The Daily Pennsylvanian is a student-run nonprofit.

Please support us by disabling your ad blocker on our site.

Columnist Brett Seaton encourages Penn students to think probabilistically when searching for their future jobs. Credit: Abhiram Juvvadi

Starting your career is a lot like building a house. You start with nothing. All you’ve got is an idea, a rough blueprint, and the conviction that one day this barren piece of land will be a home. Currently, I am deciding between internships for next summer, which will likely become my career after I graduate in 2025 and the tract of land that I will build my house on … and I have no idea what I am doing.

How should I choose what career path to take? When I googled this question, I got a list of things to consider, such as “your passion and skills,” “your goals,” “your values,” and on and on. The question to me is not what my personal factors are, it is what I should optimize for. I have come to believe that we should optimize for the expected value of our careers. 

Expected value is the weighted average of all possible outcomes. When I apply this concept to a career, I mean the probability of earning an amount, multiplied by the amount earned, summed across all possible outcomes. Optimizing for expected value would mean that when choosing between a career with a 10% chance of earning $1 billion and a 90% chance of earning $1 million versus a career that has a 50% chance of earning $2 million and a 50% chance of earning $3 million, you should choose the former.

This is because 10% * $1 billion + 90% * $1 million is a little over $100 million. In the latter case, using the same formula, the expected value is $2.5 million. Expected value takes asymmetric, low-probability outcomes into account when deciding on a value, which people often miscount as 0 because of their low probability. 

Prioritizing expected value may seem like I am urging people to get as rich as possible, but I am not. Prevailing perspectives on this decision often take one of two forms. Modern Western culture would say “Whatever makes you happy,” Penn culture would say “Whatever gets you the richest,” and parents say something in between those two extremes (mine are definitely in the “happy” camp). My position is that doing something that makes you happy is one of the variables influencing expected value.

Paul Graham, the founder of YCombinator, addresses this topic in his essay, "How To Do Great Work." “The work you choose needs to have three qualities,” he writes. “It has to be something you have a natural aptitude for, that you have a deep interest in, and that offers scope to do great work.”

But why are these factors important? In Graham’s case, he is emphasizing the importance of these qualities to do “great work,” or, in the context of his career, build generational technology companies. But these qualities are true across any great work that you do and are relevant for any goal. Generalizing the principle from building technology companies: It is maximizing the expected value of your career.

Graham focuses his piece on the first two, aptitude and interest, because almost every type of work provides scope to perform exceptionally well. Doing great work and maximizing the expected value of your career is dependent on an infinite number of factors, many of which have nothing to do with you. 

If the industry you choose just happens to face macroeconomic headwinds such as technological disruption, you are more likely to have a less valuable career. If the country you are working in gets plunged into war, you similarly have a smaller capacity to create value. 

So let’s focus on the factors you can control right now when choosing between internships and career paths. How much is the company paying you and what is the starting salary for your role? This factor is important because it is a good proxy for the long-term expected value of that position. If you are being paid $100,000 per year out of college, then when you are managing people making that amount, your compensation can increase drastically as you move up a pyramid composed of higher earners at the bottom. 

But not everyone moves up that pyramid as far or as quickly as others. This is where Graham’s point is important–doing great work can help you ascend faster and farther than others. And great work, as he argued, is dependent on interest and aptitude. So pick a job that you know or think you will be good at and interested in. Interest will lead you to know more than is required, and aptitude will enable you to become a true expert at that role.

Those are the only factors that you can fully control. But there are factors outside of these that you can predict to a certain extent. Technological innovation can be predicted fairly accurately in the short term, but cannot be predicted at all in the long term. So let’s focus on the short term. 

Advances in natural language processing present an imminent threat to all knowledge work. According to a paper published earlier this year by OpenAI – financial quantitative analysts, accountants, tax preparers, web designers, writers, and mathematicians all have 100% exposure to AI. This is frightening research, especially for Penn students who disproportionately go into technical positions like investment banking and software engineering.

AI has moved 360 degrees in terms of impact on jobs since it began. Initially, most thought it would come for technical jobs first, such as math, science, and computation, all work that has objectively correct answers. Then, when ChatGPT came out, many thought that human-generated creative work would be the first to disappear because ChatGPT could write so well and create seemingly unique art. Today, I would argue that it is more likely that technical jobs will be disrupted than non-technical jobs. 

This is because ChatGPT is really bad at writing. It writes by predicting the most likely word to come next in a sentence. It turns out this can get very boring very quickly in terms of vocabulary and structure. However, as models become more multi-modal and are fed more accurate quantitative data, their technical abilities, which are already outstanding, will get better and better.

Thus, as AI obtains better data and improves over time, it will actually get better at questions with a correct answer, and worse at questions that require creativity, or less probable responses. 

That means that less technical jobs, jobs that require creativity, more humanity, and less probabilistic reasoning, are more likely to be valuable in the future. Jobs like venture capital, sales, and writing are likely to be more valuable while quantitative skills like software engineering and quantitative finance are likely to be less.

AI is just one example of a change that could impact earning potential throughout our lifetimes. Anything could happen. It’s important when making your decision to determine what you think is the most likely to happen, then make your choice with that bet in mind. Interest and aptitude are certainly some of the most important factors toward maximizing the expected value of our careers, but the industry you choose is enormously impactful too. 

Building a career that maximizes expected value by choosing a great industry, getting paid well, remaining engaged, and being happy should be everyone’s goal. Coming to your own conclusions as to the probability that these foundations will stay strong as you build your career is important when you make your decision. It is much easier to build a castle on concrete than quicksand.

BRETT SEATON is a Wharton junior studying finance, real estate, and computer science from Manhattan, Kan. His email is