What leaders still get wrong about AI

Few organizations have successfully parlayed artificial intelligence experimentation into large-scale initiatives that move the needle on critical business metrics like revenue growth or productivity gains.

There are myriad reasons organizations have yet to derive value from AI projects, but they share a theme: Firms embrace AI as the next flashy technology initiative rather than starting with a defined business problem and then working through necessary organizational changes. 

“A common pattern we see is that organizations are applying yesterday’s best practices to an inherently different technology,” said Nick van der Meulen, a research scientist at the MIT Center for Information Systems Research. “They govern AI like legacy IT, mistake productivity shaves for enterprise value, and treat AI as another skill to acquire when it’s actually redefining what skilled work looks like.” 

The organizations making progress have realized that rethinking how they operate is  the AI strategy. Van der Meulen and his colleagues at MIT CISR have identified five common mistakes that impede AI success, and they’ve articulated what organizations should do instead.

1. Treating AI as something you do, not a tool to get results

Organizations can be so consumed with doing something AI-related that they forget that the technology is another tool in the toolbox — and a complex one at that. To generate real value from AI, organizations must invest in the right capabilities and practices to do the work properly as opposed to expecting instantaneous results.

“AI is advanced data science, and you need to have the right capabilities in order to work with it and manage it properly,” said Barbara Wixom, a principal research scientist at MIT CISR. “You need the right capabilities in order to work with it and manage it properly.”

Wixom, along with MIT CISR research fellow Cynthia Beath, outlined three principles for implementing successful AI projects:

• Invest in advanced data capabilities, including data science, data management, data platforms, and acceptable data use.
• Involve all stakeholders in the AI journey to fully understand AI’s potential and costs while providing useful feedback on AI model performance.
• Focus on value realization in the form of increased revenue or reduced expenses, and back initiatives that are clearly aligned with those goals.

2. Starting AI projects without a clear path to value

While many knowledge workers tap generative AI tools for quick productivity bursts — to write emails or create presentations, for example — those quick-fix use cases shouldn’t be conflated with broader projects that will deliver enterprise value. 

In their book “Data Is Everybody’s Business,” Wixom, Beath, and Leslie Owens identify five steps to help companies identify projects that will create value with AI. Their approach is framed in the context of a common healthcare problem and solution: predicting which patients are at higher risk for a fall, and establishing preventive measures. 

• Step 1: Collect the right data. In this case, that means patients’ medical histories, using electronic records and bedside medical devices.
• Step 2: Generate insights. Employ AI to analyze data assets and drive predictions.
• Step 3: Take action. Apply insights to improve best practices. For example, a hospital could create a policy so that a system alerts the nearest nurse when an at-risk patient’s movements indicate that they may try to get out of bed.
• Step 4: Create value. New policies result in fewer preventable falls, increased patient satisfaction, and shorter patient stays.
• Step 5: Monetize value. Changes are linked to a specific value objective, such as reducing cost of care, which is a positive for hospitals operating under performance-based contracts.

3. Getting stuck in pilots instead of scaling

Organizations see the biggest financial gains when they make the leap from pilots to new ways of working built around AI. Those that mature beyond use cases and automating select processes have a better chance at scaling AI across the business and embedding it in all kinds of workflows. However, that process often stalls out due to cost, training, and governance challenges.

CISR researchers Stephanie Woerner, Peter Weill, Ina Sebastian, and Evgeny Káganer identified four steps that must happen before organizations can move to the next stage of enterprise AI maturity:

• Ensure that AI investments are aligned with strategic goals so they can offer measurable, scalable value.
• Architect modular, interoperable platforms and data ecosystems to enable enterprisewide intelligence.
• Create AI-ready people, roles, and teams while redesigning work around AI capabilities.
• Embed compliant, human-centered AI practices by design.

“Now is the time for executive teams to align, commit, and lead the charge toward enterprise-scale AI by developing a playbook for strategy, systems, synchronization, and stewardship,” the researchers write.

AI Executive Academy

In person at MIT Sloan

Register Now

4. Overlooking how AI changes the business itself

AI is a key enabler for new business models. Weill, Woerner, Sebastian, and CISR industry research fellow Gayan Benedict updated their earlier findings about digital business models to create four new categories for the AI era. 

With the explosion of technologies like machine learning, generative AI, agentic AI, and robotic AI, the researchers expect business models to become increasingly outcome oriented and enabled by autonomous AI. The four new models are:

• Existing+. Here, companies augment an existing business model with AI capabilities. Take, for example, a financial services company that enhances its advisory services by using AI to provide personalized recommendations. 
• Customer Proxy. Companies applying this model achieve customer outcomes through predefined processes executed by AI — for instance, setting parameters that would automatically manage a customer’s investment portfolio with minimal to no human intervention.
• Modular Creator. These companies may use AI to assemble reusable modules to help achieve customer outcomes without a predetermined process. This could take the form of a financial services company creating a bundle of investment, insurance, and credit products that align with a customers’ goals.
• Orchestrator. Orchestrators achieve customer outcomes using AI to assemble an ecosystem of complementary products and services with no predetermined process. Consider, for example, a fully managed and automated wealth solution that continuously optimizes a customer’s investment portfolio.

5. Mistaking productivity gains for value

Early interest in generative AI was fueled mostly by individuals looking to work more efficiently. While some elevated their personal productivity, larger enterprise gains were more difficult to achieve. According to Wixom and van der Meulen, organizations are failing to recognize that there are two types of generative AI: tools that enhance personal productivity, and tailored solutions used to achieve strategic business goals. 

Generative AI tools help users more efficiently summarize documents or brainstorm ideas, which saves a few minutes of effort with each task. Generative AI solutions operate on a broader scale, like a large language model that provides real-time coaching to call center agents, helping to generate increased efficiencies or revenue growth. Generative AI solutions need to be integrated with processes and systems, which is far more complex than employing generative AI for personal use.

Organizations need to pursue both types of generative AI, “creating a virtuous cycle where increased employee awareness and proficiency with generative AI tools drives new generative AI capability building and inspires innovation with generative AI solutions,” the researchers write.

Cynthia Beath is an academic research fellow at the MIT Center for Information Systems Research and professor emerita at the University of Texas. Her research interests include organization redesign for the digital era, the management of data assets, and the organizational impacts of AI.

Gayan Benedict is an industry research fellow at MIT CISR and a technology partner at PwC Australia.  

Evgeny Káganer is a professor at IESE Business School and a research collaborator at MIT CISR. His research explores how digitalization and AI reshape business models and organizations.

Leslie Owens is the associate director of the Martin Trust Center for MIT Entrepreneurship, where she manages strategic initiatives. She was formerly the executive director of MIT CISR and a senior lecturer in IT at the MIT Sloan School of Management. 

Ina Sebastian is a research scientist at MIT CISR. She studies how large enterprises transform for success in the digital economy, with a focus on digital partnering, value creation, and value capture in digital models.

Nick van der Meulen is a research scientist at MIT CISR. He conducts academic research that targets the challenges of senior-level executives, with a specific interest in how companies need to organize themselves differently in the face of continuous technological change. He is one of the faculty members who teaches the MIT Sloan Executive Education course AI Executive Academy. 

Peter Weill is a senior research scientist at MIT Sloan and chairman of MIT CISR. His work explores future trends, such as digital business models, IT investment portfolios, and AI maturity models, to help organizations maintain a competitive edge.

Barbara Wixom is a principal research scientist at MIT CISR. Since 1994, her research has explored how organizations generate business value from data assets. Her methods include large-scale surveys, meta-analyses, lab experiments, and in-depth case studies. She teaches the MIT Sloan Executive Education course Data Monetization Strategy: Creating Value Through Data.  

Stephanie Woerner is a principal research scientist at MIT Sloan and the director of MIT CISR. She studies how companies use technology and data to make more effective business models, as well as how they manage associated organizational change, governance, and strategy implications.