Artificial intelligence shifts how and what we teach, as well as expectations and imperatives for our graduates. How do we best educate future industry leaders at a moment of astonishing technological change, a moment when they must be prepared to harness AI to create new products and solve pressing problems, and to help lead organizations through transformation?
To address that challenge, Columbia Business School convened leading faculty from CBS, NYU, Yale, MIT and other prominent institutions. Together we discussed the deep and lasting ramifications of AI and how to make sure graduates are well-equipped to address the challenges and opportunities of what comes next.
What will an entry-level management job look like in a few years? How much technical experience do students need? What is the right role for AI in the classroom?
In a world where the people in charge need to draw on more than expertise and analysis, how do we cultivate in our students real-time human judgment, deep critical thinking skills, and management skills that extend to autonomous AI agents?
Explore the transcripts, videos, and session breakdowns below for more insights into some of the most crucial, cutting-edge conversations in business education today. And stay tuned as we continue to explore these themes in the months ahead.
AI’s role in the classroom
From harnessing AI to create more productive classroom conversation to creating AI-proof assessments, artificial intelligence is rapidly disrupting the way we teach.
GenAI-driven oral assessments "I was getting back assignments that were great—very well written, touching all the important points…" “I was getting back assignments that were great — very well written, touching all the important points. But when we tried to have a class discussion: silence. The ones with perfect assignments were silent. We didn't want to get into the detection of AI, but written work does not convey what it used to.”
Using voice AI to reinvent the case study "The way that we've thought about CAiSEY is…" "The way that we've thought about CAiSEY is that it's a way not to make learning more efficient, but also to make learning have more frictions."
Socrat: AI-powered learning for the modern classroom "All of your classrooms right now have a real data opportunity cost..." "All of your classrooms right now have a real data opportunity cost: every student is having probably hundreds of conversations about your class in these isolated AI environments, and that is a real loss for the instructor. It used to be that when students didn't understand a concept, it surfaced via assignments or office hours. Now, it's harder to see where people are falling off or where you need to dig in further. By moving those conversations into a single, closed ecosystem, you get that data back."
Developing leaders who drive change
Performing analysis can no longer be the goal. In their first roles, MBAs will be interrogating models and their underlying assumptions, extracting actionable insights, and helping organizations navigate their AI adaptation journey.
Preparing our students for an agentic workforce "Every entry-level job is now a management job…" "Every entry-level job is now a management job, not necessarily a people management job, but it's about delegating work and learning how to be a good delegator. How do you scope the work? How do you assign work, and to who? How do you sort of group work streams thematically?"
Teaching agentic AI "The human plays two key roles: one is they're the guardrail…" "The human plays two key roles. One is they're the guardrail — they provide the judgment, the ethics, and accountability. But they're also the bottleneck. This is like Operations 101. They are the ones slowing down everything. You're not moving at 'agent speed' if you have a human in the loop; you're still working at human speed. The question is, what is the speed you need to be working?"
Building AI fluency "Who's going to pay for 'good enough'…" "Who's going to pay for 'good enough' when anyone can do 'good enough'? There's a risk of becoming a commodity of human labor. … Think of GenAI as a source of differentiation. You want to actually use it in a way that would make people want to pay you more for your work product."
We're all builders now
Even non-technical MBAs must move from working with builders to becoming independent creators who can build MVPs using natural language.
Teaching MBAs to build with AI from day one "Once I build that magic, that sense that we can do a lot…" "Once I build that magic, that sense that we can do a lot, I get every student to be excited to do more, and I want the course to feel like an adventure. I want every module to be like a new power they're learning that will help them be a hero, someone who can make their dream come true. So I really shape the whole course around that feeling."
Making AI more than just a chatbot "Coding gives you a structured way of thinking…" "Coding gives you a structured way of thinking. An ability to think like a computer does. And that is something I really don't want to throw out. And the problem with vibe coding is it throws the baby out with the bathwater. On the one hand, [users] don't need to know the syntax, but they also basically lose that structured way of thinking ... the idea that you need to see how to build something from scratch at least once."
Building future-proof startups with LLMs "Traditionally you're taught in entrepreneurship…" "Traditionally you're taught in entrepreneurship that you shouldn't build a product until you validate it. But with LLMs you could build five products, right? Building things is essentially free, right? Requires so little knowledge. So that's not the bottleneck. The bottleneck was how much feedback they got from potential customers. … The real driver was how much time [entrepreneurs] had with relevant people to give them feedback, and LLMs didn't help them connect to other humans."
Why MBAs need to learn AI one level deeper "Only in the tinkering do the good ideas…" "I had learned from my days as an entrepreneur that the best ideas often come from hands-on tinkering. Most initial ideas are really bad and only by tinkering do we generate good ideas from bad ideas. So, we wanted to make sure that students didn’t just call their techie friends for help but actually tinkered with ideas themselves."
Check out the full day's agenda and watch the full sessions →