There’s a lot of anxiety around artificial intelligence right now, and much of it centers on jobs.
People are watching companies reduce headcount. Recent graduates are struggling to find entry-level positions. Entire professions are wondering how much of their work AI will eventually be able to handle.
At the same time, we hear another argument: technology has always disrupted jobs, and new jobs have always emerged.
Both ideas contain some truth.
The part that matters most is what happens between the disruption and the recovery.
That transition has happened many times before. New technology changes how work gets done. Some tasks disappear. Some jobs shrink. Businesses become more productive. New products, companies, industries, and occupations eventually emerge around the new technology.
AI appears to be following that familiar economic pattern.
What makes this moment unusual is the speed.
We may be experiencing decades of technological change compressed into a much shorter period of time.
And that changes everything.
We’ve Been Here Before
Every major technological shift has changed the workforce.
Mechanization transformed agriculture. Industrial machinery changed manufacturing. Computers automated clerical work. The internet disrupted retail, publishing, travel, communications, entertainment, and countless other industries.
Those changes created winners and losers along the way.
They also created work that would have been difficult to imagine beforehand.
Think about how many jobs exist today because of technologies that once threatened existing industries: software developers, cybersecurity specialists, web designers, digital marketers, app developers, cloud architects, social media managers, data analysts, e-commerce specialists, and thousands of others.
Economists sometimes describe this as the interaction between displacement and reinstatement. Automation reduces the need for human labor in certain tasks while technological progress also creates new tasks where human labor becomes valuable again.
That process doesn’t happen evenly, and it certainly doesn’t happen painlessly.
A factory worker whose job disappears doesn’t automatically become a software engineer because the economy created a new technology job.
Training takes time. Businesses need time to reorganize. Schools need time to update curricula. Workers need time to develop new skills. New industries need time to grow large enough to employ significant numbers of people.
Historically, that time has acted as a buffer.
AI is shrinking the buffer.
The Big Difference Is Speed
The personal computer is a useful comparison.
After the introduction of the mass-market PC, adoption grew steadily over many years. Research from the National Bureau of Economic Research found that PC adoption rose from roughly 20% three years after introduction to around 70% more than two decades later.
The internet followed its own multi-year adoption curve.
Generative AI is moving differently.
NBER researchers found that workplace adoption of generative AI was already moving about as quickly as early PC adoption, while overall adoption was faster than both the PC and the internet.
There’s a simple reason.
Previous technologies often required infrastructure.
A computer had to be manufactured, purchased, installed, and learned.
Internet access required telecommunications infrastructure, hardware, and service.
Mobile computing required networks, smartphones, and years of device replacement cycles.
AI can arrive through software.
A company can activate an AI service, connect an API, deploy an agent, or add an AI feature to software its employees already use.
Millions of people can receive a new AI capability overnight.
There’s very little friction between invention and distribution.
That changes the economics of adoption.
The AI Adaptation Gap
This leads to what I believe is one of the most important concepts for understanding the AI economy.
I call it the AI Adaptation Gap.
It’s the growing distance between how quickly AI capabilities are advancing and how quickly the rest of society can adjust.
Think about the different systems involved.
- AI capabilities can improve in months.
- Companies can adopt software in weeks.
- Workers may need months or years to develop new skills.
- Schools can take years to redesign programs.
- Government policies and labor systems can take even longer.
The wider that gap becomes, the more disruption we’re likely to experience during the transition.
That helps explain why the current moment feels so unsettling.
People can see the technology changing faster than the systems they’ve traditionally relied on to prepare them for change.
A degree that once prepared someone for a career may now prepare them for a job that’s changing before they graduate.
A skill someone spent 20 years developing may suddenly be partially automated.
A company that once needed 20 people to perform a function may discover that 12 people using AI can produce the same amount of work.
Human work will continue. The mix of human work and machine work can simply change faster than our traditional systems are designed to handle.
Why Job Loss Can Feel Like a Cliff
Imagine two companies adopting two different technologies.
In 1985, a company wants to computerize an office.
It needs computers. Software. Servers. Training. IT support. Capital budgets. Installation. Employees have to learn completely new systems.
The transition naturally stretches over time.
Now imagine a company in 2026 deciding to automate part of its customer service, marketing, research, accounting, or software-development workflow.
The technology may already exist.
The company can subscribe to it.
That difference matters.
AI can potentially move from technological capability to economic deployment extraordinarily quickly.
Anthropic’s economic research also shows AI moving beyond individual chatbot use into API-based automated workflows, including areas such as customer service, software development, sales outreach, and other business processes.
Companies don’t necessarily have to wait for another generation of hardware before increasing automation.
Software can change continuously.
And when software changes continuously, the workforce can change continuously too.
I created the AI Shift Dashboard to help make that change visible. It tracks documented AI-related workforce displacement, hiring trends, economic indicators, and other signals so people can follow what’s actually happening instead of trying to make sense of isolated headlines.
You can also visit the AI Shift Dashboard overview for background on the project, methodology, and purpose behind it.
The Jobs We Don’t See Disappearing
Layoffs get headlines.
There’s another form of displacement that’s much harder to see.
It’s the job that never gets posted.
Suppose a department traditionally hires five junior employees each year.
AI allows the existing team to handle more work, so this year the company hires two.
Nobody was fired.
There’s no layoff announcement.
There are simply three fewer entry points into the workforce.
Multiply that across thousands of companies and the impact becomes significant.
That may become especially important for younger workers.
Entry-level positions have historically done much more than provide income. They’re where people learn how an industry works. They develop judgment. They learn from experienced colleagues. They make mistakes. They build relationships. They develop the skills required to eventually become senior employees.
If AI absorbs more entry-level tasks, we’ll also have to rethink how people gain entry-level experience.
That’s a workforce-development problem, an education problem, and an economic problem.
It’s also something we can solve.
AI Will Create Jobs Too
Technological change creates demand along with disruption.
We can already see new demand surrounding AI development, implementation, infrastructure, governance, security, training, and integration.
There will almost certainly be jobs ten years from now that have titles we rarely use today.
History gives us good reason to expect that.
The challenge is timing.
Imagine 500,000 jobs disappear or change substantially while 500,000 new jobs eventually emerge.
On a spreadsheet, the numbers balance.
For the people living through the transition, they may have very little relationship to one another.
The new positions may require different skills.
They may exist in different industries.
They may pay differently.
They may appear several years later.
That gap is where economic policy, corporate training, and education become incredibly important.
The question becomes less about whether AI will eventually create economic opportunity.
The more urgent question is:
How do we help people move from the work AI is changing into the work this new economy will need?
Productivity Will Change the Economy Too
Employment is only part of the story.
AI also has the potential to dramatically increase productivity.
One person may be able to accomplish work that previously required several people.
A small business may gain capabilities that once required an entire department.
An entrepreneur may be able to research a market, develop a business plan, create marketing materials, build software, analyze data, and communicate with customers using tools that previously required specialized teams.
That productivity can create new companies, products, and services.
It can also allow existing companies to grow without adding employees at the same rate they once did.
That means traditional measures such as unemployment alone won’t tell us everything we need to know about AI’s economic impact.
We also need to watch:
- Hiring
- Wages
- Entry-level opportunities
- Productivity
- New business formation
- Changes in job descriptions
- Demand for specific skills
- The amount of work being performed by humans compared with machines
These are some of the signals the AI Shift Dashboard is designed to bring together in one place.
Education Is Our Best Tool for Closing the Gap
This is the part of the conversation I believe deserves much more attention.
We spend enormous amounts of time debating what AI might do.
We need to spend just as much time helping people prepare for what it’s already doing.
AI literacy needs to become a basic professional skill.
That doesn’t mean everyone needs to become a programmer or understand how large language models are built.
Most people never needed to understand how a CPU worked to benefit from a personal computer.
They learned how to use one.
The same principle applies here.
You need enough understanding to recognize what AI can do, where it can help you, where its limitations are, and how it changes the value you bring to your work.
I created The AI Shift: What’s Happening, What It Means for You, and How to Prepare as a practical starting point. It explains what’s changing, which types of work are most exposed, how to audit your own job, how to start using AI effectively, and which human skills are likely to become more valuable as automation expands.
The AI Shift Learning Center takes that a step further with resources designed to help people understand AI and start building practical skills.
Start With Your Own Job
One of the most useful exercises you can do is break your job into tasks.
Ask yourself:
- What do I do repeatedly?
- What involves gathering or organizing information?
- What follows a predictable process?
- What could I teach someone else to do with clear instructions?
- Where does my experience genuinely change the outcome?
- Where do people depend on my judgment?
- Where do relationships matter?
- Where does taste, leadership, empathy, or context matter?
Then experiment with AI on the first group.
Use it to research.
Summarize.
Analyze.
Draft.
Organize.
Generate options.
Automate repetitive processes.
The goal is to understand the technology through experience.
As those tasks become easier, spend more energy developing the second group: judgment, creativity, communication, leadership, relationships, and decision-making.
Research on AI use is already finding that experienced users develop better ways of working with these systems and are more successful at getting useful results from them.
AI fluency can be learned.
You Have More Control Than It Feels Like
Nobody knows exactly where this transition ends.
There are simply too many variables.
But you don’t need to predict the entire future to prepare for it.
You only need to understand the direction of travel.
AI capabilities are improving.
Businesses are adopting them.
Work will continue changing.
New opportunities will emerge.
And people who understand how to work with these systems will have more options as the transition unfolds.
That’s something you can act on today.
- Spend an hour learning how an AI tool works.
- Use it on a real task.
- Find one repetitive part of your job and see how much of it you can automate.
- Identify one human skill worth strengthening.
- Help someone else learn what you’ve learned.
- Then keep going.
Small steps compound.
Understanding Changes Everything
Fear grows easily when something feels unpredictable and outside our control.
Understanding gives us perspective.
The economic changes surrounding AI make much more sense when we see them as part of a technological transition that has happened many times before, combined with a rate of change we haven’t experienced before.
That distinction matters.
We have historical evidence that economies adapt to technological change.
We also have good reason to take this transition seriously because the adaptation period may be much shorter.
So our response needs to move just as quickly.
We need better education, better workforce training, more AI literacy, more honest conversations between companies and employees, education systems that prepare people for the economy they’re entering, and individuals willing to keep learning throughout their careers.
The future of work isn’t something that suddenly arrives one day.
We’re building it now.
Understanding what’s happening is the first step.
Learning how to participate in it is the next.
And once you start doing that, the AI shift becomes a lot less frightening.
It becomes something you can prepare for.
Sources
- Alexander Bick, Adam Blandin, and David J. Deming, “The Rapid Adoption of Generative AI”, https://www.nber.org/papers/w32966, Published: September 2024, Revised: February 2025, Accessed: August 11, 2026
- Daron Acemoglu and Pascual Restrepo, “Automation and New Tasks: How Technology Displaces and Reinstates Labor”, https://www.aeaweb.org/articles?id=10.1257/jep.33.2.3, Published: Spring 2019, Accessed: August 11, 2026
- Anthropic Economic Research, “Anthropic Economic Index Report: Learning Curves”, https://www.anthropic.com/research/economic-index-march-2026-report, Published: March 24, 2026, Accessed: August 11, 2026
- Sabrina Wulff Pabilonia and Jill Janocha Redmond, “The Rise in Remote Work Since the Pandemic and Its Impact on Productivity”, https://www.bls.gov/opub/btn/volume-13/remote-work-productivity.htm, Published: October 2024, Accessed: August 11, 2026