AI in the classroom: Teaching MBA students to harness AI for business success

“Classified” is an occasional series spotlighting some of the more powerful lessons being taught in classrooms around Haas.

It’s a Sunday morning at Berkeley Haas, and Lecturer Roopesh Varier is preparing MBA students in his weekend class to use AI tools to analyze customer satisfaction metrics at fictitious startup Bling Data. 

Varier, who teaches “Harnessing AI for Business Success,” gives the group a large data set of customer feedback for Bling, which helps consumers track spending. The class discusses Bling’s poor Net Promoter Score (NPS), used to measure customer loyalty and satisfaction, before Varier suggests how to write detailed AI prompts to get the information they need. Then the case work starts.

A veteran data leader, Varier is bringing AI into the classroom in real-time through sharing his own workplace experiences and inviting speakers who also work in the industry to share how they’re using AI. In the first week, the class addressed AI basics, cutting through jargon to help students understand how AI can be used in marketing, finance, and operations. He then moved on to show how a smart AI strategy can build a competitive advantage through measurable business outcomes.

In today’s class, the students consider how AI could help Bling understand patterns in user feedback. One student says that the company’s support staff, while empathic, are not solving the real problem: technical issues with the Bling customer support model. 

“The story, I think, is that there’s a relatively low loyalty to this company, specifically because of the low NPS score,” Varier, MBA 09, and a member of the Haas Professional Faculty responded. “But also, another company could just as easily offer these same features like ‘spend alerts’ and customers would immediately switch. … Let’s say you are a competitor. What would you do if you had this information? Where would you focus?”  

 “Lower pricing,” the students reply.

Varier agrees but then takes the conversation deeper, pulling in the story of how Facebook took on MySpace by clobbering them technically.

“Some of you may not even have heard of Myspace because Facebook made it so,” he said. “As an example—MySpace was not reliable. It was crashing every single day. Users were frustrated. Facebook focused on making sure its system was reliable. Facebook listened to the customers’ pain points. … And that made all the difference.” 

The key takeaways from the exercise? AI tools can be powerful allies in making product roadmap decisions, but always double-check the math when you use AI tools, which can be unreliable. Also, customer feedback is key to helping you understand patterns. 

Catching up on trends

The case work appealed to Hayden Xu, MBA 25, who works in consulting and says he’s an early adopter of ChatGPT. “I took this course to catch up on all of the trends,” he said. “Most of what I am learning here is new, including the powerful data analytics I am using with the AI assignments.”

An industrial engineer before she came to Haas, Simay Gokalp Altin, MBA 26, said she is hoping to work as a product manager after graduation and needs to have a deeper understanding of AI.

Moving on to a second case, the class considered the strategy of a healthcare startup founded by a doctor who lost her daughter to cancer. Students were asked to map the customer journey for the startup and identify pain points. Then they had to decide whether generative AI could be used to replace customer service representatives. The case involved both ethical and organizational considerations when implementing an AI strategy.

The class broke into groups. Half of the groups provided reasons for replacing the customer service representatives at the company, which provides support for families to help with cancer treatment. The other half had to argue against using Voice AI, a type of artificial intelligence that can interact with humans via the phone in real time, after learning from existing data.

Fernando Gonzalez Demaria, MBA 25, said that using AI for customer service doesn’t align with the founder’s desire for a human touch. “There’s also a high risk of mistakes, and then who is going to be liable if the AI makes a bad decision?” he said. “We think that going full AI contradicts the founder’s vision of holistic care.”

The pro generative AI team argued that replacing the reps could save the organization a lot of money and improve accuracy when recalling past customer data. The company could also retrain customer service reps to do other jobs that might be less stressful for them, the team said.

A bright future

After this exercise, students formed groups to present how they would incorporate AI marketing and organizational strategies into an industry of their choice.

Dhruba Borthakur, a technical leader of infrastructure at OpenAI, also joined the class to talk about his experience in entrepreneurship, the latest in the world of AI, and his thoughts on where AI is going.

Looking at the future for AI, Varier argues that the technology’s biggest threat isn’t that AI will take over the world; it’s human brain atrophy.

“If you don’t need to think about it, you’re going to let it go,” he said. “Suddenly now you don’t have to write essays. You don’t have to do a lot of thinking. Brain atrophy is one of the biggest issues. Creativity loss is one of the biggest issues. That is the risk for humans from AI.”

Done the right way, Varier sees a bright future where humans and AI coexist. “AI will help humans be more efficient in all phases of life,” he said. “But what we do and how we do things in the future will be redefined.”

But when asked what tool is best to use for AI analysis, including Claude and ChatGPT, Roopesh’s reply is simple. “The answer is you. … Nothing can be more valuable than the human perspective,” he said.