How Stanford GSB's Student-Led AI@GSB Initiative Came Together and What It Changed for Students

The panel began with applause from the audience as the session got under way with a welcome from Scott Brady.

Scott Brady said he has been back at the GSB for well over a decade. He teaches several courses, including the capstone entrepreneurship class and a research class, and most recently serves as faculty advisor for AI@GSB along with Brett Jordan. He said the context and genesis of the initiative were worth sharing.

Brady recalled that about 18 months earlier, Sarah, who was preparing to be seated as the new dean, had contacted him after he and Brett Jordan taught a new research class heavily based on applied AI. He said that when he received an email from her asking to talk, he worried he had done something wrong, but he found that the dean was excited. She saw AI as a formative moment and an opportunity for the GSB to take a leadership position, and she wanted the faculty to experiment. She then identified four rising MBA second-years who became AI scholars. The initiative was a dean's initiative and was faculty sponsored, but it was led by an extraordinary group of students. Three of the four were on the panel.

Jenni Steiger, who was among the students Sarah spoke with during her listening tour, described why business school is a major decision: students take two years out of the workforce and hope the experience will make them better off in the long term. When they arrived at the GSB, ChatGPT was just getting very good at knowledge work in a way it had not been before, and the class had a moment of realizing that work would be different when they returned.

Steiger said the students she and Sarah heard from were anxious about jobs and whether they would have the skills needed to return to the workforce. Constant news and social media coverage of new AI releases made them feel insecure about whether they knew what they were supposed to know and were keeping pace with their peers. Many were learning AI on their own in their dorm rooms and were looking for resources, an approach and help.

She said students were really, really hungry for something from Stanford.

Steiger said that when the team spoke with classmates, the concerns fell into three groups. First, students did not know how to use the tools, or knew how but were unsure whether they were using them properly or efficiently. Second, they were not sure how AI would change their role and the industry they wanted to enter. Third, they wanted to know how to make sense of what was happening across the AI landscape.

To address these, AI@GSB aimed to teach students how to use the tools, how to set up their environments, and how to become smart, efficient users. Steiger said the goal was for students to be able to tell interviewers what they had built, show artifacts, and demonstrate the capability in job interviews.

As part of the program, the team set out how AI looked six months earlier, how it looked at the time, and what was still changing. Steiger said the purpose was that students would have an informed view of what they were going back into instead of going blind.

The third element of the program was bringing in thought leaders. Steiger said the team invited someone like Derek Thompson to talk about whether AI is really a bubble. She said students going into investing, or anyone who would hold a senior leadership position, need a grasp of what is happening and how it will affect their industry, their role and society at large. She said the team tried to deliver those three elements through AI@GSB.

Bean noted that one of the benefits of the GSB is that it is very experiential. She said many students told the team they wanted to do things themselves, and that was the first thing the program built out. This took shape as a series of sessions in which the team taught specific tools and how to think intelligently about whether something should be built with software. The sessions also covered the implications of having a probabilistic model under the hood of what students build, and made sure students knew where to start, along with tools to pursue their own creative ideas.

Bean described three pillars. The first was making sure students knew how to use the tools themselves and had to put their hands on the keyboard. The second was bringing in practitioners who could talk about the mistakes they made and how they fixed them, since hearing about a problem is very different from making the mistake and having to change it so it works. The third was addressing how to use the tools strategically, ethically and inclusively, and the broader implications of these technologies, which led to bringing in leaders thinking at a further horizon.

Scott Brady called students teaching classmates one of the most distinctive choices in the program. Bean said the sessions often began with a demonstration, for example by Freddie or herself, of how to use a tool. The team aimed to teach students to be tool-agnostic, to think beyond the particular demonstration. This deputized students to transmit their knowledge to classmates.

Bean said this had two benefits. The first was that the best way to learn is to teach, so students had to think more deeply about what they were saying and grapple critically with the concepts they had encountered. The second was that it allowed students to move into an identity where they could build things themselves, applying an example from a session to their own subject-matter expertise.

Brady asked Alfredo Méndez what difference AI@GSB made to students' careers and for a specific example. Méndez said the program brought about an identity shift. Before, the perception of technology, particularly among those without a technical background, was 'I don't belong here.' He said that place seemed to be for engineers and computer scientists. As people began to see that everybody can build now, and that their opinion matters, they felt welcome.

Méndez said that in the first week of Admit Weekend, some classmates described traditional backgrounds and said they would love to go into tech but did not think they had what it takes. By the end, they believed they had what it takes, and he said seeing that shift in classmates was very powerful.

Méndez said that in a class on entrepreneurship and venture capital taught by Eric Schmidt, Schmidt said that while he was head of Google, there was a golden ratio of one product manager for every ten engineers. That ratio has now shifted to two product managers for one engineer. Méndez asked who the bottleneck is now.

Méndez said many students who had been told they were not welcome in tech came to see themselves as the missing piece. This affected the careers, industries and even the companies students aspired to, and he said it moved them from 'I'm not welcome in here' to 'I am the missing piece.'

Scott Brady asked how the bar has moved for the best jobs students want. Méndez said competition is intense and the bar has been moving fast. He described three moments. In summer 2024, when the MBA started, companies asked, 'Tell me what you know about AI?' Generally, a good grasp of technologies such as transformers, and how they work, including next-token prediction, was enough, especially for students without a technical background.

During the summer internship, the question became how candidates were using AI, and to tell and teach the company. Méndez said he and many classmates had internships outside the Bay Area, including internationally; he was doing one in Tokyo. He said that coming from a place like Stanford and representing the GSB meant the expectation was that you would be the expert even without that background. Companies relied on MBA interns to define their AI transformation strategy. In his words, literacy was assumed and fluency was expected.

By the spring of this year, when Méndez was interviewing for a fund led by a well-known AI researcher and Stanford professor, the process went beyond the standard tasks. He said that in his final round the usual assignments were included, such as writing an investment memo, analyzing trade-offs and drafting a product requirements document, which is very standard for a product manager. The instructions went further, however, asking candidates to build a demo, ship it, get customers, integrate their feedback and explain what they learned, all within about 24 hours. Méndez said the goalposts had moved significantly, and he summed up the shift as going from 'tell us what you know' to 'how do you use it' to 'show me what you've done' in a constrained amount of time. Brady noted that Freddie got the job.

Brady said the question of what is most valuable when building becomes easy might be the most important one. Méndez said many people are still asking it, and the answer changes. He summed up three things. The first is conviction: having an opinionated but well-informed vision of the world and the future, and being able to stand by it. The second is curiosity, since everyone can be technical but not everyone is curious, and much of this is learned by doing.

The third is the human component. In Méndez's experience in the first months after graduation, what he learned at the GSB, including interpersonal dynamics and understanding motivators, helps decide what to build and what not to build, and helps one stay a step or two ahead of the implications. He said this is where leadership, management and interpersonal training seem to become more valuable than ever.

Brady said he teaches many executive education programs on AI and typically gets two questions. The first is about how fast things are moving and what to do first, and he said the answer is typically not about technology but about talent. The second is what to tell one's kids.

Brady described a talk by Derek Thompson, a writer for The Atlantic, which he said was one of the most impactful events he attended because it challenged his perspective. Brady said Thompson's writing is generally apocalyptic. After a Q&A question from a student asking what she should do, Thompson said she should treat AI like her health. He asked how many attendees had brushed their teeth that morning, and why, since having teeth 20 years from now was important, and then asked who watches what they eat or works out, and everyone raised a hand. Thompson said AI is not an event like the things that got people into the room, and one should build the habit and attention to it over time.

Asked how the GSB benchmarks its progress against peer schools, Steiger said the GSB was recently ranked the top MBA program for AI learning, and the audience applauded. She said students have a lot of activation energy and are agentic themselves. The new cohort of AI at GSB scholars received the same number of applications that View from the Top received in its first year. Every session the team ran was a packed room, and organizers had to block out larger venues.

Steiger said the dean backs the program, and a new set of scholars is working with Brady on the programming, with Brett Jordan providing mentorship. She said she believes the infrastructure is in place to keep the program going and improving.

Steiger said she would not claim the team had solved students' anxiety about AI. She said everyone is anxious about what is happening in the world, and that it is hard to go to business school, especially now, and to take years off. She said it is an amazing time to be at Stanford because of the GSB's work on AI, as well as the research happening across the street, where there are, in her words, the best researchers in the world.

Responding to an audience question about what the GSB's faculty and teachers could do differently, Bean said AI@GSB is a fully student-led initiative. She said much of the content it produces is not evergreen. For faculty, one challenge is that they build an excellent curriculum but changing it every time they teach a class is a lot of work. Bean said there is a benefit to a nimble organization like AI@GSB that leverages each incoming cohort's proximity to the technologies, since each subsequent class is becoming more technical.

Bean said faculty have been dynamic in a couple of ways. One is using a flipped classroom, making use of synchronous time together to take advantage of the many smart people in the room with different experiences, moving away from an asymmetric lecture style.

Bean said changes are also happening on the evaluation side. A take-home exam does not mean the same thing if students are just competing on who has the better Claude subscription, and that starts to lose its salience as a way to reflect how much someone has understood. Faculty have responded with a number of creative mechanisms. On the less creative side is going back to blue books, which Bean said she personally favors, because students take the test in person and write it by hand. Bean also described a move toward oral presentations to make sure students are responsible for the material, which goes back to the idea that the best way to learn is to teach.

Brady said he teaches Formation of New Ventures, a course Irv started close to 40 years ago that explores the arc of a startup and was the quintessential case-method class. He said the case method has evolved, and that at some schools it is incredibly rigid. Over the last two years, Brady said, he has had perhaps the most extraordinary teaching experience of his career.

Students are told to prepare using whatever tools they need before class. Instead of using the historical case method, with five questions the class pivots around, the course is taught extemporaneously. Students can do anything they need before class, but when they walk in there are no computers. It is just students and faculty in an unscripted conversation, with cold calling. Brady said he does not know where the discussion will go, and neither do the students. He said the approach leverages these technologies to make students show up the way they would in the real world. He said the quality of the environment and of the learning, as well as his own interest in teaching, has risen tremendously, and he believes the post-AI world offers an opportunity to do something extraordinary.