davidbau.com

Freedom and Purpose

I spent 20 years making products in industry before switching to teach in academia, so I am frequently asked to compare the two paths by PhD students who are facing the choice between them. Here is my answer: industry and academia fundamentally have two different missions, and when choosing between them you should think about what kind of impact you would like to have on the world....

Differences Blur in Prosperous Times

It can sometimes be hard to tell the differences between academia and industry when times are flush.

If you work at University of Toronto in AI when the field is exploding, or if you work in Harvard at the height of the biotech boom, or Stanford during the Internet boom or at Caltech during the space race, lots of money will be pouring in to your academic research lab. Every year, professors and students will be spinning off startups, lots of product development ideas will circulate in the hallways outside classrooms, and the whole academic environment will feel similar to a startup. You will see Google and Facebook getting started in a dorm room.

Or if you work at Xerox PARC, Bell Labs, DEC SRC, Google DeepMind, or OpenAI during those companies' monopoly dominance, you will have lots of freedom, with researchers wandering the open hallways and writing on topics from philosophy to neuroscience, with open publishing, open source, full transparency, and collaborations with professors and PhD students everywhere. You may actually teach university classes, conduct open-door seminars, and entertain lots of risky viewpoints. You will see intellectual revolutions like Shannon's information theory or LeCun's convolutional nets emerge from your industrial environment.

During these prosperous times you will think "this is wonderful—this company is like an academic lab with all the freedom of university, but with more money." (Or: this academic department is just like a startup incubator, but with more freedom.)

The Real Job is Revealed When the Going Gets Tough

But when the going gets tough, you will find that, in industry, freedom is not essential. And in academia, the money is not essential either.

When competition heats up, industry labs will return to their strength and their mission. They will jealously guard their resources: their money, their hardware, their army of employees, and they will trade away freedom when necessary to pursue their mission. I have explained this to many students who have gone to industry labs, and they often tell me that I am wrong and that "Google is different" or "OpenAI is a nonprofit" etc. But I am not wrong: in recent years we have seen the dynamic of fiduciary responsibility playing out very dramatically in the AI industry, where a veil of secrecy has descended on the corporate labs. All the big industry AI labs have turned on a dime, now focusing on preserving and raising capital rather than freedom. Researchers who were recently free to speak and collaborate and work on any direction they chose are now mandated to keep secrets and focus their attention on projects that directly create value. There is nothing wrong with this! I loved working in this setting for 20 years. That is the job.

Similarly, in a university lab, when interest in an area dries up, it will reveal itself to be different from a company. Professors will return to their strength and their mission. They will jealously guard their freedom: their ability to publish, to collaborate, to speak, the freedom to teach students, and they will trade away their money when necessary to pursue their mission. When the tide turns, don't be surprised when you are no longer consulting for venture capitalists and you cannot afford fancy equipment. You will need to spend more time teaching more courses. That is the job.

At least: if academics are good at their mission, that is what they should do.

Capability versus Credibility

Long-term investments look a bit different at companies and universities.

In a company you must be a steward of your capability, your ability to design and produce valuable products. In recent decades, it has been fashionable for fresh-faced MBA managers to goose fiscal efficiency by outsourcing expensive employees or by selling off costly real estate, but in the long run such moves are often losers for a company. If you reduce your corporate capabilities, eventually you will destroy your company.

In academia, you must take care of your creditibility, your ability to uncover and teach purposeful knowledge, In the short run it might look good to put a shine on your research, to fudge the truth and tell an exciting story. And it might seem smart to adjust your teaching to fit the agenda that wealthy funders are looking for. But often such moves will be losers for a university, because if you sell off your credibility, eventually you will society to ask "why should we believe anything you say?"

Just as companies are sometimes short-sighted, somtimes universities make terrible mistakes.

If Not Value Creation, What is the Academic Goal?

If industry is about creating value and academia is not, then maybe you would conclude that academia is just a playground where people spend their time doing nonsense. But far from it. The work of academia is crucial in society and just as important as anything done in industry.

The essence of being human is to decide what we want in the world. This sounds easy—most economists assume that people inherently do this—but actually, knowing what you want is remarkably difficult.

Consider your deliberation every time you decide to start a new job, buy a new house, or go on a date with a partner. Deciding is not easy!

Your ability to really know what you want is intertwined with your ability to predict the future, to perceive the conditions in the world, understand your own psychology, strengths and weaknesses, and to anticipate your effect on your self, your family, your peers and society. It can help to understand your principles, and it can help to have a sophisticated understanding of how the world works. If you are taking on a technical job, your decision might depend on your opinion of CMOS manufacturing processes or your view on the usefulness of RL optimization objectives. Without that knowledge, it would be very difficult to know what the job is, and whether you actually want to do it or not.

Everybody understands that industry's role is to provide customers, who buy products, with some value. What is less well-understood is the role of academia, but it is analogous.

The role of academia is to create purpose in the minds of students and in society.

Posted by David at March 25, 2025 07:17 AM