The future of corporate office work

I was recently listening to the Dwarkesh Podcast, great podcast, you should check it out. Across multiple episodes the question is asked

"when will AI be able to replace remote workers."

Breaking down the question, it carries a few assumptions. The first assumption is that it is the workers that will be replaced. The second is that remote workers are the first to be replaced.

A bleak future where massive layoffs impact white collar workers replaced by AI agents is a real possibility. The need for people fluent in programming languages and skilled at the day to day writing tasks will quickly dry up. How many ActionScript developers are still employed off of that specific skill? Oddly I wrote about this in 2017 not thinking about AI at all at the time but rather about the rapidly falling barriers to entry into the career of a software developer, or more rightly, the function of a software developer. The story is the same though.

I'm tempted to remind us all that there used to be rooms filled with people hand copying legal contracts, followed by rooms full of people typing on typewriters, followed by photocopy businesses, and now I haven't touched a piece of paper in over 5 years despite negotiating and signing over 200 contracts in that time span. People simply found much more productive things to do with their time than copying contracts, shuffling papers, repairing printers, storing and moving filing cabinets, etc.

Futurists are right to point out that "this time it's different." People will say that the transition from scrivners to typists to photocopiers to DocuSign was just a process and tool change. People, actual humans, were still the ones in control, still doing the thinking, still monetizing their time. The only thing that still remains in the modern world is the thinking job at the top of that series of changes and now technology is coming for that too!

I think the focus is on remote workers first because after all, they are merely virtual people, and already you interact with them in text or video through the same modalities you would an AI agent most of the time. It is easy to imagine sending a Slack message to an AI drop in replacement for a remote worker because the response or the agentic activity that happens asynchronously will be exactly the same as if it was a real human working from home. It is going to be much more difficult to replace that person that you walk to coffee with between meetings, or the IT guy that fixes the broken conference room screen, or that security guard that monitors the parking lot. AI can't do all that, am I right?

And so the logical conclusion of most people is that in fact the big companies will lay off their thinking staff and replace them with thinking machines, and this time, unlike all the previous technological revolutions, "we're cooked." But let's take a moment and pause that thought. Why do companies need most of their roles in the first place? They need a means of production and sales at its most basic level and that means they need to have people (historically at least). And because they have people, they need people to manage the people. They need a facility for the people (or they think they do). They need recruiters, they need financial managers, human resources, on and on.

Simply put, to scale a business is a big business. In 1934, Ronald Coase wrote the seminal essay The Nature of the Firm where he questions the underlying assumption for why corporations or "firms" exist at all. At the end of it all, the conclusion that Coase and future economists have reached is that firms exist to reduce transaction costs or to reduce the cost of coordination.

If you aren't familiar with this thinking, I'll try to modernize the language. Business at its most atomic level is conducted by individuals acting on their own. Someone picks lemons from a tree, squeezes them into a cup, stirs in sugar, adds ice, sells it as lemonade. In theory, every element of the economy could be run this way, but it has not been that way. Instead we have huge corporations to create efficient economies of scale.

Imagine if a CEO of a company that bought, repackaged, and sold collateralized fixed income insurance portfolios had to actually do all of the work herself. She had to source the portfolios, she had to liquidate things, she had to negotiate contracts, when she got sued she had to defend herself, etc. She could do it but it wouldn't scale well. She could hire out for certain tasks. A lawyer for this, an accountant for that, a marketing expert for something else, etc. But, for each of these things there is a cost of time and effort to find the external vendor, negotiate the price, verify the results, establish trust, etc. A firm will instead align its needs such that if it makes sense, there will be internal accounting departments talking to internal marketing teams talking to internal legal teams and the CEO can focus on the next highest order value creation and the opportunity cost of coordinating everything will be traded for the cost of payrolling a team.

Now, I would like to put two things together and postulate that rather than ask "when will AI replace remote workers." I would rather us be asking, when will AI lower the transaction and coordination cost of doing business to such a drastic extent that we no longer need large firms.

The result is the same to start with. Large firms shrink drastically, laying off millions of knowledge workers. Where I would like to focus is on what happens next.

If AI has equipped knowledge workers with a massive force multiplier, they will no longer need the firm because they will no longer need to coordinate nearly as much. If AI can write the code, do the accounting, pay the bills, maintain legal compliance, provision and optimize infrastructure and security, do customer support, generate marketing content, etc. There very quickly becomes less and less need to payroll a staff to build a business at scale.

A new artisan class can emerge in a post firm era where the so called soloprenuer or small teams of 3-5 people can run multi-million dollar or billion dollar busineses. In hindsight the expenditure of time, energy, and capital spent on large corporate functions will seem ridiculously inneficient and wasteful. It wouldn't be worth the rubber worn off the tires of your VP of Legal Affair's Porsche to drive into the office.

If, as many have questioned, this era of corporate scale has been in fact a huge waste of resources, then this next era could be a golden age like we have never seen before. It won't be the people that get replaced, it will be the firms that will be replaced.

In a less happy note there will likely be a great number of people left behind by this rapidly accelerating transition because most people are not mentally or spiritually prepared to fend for themselves as entreprenuers. It will take a level of mental plasticity that this generation has not yet encountered, even with the rise of the PC, the internet, mobile phones, all within my lifetime. For some people there will be absolute wins and for some there will be a desperate level of fear and uncertainty. With that I genuinely fear political instability and unrest at a scale we haven't encountered since the 1920s and 30s and unfortunately. Given there is little that can be done about that risk, the cat is out of the bag so to speak, if you want to be on the right side of history, my thesis is that we should not ask "when will all of our jobs be replaced" but instead we should ask "when will all of these firms be replaced?" Go start something new.