Professor Drake

Director, Center for AI in Business

The Best Business Majors When AI Takes Over

by ·

With two kids in college and another a junior in high school, talk around our house often revolves around careers, majors, and classes. And with my obsession with all things AI, one topic that often comes up (by me) is the future of those jobs due to AI automation. As Director of the Center for AI in Business, I’ve done a bit of thinking on how business careers will change due to AI. Threats of job lose due to AI are largely over-blown. Jobs will not go away. However, they will change. These changes directly impact what college students today should major in. Three trends shape my thoughts on the best business majors in the next 5 years. 

Automation of Knowledge Work

The first trend addresses what AI models can do. One measure of their ability is GDPval, which calculates the percentage of expert level knowledge work tasks that the model can do as well or better than actual experts in that field. It’s not a perfect metric, but does a better job than other metrics in capturing potential economic impact of the tools.
 
Of note, OpenAI’s latest model, GPT-5.2, achieved 71% on the GDPval, the highest yet achieved. This rating suggests it can beat or tie human professionals on ~71% of evaluated knowledge work tasks. Those tasks could potentially be automated today.
 
Not to be outdone, other AI platforms are rolling out products and features that also significantly impact knowledge work. Anthropic’s products, such as Claude Cowork and Claude Code, further push the boundaries of automated work. While GDPval can’t assess the efficacy of these products (it only measures the underlying model), antidotal stories suggest it pushes the boundaries of automation.  
 
It’s important to be clear that just because these tools exist, doesn’t mean all tasks of knowledge workers will be automated immediately. There’s always a lag time between capability development and organizational implementation. Furthermore, automation will likely start with the simplest tasks, to provide proof of concept that AI can be implemented effectively.  These tasks are usually completed by entry level workers. And who generally does entry level work? Recent graduates. 
 
Since I’m no economist and because ChatGPT can automate that knowledge work for me, I asked ChatGPT to assess which undergraduate majors would be at the greatest risk due to automation in the next 3 years. Here’s what it said:
 
Undergraduate Major / Field
Risk Level
Key AI-Impacted Tasks (Examples)
Computer Science & Software Eng.
High
Routine coding, debugging, documentation; AI generates boilerplate code
Accounting & Finance 
High
Financial reporting, modeling, auditing; AI can prepare financial statements and analyses
Business Administration / Management
High
Business reports, project planning, data analysis; AI drafts proposals and schedules
Marketing & Sales
High
Marketing copy, data-driven market analysis; AI generates ads, customer emails
Law & Legal Studies (Pre-Law)
High
Legal research, drafting contracts, document review; AI automates routine memos and discovery
Journalism & Communication (Media, English)
High
Article writing, content summarization; AI produces news briefs and social media content
Data Science & Statistics
Moderate
Data cleaning, visualization, initial analysis; AI accelerates analytics but not strategic insights
Economics (Policy/Research)
Moderate
Economic analysis, report writing; AI assists in data crunching and forecasting
Engineering (Non-Software)
Moderate
Routine design reports, simulation setup; AI offers CAD and analysis assistance
Human Resources / Office Support
Moderate
Resume screening, scheduling, memo drafting; AI automates admin tasks
Education / Training
Moderate
Lesson planning, grading, content creation; AI generates materials and quizzes
Nursing / Healthcare Administration
Moderate
Documentation, coding, scheduling; AI handles admin work but not patient care
Arts & Humanities (Fine Arts, Social Work, etc.)
Low
Ideation, research assistance; AI provides creative inputs or summarization, but core tasks stay human
Trades & Vocational (Tech & Labor)
Low
Manuals, technical documentation; AI may draft manuals but skilled labor tasks are hands-on
 
Full report: https://chatgpt.com/s/dr_6940053aeff08191b5b9c798fad610ae
 
Obviously many majors are missing, but it provides the general trend. Many traditional business majors are at risk. I suspect over the next couple years, job reports will start showing a softening of hires of entry level business employees. Once the job numbers come out, enrollment will start declining in these majors (3-5 years). 
 
What’s not mentioned in the ChatGPT report is the need for people to do that automation. While existing software engineers and computer scientists will back fill some of that work, their skill set doesn’t align well with this need. Software engineers excel at coding. Excel at defining logical steps to accomplish tasks. Excel at deductive reasoning.
 
Automation of knowledge work requires a different skill set. It requires the ability understand business processes, work with uncertainty, and analyze ambiguous tasks in order to design systems that ensure the automation correctly addresses the business need. This skill set is exactly what MIS professionals excel at. 
 
As vibe coding becomes more mainstream, MIS professionals will gain increased capabilities of building applications themselves, rather than waiting for software engineers. While parts of the MIS skill set will be automated, there should be sufficient demand in the overall major to offset those automated parts. 

Agentic AI in organizations

While most people are aware of the AI chatbots, fewer are aware of AI agents and their capabilities. If a chatbot can double a single employee’s productivity, a series of AI agents properly designed and working together could 10x the productivity of the entire organization. Dare I say 100x? Will we have a billion dollar company with 10 or less employee in the next 5 years? YES! 

But that’s getting off track. What will agentic AI mean for employees?  Each employee will need to master managing multiple agents. Certainly there will be orchestration between agents, such that you talk to one agent who then knows what other agents to contact to get work done. But once an agent is assigned a task, it make take a while to complete. In the mean time, what would an employee do?

Consider this scenario:

Say I need to create a marketing campaign. I might ask a marketing agent to develop the campaign framework, which coordinates with a legal agent, then with an advertising agent, and lastly with a product agent. With that information and occasional feedback from me, it starts building the campaign based on previous successful campaigns. 

As I wait for the marketing agent to complete it’s task, I request a scheduling agent for meetings with five major customers about a new product. The scheduling agent coordinates with the customers’ agents to find the ideal time for both parties and add it to our respective calendars. Details of the new product are shared from an engineering agent and added to the agenda – along with an economic forecast from an economic agent, and a timeline for production from a supply chain agent. 

Now that I have two agents workflows processing in the background, I spin up an app building agent to create a new agent to track competitor products. The app building agent must get approval from IT security agent, IT infrastructure agent, and IT database agent. As I’m working on this agent, I get five notifications from the marketing agent, two of which needs my feedback, and one from the scheduling agent. 

All three tasks are done before lunch. 

Skills necessary in this environment have less to do with specific specializations and more to do with general business knowledge and the ability to manage. The difference, however, from traditional management is that AI agents don’t have feelings. In some ways that’s empowering, in that less effort needs to be placed in managing feelings and more energy put towards fully leveraging the agent abilities. On the other hand, there’s a danger in too much automation unless ethical safeguards are baked in to the system in deep and meaningful ways. Without the normal checks we place on ourselves when interacting directly with people, the last major check on behavior is our moral foundation. Without such moral checks, narcissists will thrive. 

As strange as it might seem, hiring majors in the humanities might provide some of this moral check. I say this with a major caveat – it depends on the moral foundation taught in these programs. Not all moral foundations are equal. But that’s a can of worms I don’t want to get into for this post. At minimum, business majors still need that humanities foundation in their curriculum. 

Besides the moral foundation, employees need an ability to rapidly adapt to changing environments. They need an innovation mindset and skillset. Many tech companies already recruit based on that mindset, often hiring developers that have started their own business because it shows the developers willingness to take chances by looking into the abyss of nothingness to create something new. 

However, not all entrepreneurs make good employees as some keep grasping at the new instead of shipping the current product. Few majors focus on innovation mindset. However, we do see it some in the arts and occasionally in computer science and MIS. And of course entrepreneurship programs. To the extent majors can transition to an innovation mindset, the better they will prepare students and likely flourish in this environment. 

The AI interface

Jensen Huang, the CEO of NVIDIA, in his keynote speech at CES 2026 stated we are witnessing a change in fundamental structure of software. Software longer consists of apps built with AI, but instead apps built on AI. Within a few years, the new form factor for interacting with computers will consist primarily of an AI. Similar to how windows interface replaced command line interface, AI interface will replace windows.

Coupled with agentic AI above, technologists expect software-as-a-service to die. Their reasoning? Eventually, organizations will create agents that connect to all the tools and databases within an organization. When an employee needs something, they just ask the corporate AI. The AI will access the data, perform the actions, reference the tools, call other agents, and accomplish the task. No other software will be necessary. Even the CEO of Microsoft has admitted the danger to their SaaS business model. 

The key word there is “Eventually”. Changing all this legacy software to the new AI interface encompasses a massive upgrade that will take years to accomplish. Slow moving incumbent companies will be vulnerable. While this truism has always been true, the difference this time is the scope of impacted companies and industries. Small teams of motivated entrepreneurs can leverage AI from the start and rapidly grow, pivot, innovate, and crush the competition. With the knowledge of the world built into the AI, entrepreneurs have the advice of industry experts at their finger tips.

Furthermore, less time will be necessary for administration, as AI automates it, and more time on customer and product development. This gives new businesses a huge advantage. The major advantage existing organizations have is their proprietary data saved in their databases. Will that be enough to fend off the onslaught of new businesses? Only time will tell. 

Again, a business major such as entrepreneurship creates a solid background to succeed in the AI interface world. 

Best Business Majors 

With AI automation, changes are coming. Massive global changes. Traditional majors may struggle to find fluency in this world. Yet, two business majors should withstand and even flourish, MIS and Entrepreneurship. These changes may take a few years, but freshmen would be wise to consider these changes and their place in it. 

Tags:

You may also like...