INFORMS Open Forum

  • 1.  Will AI eliminate OR jobs

    Posted 20 days ago

    I'm teaching an intro to OR course to college freshmen.  Several students, who love the material and are showing real talent for it, are concerned about future job prospects, and are planning to major in an engineering field they think is more robust to AI-based job elimination.  They are choosing electrical and computer engineering and mechanical engineering.   I'm curious as to what's happing at other universities, and whether OR professionals share the students' view on future job prospects.

    Let's discuss.



    ------------------------------
    Brenda Dietrich
    Arthur and Helen Geoffrion Professor of Practice
    Cornell University
    Ithaca NY
    ------------------------------


  • 2.  RE: Will AI eliminate OR jobs

    Posted 18 days ago

    OR is a great field, but there is less of a tangible connection to job opportunities.  It can range from almost pure math to the embedded OR we see in operations management.  In the old days OR was deprecated as 'efficiency experts' who went around speeding up people's factory jobs to intolerable levels.  Some of that pastiche remains, but now usually if you tell someone your field is OR they just blank out--- though if you call it analytics or AI they get the feeling.

    My university (Princeton) solved this dilemma by making a separate Operations Research and Financial Engineering department.  This resonated because at an Ivy like Princeton, something like 50% of graduates go into finance--- ridiculous, but it's where the money is (pun intended).  It does give the field a very positive cachet, though.  I've often wondered what counts as financial engineering.  The graduates, in my opinion, are not trained in any way like the engineers of my day in engineering school. There was always lots of experimentation, building, hands-on work, and deliberate consideration of costs and physical (and today, environmental) constraints.  Maybe they do that in Princeton's ORFE with practical work, labs, internships, practical experience.  Warren Powell, the prime mover there,  certainly believed in logistics as a key application.

    So if students think they want to be engineers they should think about how much they are into that practical 'maker' discipline those engineering fields train for. Much of that, like most conventional industrial engineering or operations management jobs, involves applying that mentality to mundane problems. Much will certainly be replaced by AI agents.  That's just as true as what agents, salesmen, brokers, even doctors (or nurse practitioners today) do. The niche for new graduates is their deep general knowledge of the field's intellectual reach and limits, and the judgment they can bring to help make AI agents work in the service of man.  

    I'm enough of a liberal arts advocate that I think you can get that through any of our major disciplines, including some of the liberal arts, though a good slug of technical training in STEM-related disciplines, especially math, and certainly OR, is certainly a benefit in finding jobs. The best jobs will go to the versatile. A Cornell degree, in whatever, will be a major asset.



    ------------------------------
    Bruce Hartman
    Professor
    University of St. Francis
    Tucson, AZ United States
    bruce@ahartman.net
    website: https://sites.google.com/ahartman.net/drbrucehartman/Home
    blog:http://supplychainandlogistics.org
    ------------------------------



  • 3.  RE: Will AI eliminate OR jobs

    Posted 17 days ago

    I tend to agree that AI is more likely to change the nature of OR work than eliminate it.

    I read this recent article by Tom Davenport and Miguel Paredes in Harvard Data Science Review. It makes a useful point - we should be cautious about predicting which jobs AI will "take." Many forecasts start by estimating which tasks can be automated, but task automation does not translate cleanly into job elimination.

    That distinction matters for OR. AI may automate parts of the job (think of model coding, scenario generation, documentation, data preparation, and first-pass analysis). But the more valuable parts of applied OR often involve judgment - defining the decision, choosing the objective, understanding which constraints are real, evaluating tradeoffs, and helping decision-makers act on the results.

    So I would not advise students to avoid OR because of AI. I would advise them to learn OR deeply, learn AI tools, and focus on becoming the professional who can connect models to real decisions and operational systems.

    The safest career path is probably not to search for an "AI-proof" discipline. It is to build technical depth, practical judgment, and the ability to work across models, systems, organizations and decisions. OR is a very strong foundation for that.



    ------------------------------
    Jiaxi Zhu
    Head of Analytics
    Google, Google Customer Solutions
    Mountain View CA
    ------------------------------



  • 4.  RE: Will AI eliminate OR jobs

    Posted 15 days ago

    I agree with Dr. Jiaxi Zhu. The safest career path is to demonstrate one's value to the prospective employer, AI or no AI. What tangible can one bring to the table? What employer's problem can one solve, with AI or no AI help? What's the vision for moving the organization forward? What's one's ability to communicate and interact with others? Having a major in the OR field is an indispensable foundation. However, the critical factors here are a deep non-formal understanding of the OR field, the ability to translate OR methodology to real-life problems, and the ability to ask the right questions. 



    ------------------------------
    [Alexander] [Kolker]
    [
    ------------------------------



  • 5.  RE: Will AI eliminate OR jobs

    Posted 18 days ago

    My view is the opposite: AI is more likely to expand the field. Spreadsheets did not eliminate analysts. Low-code/no-code ML did not eliminate data scientists. Those, and other disruptors, expanded the market for their work, respectively. AI will take over more of the routine work in OR, such as data preparation, model coding, scenario generation, and documentation. This will allow analysts to spend more time on the more complex and more valuable parts of the job. Rather than replacing OR analysts, AI will move them up the value chain toward broader, messier, higher-impact decisions.

    Since 
    more organizations will attempt to automate decisions, it will raise questions that OR analysts are trained for:

    • What decisions should be made?
    • How do we measure good/bad?
    • What objective should be optimized?
    • How do we handle uncertainty?
    • What constraints are real?
    • What risks are acceptable?
    • How do we evaluate, audit, and improve performance? 

    These are OR questions. 

    For talented freshmen who like the material, I would encourage them to stay with it. 
    I don't know much about ECE and ME, but my guess is that they are not necessarily safer from AI disruption. Every field will be affected. The better strategy is not to choose a field that avoids AI, but to choose one, like OR/MS, where AI becomes a force multiplier. 



    ------------------------------
    Warren Hearnes, PhD
    Founder, OptiML AI
    INFORMS Board Role: VP Technology Strategy
    ------------------------------



  • 6.  RE: Will AI eliminate OR jobs

    Posted 17 days ago

    This is a great question and there are already some good answers. Here is some additional food for thought.

    First, according to the US Bureau of Labor Statistics, demand for industrial engineers is much higher than average. It is expected to grow 11%, which is higher than every other engineering discipline (https://www.bls.gov/ooh/architecture-and-engineering/industrial-engineers.htm). This reflects a number of factors and suggests that operations research and industrial engineering are great career bets.

    I've been reading various reports on the "AI exposure" of various jobs. While many white collar jobs are at risk, the skills we use in operations research have less exposure than most (translating messy real‑world systems into models, balancing multiple objectives and constraints, addressing uncertainty with multiple human stakeholders). 

    All in all, I agree with Warren:  The better strategy is to choose a field like OR/MS, where AI becomes a force multiplier. 

    Laura



    ------------------------------
    Laura Albert
    Industrial and Systems Engineering
    University of Wisconsin-Madison
    ------------------------------



  • 7.  RE: Will AI eliminate OR jobs

    Posted 15 days ago

    This is a great question, and there are already some good answers. Here are two pieces of information that suggest that operations research and industrial engineering are in fact the safer bets among engineering disciplines.

    According to the US Bureau of Labor Statistics, demand for industrial engineers (which includes operations research) is much higher than average. Demand for IEs is expected to grow 11% in the coming decade. This is higher than every other engineering discipline (https://www.bls.gov/ooh/architecture-and-engineering/industrial-engineers.htm)

    There have also been some reports on AI exposure. Roles that involve core operations research skills (problem formulation, translating messy real-world systems into models, balancing multiple criteria and constraints, and decision-making under uncertainty with human stakeholders) are much less likely to be eliminated. 

    I believe operations research and industrial engineering are great bets for college freshmen. I wholeheartedly agree with Warren: in the coming decade OR expertise will move them up the value chain and make them a force multiplier.

    Laura

     



    ------------------------------
    Laura Albert
    Industrial and Systems Engineering
    University of Wisconsin-Madison
    ------------------------------



  • 8.  RE: Will AI eliminate OR jobs

    Posted 10 days ago

    I was on vacation for a week, so my response is delayed.  The responses from Jiaxi Zhu, Warren Hearnes and Laura Alpert cover the needs for OR skills pretty well.

    I'd also look at it from another angle. If you consider the recently published INFORMS Analytics Framework, and the tasks involved in each domain, I don't foresee that many of those tasks will get replaced by AI, especially the ones in Domains I (Business Problem/Question Framing), II (Analytics Problem Framing), VI (Deployment) and VII (Analytics Solution Life Cycle Management).  The concept of being a "translator" - one who does business problem/question framing and then maps that to an analytics problem are something that I don't think AI will replace.  It requires interactions with stakeholders and an understanding of the underlying technology and how it can be leveraged to deliver value.  It may be true that AI can eventually write decent models and do some level of data analysis that determines valuable insights. But you have to gain the skills to verify that those models and analyses are correct.



    ------------------------------
    -Irv Lustig
    Optimization Principal
    Princeton Consultants
    ------------------------------



  • 9.  RE: Will AI eliminate OR jobs

    Posted 2 days ago

    Hello. I'm also late to the party, someone just pointed me to this thread.

    POV: my background is operations research, my job title is data scientist, and among other things, I have had several projects involving Generative AI where I work with business partners (internal customers) to use Generative AI as part of their jobs.

    1. In every Generative AI project I have had, part of the development and testing phase involves working with the business partners to find what Gen AI gets wrong about their setting. Every business partner I have worked with has successfully identified classes of mistakes the Gen AI makes. We can then add context to mitigate this, but it never goes away so even at the domain expertise level, Gen AI becomes a non-expert aide. (My webinar/workshop covers classes of mistakes)
    2. In every OR model (simulation or optimization) that I have done in a practice setting, I have never used a textbook model out of the box (my career has been in both industry settings and industry facing academia).  I have always had to add in a context specific detail to the model that made the model specific to the organization and setting.  Gen AI tends to give generic answers, and generic models. It makes for a good starting point, but modifying the model to a specific setting still requires a person to recognize that this is needed and then to do it.
    3. When I talk to Gen AI consultants, they have encountered a similar situation. When a client company has characteristics that differ from others in its industry, Gen AI gives the answer that is like a generic organization. In my case, my company claims to be different in approach, strategy, and organization than most others in its field. And I have had Gen AI projects where we realized the answers we did not like would probably be accepted by our competitors.  Gen AI has trouble with goals and intent, and companies using Gen AI needs people who can translate values, purpose, goals, and intent into models to supplement what Gen AI will do.


    ------------------------------
    Louis Luangkesorn
    INFORMS-Pittsburgh President
    Lead Data Scientist
    Highmark Health
    Pittsburgh PA
    ------------------------------