New leadership and management thinking from MIT experts

Use imagination to make the most of generative AI

Lending standards can be too tight for too long, research finds

Ideas Made to Matter

Artificial Intelligence

Implementing AI and managing relationships: 5 ideas from MIT Sloan Management Review


As artificial intelligence matures and expands within enterprises, leaders across industries are struggling to get everyone on board. At the same time, they must manage customer and employee relationships amid shifting expectations in an era of digital transformation.

The latest ideas from MIT Sloan Management Review consider how to overcome the barriers of AI implementation and go all in on putting AI tools into production. Leaders will also learn how to know what customers want, how to avoid a toxic workplace, and how to run effective brainstorming sessions.

Overcome 3 common barriers to using AI tools

AI-powered decision-making tools have the potential to increase efficiency, improve service quality, reduce costs, and boost revenue. But this only happens if workers use the tools. Often, they do not.

AI projects face resistance from front-line workers in industries ranging from health care to retail, MIT Sloan professorwrites, along with co-authors Mark Sendak and Suresh Balu. This resistance typically stems from three conflicts of interest among AI developers, corporate leadership, and end users. A more holistic approach to implementation can break through these barriers.

Problem 1: AI tools benefit the organization, not the end user. This is common when organizations use predictive analytics to increase value downstream, as it forces end users to input data or make decisions unrelated to their role. To address this, AI developers should focus on problems that end users face in their own day-to-day work, while managers should offer tangible incentives for using the tools.

Problem 2: Tools require additional end-user labor. Increased engagement with AI tools, especially those outside workers’ typical technology workflows, only makes the job harder. Tools that can automate data retrieval, testing, and validation, and that can provide insight within the applications that workers already use, should minimize the impact on end-user workload.

Problem 3: Tools curtail end-user autonomy. Prescriptive tools that offer evidence-based decision support — and track whether someone accepts those recommendations — infringe on end users’ intuitive judgments. AI tools should assist end users in making decisions while leaving the “final call” to them. Understanding this give and take requires end user involvement early in the development life cycle.

Read: AI on the Front Lines

Know the 5 traits of an ‘AI powerhouse’ company

Nearly two-thirds of enterprises have yet to see value from their AI investments, and 45% perceive AI as a risk to their business in some way. That’s because these firms tend to be dabbling with AI and have yet to put their AI tools into production, write Thomas H. Davenport, a visiting scholar at the MIT Initiative on the Digital Economy, and Randy Bean, CEO of NewVantage Partners.

Mastercard isn’t one of those companies. Support for AI starts with CEO Michael Miebach and has been built through a combination of acquisition and internal talent development. The example of Mastercard, a self-described “AI powerhouse,” suggests there are five pillars of a company that is all in on AI:

  1. Powering products and services. Mastercard started with fraud detection but plans to apply AI to all components of the payment cycle.
  2. Powering internal business operations. Predictive applications support processes ranging from business forecasting (with 99% accuracy) to server maintenance.
  3. Supporting customers. Mastercard works with corporate customers to identify their own use cases for AI — and to create a proof of concept in as little as six weeks.
  4. Pursuing AI for good. The company is running AI projects targeting community development, microfinance, and building data science talent in underserved areas.
  5. Prioritizing ethical AI. In its AI development efforts, Mastercard emphasizes customers’ ownership of, control over, and ability to benefit from their own data.

Read: Becoming an ‘AI powerhouse’ means going all in

Rethink these assumptions about what customers want

Assumptions can steer business leaders toward beneficial decisions and align stakeholders on common viewpoints, MIT Sloan principal research scientist explained in a recent webinar. However, assumptions also reinforce unconscious bias and screen out innovations that go against the grain. In an ever-changing world, companies need to rethink these 5 common assumptions about customer expectations.

Customers value the human touch. Many customers actually prefer self-service and believe that a human slows down the interaction.

In-person experiences are better than digital. The digital experience lets companies reach a worldwide audience while offering better convenience at a lower cost.

People won’t pay full price for digital. People pay for value, and convenience is an important part of that value. By providing convenience, digital can provide significant value.

Pandemic-era service restrictions are only temporary. It turns out that certain services cut back during the pandemic, such as daily turndowns in hotel rooms, may be overvalued.

The old way was the right way. Traditional business models not only predate the pandemic; they may predate the smartphone, the Internet, or even the telephone.

As business leaders challenge these assumptions, these are the questions they should keep in mind:

  • Who’s the human salesperson helping — the customer, or our inefficient and outdated internal processes?
  • How do we optimize in-person and digital interactions for customers, not just ourselves?
  • How do preferences vary by customer segments?
  • When is digital better — and can we charge more, not less, for it?
  • How much did customers really need the services we cut back during the pandemic?

Watch: Rethinking customer expectations for 2022 and beyond

Avoid the 5 attributes of a toxic corporate culture

Many characteristics of a company can contribute to a bad culture, but there’s a difference between elements of culture that are irritating or disappointing and those that are truly toxic. In a study of more than 1.3 million Glassdoor reviews, MIT Sloan senior lecturer and his co-authors identified five common attributes of a toxic corporate culture. These are the characteristics that have the largest negative impact on how employees rate corporate culture.

  1. Noninclusive representation of employees by gender, race, sexual identity and orientation, disability, and age — coupled with a culture of cronyism and general unwritten favoritism.
  2. Disrespectful treatment of employees as shown in a lack of consideration, courtesy, and dignity for others.
  3. Unethical and dishonest behavior — or, worse, failure to comply with applicable state and federal regulations.
  4. Cutthroat work environments in which colleagues actively undermine each other.
  5. Abusive management that openly bullies, condescends, or talks down to employees.

A toxic corporate culture comes at a high cost. Companies find it harder to both attract and retain talent, while employees who remain on staff are less productive and more likely to suffer from a chronic disease. And no one is immune: Even companies with high ratings for corporate culture are likely to contain “pockets of cultural toxicity” with business units, job functions, or geographies.

Read: Why every leader needs to worry about toxic culture

Bring constructive criticism to the creative process

The primary ground rule for brainstorming sessions — no criticism — dates to the late 1940s. But recent research suggests constructive criticism can encourage additional creativity and imagination. The key, according to MIT Sloan associate professor is understanding the context of the brainstorming exercise.

In a cooperative context where the goals of group members are aligned, criticism can stimulate creativity. In a more competitive session — one where participants are encouraged to prioritize ideas, for example, or one where groups fall squarely into two camps, such as workers and managers —criticism is more likely to trigger conflict.

Before conducting a brainstorming session, leaders need to understand the dynamics of the teams coming together. If conflict is likely, organizations may opt for one session that includes free-flowing ideas coupled with criticism, then follow that up with a session for reviewing ideas. In such a setup, team members are less likely to edit their ideas — and the creative process is less likely to be undermined.

Read: Improve creative brainstorming with constructive criticism

For more info Zach Church Editorial & Digital Media Director (617) 324-0804