Metropoulos Institute for Technology and Entrepreneurship at Babson College, as well as a fellow of the MIT Initiative on the Digital Economy. His latest book is “The New Science of Customer Relationships: Delivering the One-to-One Promise With AI.”

Randy Bean has been an adviser to Fortune 1000 organizations on data and AI leadership for more than four decades. He is the author of “Fail Fast, Learn Faster: Lessons in Data-Driven Leadership in an Age of Disruption, Big Data, and AI.”