Automation, AI, and the Labor Market

AI is a big discussion topic right now. Collective bargaining groups have presented doomsday scenarios of AI destroying industries and advocating for guardrails.

The world will change over the next decade, just as ever since the advent of the Industrial Revolution. Some jobs will disappear, and new industries will bring new jobs. Countries will adapt differently, and some countries will benefit more than others. In this article, I want to take a step back from the hyperbole, look at how automation has impacted jobs in the past, and address how I plan to remain relevant in an ever-changing landscape.

Automation Isn't New

We have seen this before. Automation and progress are a matter of life; AI is just another demonstration of progress. The rise of personal computers caused a vast reduction in the number of secretaries, and the jobs that remained vastly changed. Every new technology in farming brought a reduction in the number of farmers, and consolidation brought economies of scale that reduced costs.

The difference now is:

1. People see it coming for more white-collar jobs
2. People think the transition is faster than it is
3. LLMs make it feel more human than other technology

Watching the public reaction to AI, especially the non-reaction until ChatGPT was released, has been fascinating. Aladdin by BlackRock has a massive influence on financial markets. For years, companies have poured money into AI-based self-driving cars and trucks. People, predominantly white-collar workers on the coasts, had little sympathy for complaints at the time, saying (in a derogatory manner) that they should learn to code instead. The innovation for self-driving trucks is taking longer than doomsday people predicted, and truck drivers are still an essential part of the economy.

Ways that AI May Impact Jobs

When you look to the past, AI could have three patterns of innovation impacts on existing positions. The effect will vary by industry, role, region, and other factors.

Replacement

Replacement is the impact that most people are concerned about. Replacement is when automation mostly or entirely replaces jobs currently done by people. The most obvious example that comes to mind is grocery store check-out. When I go to Walmart, I usually use in-app Scan and Go to get through the store faster. When I head home from work, I frequently visit Amazon Go, where the store doesn't require scanning.

Automation will replace some jobs, and it is challenging to establish which jobs will see a complete replacement. However, the replacement scale will not be as severe as people worry about (especially in the short term).

Consolidation

Many industries are going to undergo consolidation in the next decade. People will still lose jobs, but not everyone will. Instead, AI will allow fewer people to be vastly more productive than they would currently.

Farming is an area that has undergone immense consolidation throughout history. We have gone from small-scale, highly manual farms to small amounts of farmers handling hundreds of square miles. Some tasks like picking are still manual (and start-ups are trying to automate), but it has been transformative in allowing people to have cheap food.

While replacement and consolidation will have some people leave and some people remain, the difference is that consolidation comes with productivity increases. You will likely see entry-level roles disappear, with more senior-level positions remaining. This trend is not a matter of greed. Since senior-level people already write detailed descriptions of expectations and ask entry-level to implement them, leaders can delegate to generative AI instead.

Consolidation for Software Engineers

A major topic of discussion that I have seen online is how AI will impact software engineering. Some people say AI is lowering the barrier of entry, while others say it will destroy the industry. The answer is in the middle: it will bring consolidation and nearly eliminate junior-level positions while keeping senior-level posts in place.

Many software organizations already operate because a senior-level person creates a task in Jira, describes what is needed, and assigns it to a junior dev. That junior dev then writes to the specification and makes a merge request, which is reviewed and approved. That puts junior devs in a precarious position where their inputs and outputs are replaceable with OpenAI Codex or other generative AI.

Software engineering has been in a bubble for the last decade. Easy venture money and near-zero distribution costs mean that teams could be more bloated than they otherwise needed to be. Much of that bloat went to entry-level positions, which people could use to learn on the job and make lots of money. The rise of boot camps and near-100% job placement rates for graduates from some universities are evidence of this bloat.

However, the bubble is about to burst. With the widespread layoffs around Silicon Valley and the investor shift to profitability, entry-level positions will dry up. The same thing will happen with security. The days when people can party in university, not do independent projects and internships, and graduate with a decent job are over. Demonstrated skills will be a prerequisite for most jobs, and many in or entering university will not be ready.

In the past, the main criterion for a job in the industry was a decent ability to write code, which students could demonstrate in the classroom. However, those days are over. Software architecture and understanding the underlying languages will be the focus for the future. The students who use CODEX to complete their assignments and don't understand architecture will have near-zero economic value. Those who survive will enter the industry with senior dev positions and receive good pay since their productivity will skyrocket.

It hurts to see first- and second-year students think they can get away with using AI to complete their assignments. They are throwing away their tuition money, and it is ubiquitous. The days when an inefficient industry could cover laziness during university are numbers. I am so aggressive about maximizing the value of my education because I know that trying to skate by and get a job at graduation is turning into Russian roulette. I don't want to gamble on when entry-level positions will disappear, so I want to skip over that stage or spend as little time there as possible.

Relocation

The third option is what we saw with domestic manufacturing: price-sensitive jobs move outside the West, and the West has to compete on skill. The collapse of manufacturing was not global. Low-skill jobs relocated to Mainland China, Vietnam, India, Mexico, and elsewhere. Higher-skill jobs, like automobiles and specialized equipment, remained in NAFTA and the EU. However, this shift to specialization brought changes in the market. We generally saw fewer employees as people worked alongside robots. Skill requirements increased, and some industries (like foreign auto manufacturers) shifted from bloated, expensive, and inefficient union labor to a more agile non-union workforce. The smaller number of workers meant that the remaining received higher pay.

Relocation opens markets to greater competition. Competition may be between countries, but it can also happen within a country. Some tech companies have shifted their headquarters from California to Texas for a more business-friendly and cost-efficient environment. Different regions benefit differently based on how they choose to compete. The "losers" are not necessarily the regions that have an existing industry but rather those that fail to adapt and compete.

With AI, we will likely relocate for many jobs that require English proficiency. English proficiency has been an inhibitor to relocation in the past, but ChatGPT makes it so that a person who may not be proficient can still communicate as if they are. Adaptation is essential in a changing economy, and a person who relies on being a native English speaker (writers, HR staff, etc.) as a critical differentiator should focus on adapting.

To Resist or to Adapt

When technological shifts change the labor market, there are two ways to respond: resistance or adaptation. Resistance means that a person or organization aims to prevent the widespread adoption of an innovation. The argument generally is that innovation will hurt existing labor market participants and that legislation and collective bargaining agreements should curtail adoption.

Change is difficult. People want to resist change, which is why politicians from populist coalitions of the left and right in almost every developed country are talking about fighting AI. Yet, the short-term interest in "protecting jobs" has severe macroeconomic consequences. Resistance to change necessitates isolationism. If one country resists AI-assisted productivity to save jobs, another country will embrace it and produce cheaper products. Those products will become popular globally, slashing the resisting country's exports. That resisting country may respond with import tariffs to support domestic industry. However, the resisting government would have permanently ceded global trade and caused widespread, long-term obliteration of their GDP.

At a national and individual level, the better approach is quickly monitoring and responding to market shifts. While distinct sectors and positions have grown and faded, the overall labor market has been highly resilient and, at least in the US, generally improved. Some countries have seen labor market contractions outside of global recessions, but those often happen in countries that have tried to resist market forces.

I am continually monitoring and adapting to the forces I see or expect. One of the reasons why I chose to pursue a Computing Security degree over a Software Engineering or Computer Science degree was that I saw the gradual outsourcing of entry-level programming positions to the developing world. I knew the more advanced CS positions were continuing to grow, but I wanted to position myself toward a growing industry. I have placed myself on a path that will retain relevancy in the future. If I start to notice factors indicating a contraction, I will strive to quickly adapt to a new direction (either within the security domain or another domain entirely).

The Dramatization of AI

The world has and will continue to evolve. Every technological change has brought tremendous resistance, and AI is no different. While technical evolution affects people differently, the fact is that the world continues to be better. Yes, some jobs will disappear. Other jobs will see significant reductions. Some regions will be negatively affected. Detroit suffered from the manufacturing relocation, and West Virginia suffered from a shift from coal to LNG and renewables.

The doomsday perspectives on AI primarily come from a place of resistance to change. People want technology to stay where it is, but macro forces mean it isn't possible. Adaptability will be vital. New job opportunities that nobody can currently imagine will open up. Worker productivity will increase, and the cost of goods and services relative to average income will decrease.

I have sympathy for the people who will feel negative impacts from AI. The people who may have to change positions or industries. Not everybody can be a "winner" in the short term from technological revolution. Yet, in the long term, I do think that it will be a net positive for the majority of people.

I feel tremendous sympathy for the West Virginia coal miner whose mine shut down due to more efficient forms of electricity. Yet cheaper and more efficient energy sources have created so much good in the world. I feel sympathy for the Detroit auto worker who lost their job as foreign auto manufacturers came in and set up shop in Canada, Mexico, and the south. Yet, people can get better cars for far less money thanks to global brand presence in the US auto market.

I know that these changes may have severe regional effects. I live in a city that has undergone painful urban decay ever since the downfall of Eastman Kodak. Nostalgia is at the center of regional politics. The City of Rochester, Monroe County, and New York State talk about upstate decay, and the brain drain from Rochester's two major universities. Kodak is largely gone, and no amount of wishing is going to bring it back. AI will make a lot more Detroits and Rochesters. But it will also make a lot more Austins and Singapores.

No amount of regulation, collective bargaining, or contract negotiation will keep the world in the past. The stronger the resistance to change, the stronger the negatives will be compared to the positives. Monitoring and adapting is a far better option in the long term than resisting. There will be some pain, as there has been every time we have seen technical revolutions before, but I am confident the world will end up better due to AI.