BBAs Versus Bots: AI in Business Education

• While most new hires will not need to have deep coding expertise, employers expect them to be able to design effective prompts and use GenAI tools and platforms.
• Employers also want to know that recent graduates have critical thinking skills that allow them to solve complex problems without over-relying on artificial intelligence.
• At Baruch College’s Zicklin School, students prepare for the tech-enabled workplace by using AI tools in the classroom and interacting with chatbots that work with them the way a professor would.

 
As today’s business school graduates enter the job market, most will take positions that require some knowledge of artificial intelligence. “It seems like every other day, I hear about something that a company is doing to prepare its workforce for AI,” notes Michael Kalish, senior associate director of recruiting and internships at the Starr Career Development Center at the Zicklin School of Business at Baruch College, part of the City University of New York (CUNY).

While many students worry that AI will take their jobs, Kalish says the opposite is true. “For example, we just held a career event with a Bank of America recruiter, and she told the students that AI will make their jobs easier—it will handle the grunt work and they’ll have higher-level responsibilities.”

However, the pervasive use of technology means that students need to “bring their A-game,” Kalish adds. “They have to be prepared for AI and be knowledgeable about the skill sets that will now come in handy.”

At Baruch, which is located in the global business hub of midtown Manhattan, roughly 12,000 undergraduates are working toward BBA degrees. Knowing the importance of AI literacy to their graduates, school leaders encourage—but don’t require—faculty to incorporate AI into their courses. Says Kannan Mohan, associate dean for academic affairs and innovation, “We acknowledge that a one-size-fits-all approach will not work for AI infusion.”

For faculty wishing to improve their AI skills, Zicklin provides professional development opportunities, such as playtesting sessions and lists of academic readings collected by the college’s Center for Teaching and Learning. The school also promotes community discussions by faculty who have been assigned Microsoft Copilot 365 licenses and are working on custom AI agents to support students in their courses. While Microsoft Copilot is the CUNY-approved AI tool, faculty also may use other AI tools as long as they do not provide the chatbots with any institutional data or student information.

The Zicklin School recently won a grant from CUNY to support a faculty group now analyzing BBA and MBA curricula to identify and address gaps in AI education. Through these combined efforts, the school is focusing on three skills employers say are more essential than ever for new hires: technological literacy, problem-solving abilities, and critical thinking skills.

1. Basic Technological Fluency

In today’s AI-informed workplace, the minimum requirement for employees is the ability to write an effective prompt. Many students already are independently refining their prompt-engineering skills, thanks to the widespread availability of free versions of Anthropic’s Claude and OpenAI’s ChatGPT. But it’s important for students to remember that they aren’t really having a conversation with generative AI (GenAI), says Chul Kim, an assistant professor of marketing. They should be directing it to uncover specific information.

“A recent paper showed that giving all the details in the first prompt yields better results,” Kim explains. “The algorithm relies heavily on the first prompt for context.” For example, a simple prompt such as “write an email following up with a client” will elicit a less useful response than “write a 100-word email using these keywords to summarize the meeting I just had with Client X.”

Many organizations want new hires who are skilled at “vibe coding”—prompting AI to write the code for them—so it’s beneficial if students can navigate code-writing platforms and GenAI options.

Most organizations will not be looking for business school graduates who can do deep coding, but many will want new hires who are skilled at “vibe coding”—that is, prompting AI to write the code for them. Therefore, it’s beneficial if students understand how to navigate code-writing platforms such as Google Colab and GenAI options such as Google Gemini.

At Zicklin, students in an inventory analytics class practice using these tools to give verbal instructions for writing code that will execute complex operational analysis in minutes.

“The ability to model uncertainty is crucial for decision-making and is done by building simulations that used to require high-level math and coding skills,” explains Tolga Aydinliyim, an associate professor of operations and decision analytics. “Students can now ask Gemini, ‘I’d like to simulate 100 days of random demands and compare the performance of different quantities of inventory ordered. What will my profit be if I order 50 pairs of jeans versus 100 pairs?’” He adds that, within the education field, this sort of analysis was only accessible to PhD students until about three years ago.

Michael Weylandt, an assistant professor of statistics, has his students use R programming language and GitHub, a web-based coding storage platform, to “skip the building blocks” and “get to the good stuff faster.” He likens these tools to the food processor a chef might use: “You can get to feeding people faster.”

Weylandt notes that many of today’s company leaders “know they need to invest in AI but don’t always know which way to go. They’re happy to hire students who can bring that direction to them by using AI on the job to perform better.”

2. Problem-Solving Abilities

Even so, employers want to be certain their workers can evaluate AI outputs and not just copy and paste “slop” from ChatGPT, says Kalish. For that reason, hiring organizations are increasingly conducting “case interviews” where job candidates are given complex business scenarios and asked to propose solutions. Zicklin professors prepare students for such interviews through project-based assignments that they must complete with the use of AI.

In one such project, Weylandt’s statistics students role-play a consultant hired by the New York City mayor’s office to find budget savings. Weylandt asks students to imagine that New York has a rule that no government employee can earn more than the mayor. This requires them to search publicly available government data on workers’ salaries, a data set so big that it requires them to use AI tools to solve the problem.

In the Zicklin School’s Executive MBA in Healthcare Administration program, Aydinliyim has students use statistical analysis to identify diabetes patients at higher risk of readmission because they don’t remember to check their blood sugar and take their medications.

To prepare students for job interviews where candidates must propose solutions to complex business scenarios, professors design assignments that students must complete with the use of AI.

“This is a well-known problem among doctors who treat diabetes,” Aydinliyim explains. “If you can do wellness checks where you call to remind them to do those things, it’s quite helpful at avoiding readmission, but it’s also costly. I have students run simulations where they compare no reminder calls to reminder calls only for high-risk patients to reminder calls for everyone.”

Students use that knowledge to determine which routine is most effective. They can highlight such activities on the portfolios they submit when they’re applying for jobs.

3. Critical Thinking Skills

Because using AI can encourage laziness, professors take steps to ensure students don’t become accustomed to outsourcing their thinking to technology. For instance, Kim requires more in-class discussions and group projects in his marketing courses. “I don’t want students to have fancy ideas or strategies—just their original thought,” he says.

Yafit Lev-Aretz, associate professor of law, emphasizes that “uncritical reliance on AI has real professional consequences.” She explains to her students that GenAI models “are at their core highly sophisticated iterations of autocomplete. They can generate plausible responses, but that same mechanism is also what leads to failures, including hallucinations.”

In her business law classes, Lev-Aretz focuses on overarching themes so students develop a more conceptual understanding of the law. In her classes, students also learn how to evaluate AI outputs and integrate those outputs into decision-making. She structures her classes as interactive discussions, emphasizing real-time engagement and encouraging dialogue to reinforce human interaction. To that end, she instructs students to address her directly “as if it’s just the two of us in a room.”

Lev-Aretz illustrates the direct consequences of relying too heavily on AI in two ways. First, she writes exam questions that are so broad AI systems are likely to answer them by citing legal doctrines or terminologies that were not part of course assignments, and she deducts five points every time a student mentions material that wasn’t covered in the class. Second, she provides real-world examples in which lawyers have been sanctioned by courts because they relied on AI-generated briefs that cited nonexistent cases.

Joshua Moritz, an adjunct lecturer of marketing, also takes steps to keep students from outsourcing their thinking to a bot. He has devised a comprehensive framework students must follow whether they are using AI to create advertisements, devise personas, or write research questionnaires. During such assignments, students must complete these tasks:

• Create detailed mission statements, project objectives, and lists of goals.
• Write ten bullet points relevant to the project—for example, ten facts about the persona they want to create.
• List ten inputs for training the AI. These could be book chapters, detailed PowerPoints, videos, scholarly articles, or similar materials.
• Write ten initial prompts and ten secondary prompts.
• List ten ways to verify that the output is correct. For instance, students might solve problems in Excel, compare results from statistical software packages such as SPSS or MATLAB, or check the persona against existing behavioral and demographic information.

Moritz also has students record their computer screens while they’re having discussions with GenAI; each recording lasts about ten minutes. Because students complete eight projects over the semester, they are “intuitively” using the framework by the end, he says. Through this approach, Moritz is able to see that students are checking AI-generated work all the way from input through output.

Teaching With Technology

While students should learn the risks of over-relying on AI, they should also be able to take advantage of its immense potential—including its role as a teacher.

With that goal in mind, information systems faculty at Zicklin are developing an AI coding partner chatbot that double-checks that students understand the underlying logic of coding. The chatbot is modeled on the industry practice of pair programming, in which two developers work on the same code simultaneously.

When a chatbot functions as a teacher, it can use the Socratic method to offer hints, fragments of code, and detailed questions instead of full solutions, leading students to find the answers themselves.

“With generative AI, it’s very easy for students to take a coding prompt, paste it into ChatGPT, and ask for the answer,” says Sonyl Nagale, a software engineer who is also an adjunct lecturer of information systems. “They don’t break it down or think about why they’re solving it, what the need for it is, or why it’s being presented. So, we came up with the idea of creating a generative AI plug-in that works with students like a professor would.”

Thus, the chatbot employs the Socratic method to offer hints, fragments of code, and detailed questions instead of full solutions, leading students to find the answers themselves. “This is how we would guide them one-on-one, if we had infinite time to sit down with each student,” Nagale explains.

Nagale employs similar techniques in his classroom—for example, by routinely asking students to explain why they solved a problem the way they did. “Just because something works doesn’t mean it’s the best answer,” he notes. “Asking why hopefully leads students to reflect and iterate on their solutions.”

Similarly, he breaks large projects into smaller components that students combine to form the final solution. However, they must reevaluate the overall problem so that the result becomes more than the sum of its parts. He says, “When students synthesize a greater whole from fragments of a solution, they practice piecing together incomplete data—like in the real world—and taking the next intuitive leap toward understanding.”

Time to Embrace AI

The bottom line is that AI is here to stay, says Kalish from the Starr Center. This means that both schools and students need to understand how it’s deployed in the workplace.

“Students should embrace ethical AI with open arms because it really can make their jobs easier,” says Kalish. “It’s almost as if everyone got a promotion because the grunt work is taken out of the mix. It is important to note that students need to continue using their own critical thinking, judgment, and interpersonal skills, if they truly want to get promotions. Students will still need to engage their brains and humanity, as AI is certainly not the solution to everything.”