AI is expected to reshape roughly half of all U.S. jobs over the next two to three years, while eliminating 10% to 15% of them over a longer horizon of five years or more. That’s according to a Boston Consulting Group model covering 165 million jobs across 1,500 roles. The key takeaway: far more jobs will change than disappear, but the changes will be significant enough that most workers will need to adapt.
Which Jobs Are Most Exposed
The first wave of AI disruption is hitting white-collar work harder than many people expected. Roles centered on data analysis, bookkeeping, basic financial reporting, and repetitive administrative tasks are among the most susceptible to automation. If your daily work involves collecting, organizing, or summarizing information in predictable formats, AI tools can already do much of that faster and cheaper.
Creative professionals are feeling pressure too. Generative AI can now produce marketing copy, graphic designs, video edits, and stock imagery at a quality level that competes with entry-level human output. That doesn’t mean every designer or writer loses their job, but it does mean fewer people may be needed for production-level creative work, while the value shifts toward strategic thinking, brand voice, and creative direction that AI can’t replicate on its own.
Human resources is another area seeing partial automation. Tasks like drafting employment contracts, generating offer letters, running payroll, managing benefits enrollment, and distributing engagement surveys can be handled or heavily assisted by AI systems. The parts of HR that require judgment, like navigating a sensitive employee situation or building workplace culture, remain firmly human.
The pattern across all these fields is consistent: tasks that follow clear rules and use structured data are vulnerable. Tasks that require nuance, relationship-building, or complex judgment in unpredictable situations are not.
New Roles AI Is Creating
While some positions shrink, entirely new job categories are emerging. Companies are hiring knowledge architects, who design how information flows through AI systems so outputs are accurate and useful. Orchestration engineers build and manage the pipelines that connect multiple AI tools into a single workflow. Conversation designers craft the way AI chatbots and virtual assistants interact with real people. And human-AI collaboration leaders oversee how teams integrate AI into their daily work without losing quality or accountability.
These titles didn’t exist five years ago. They reflect a growing need for people who understand both the technology and the human context around it. You don’t necessarily need a computer science degree for many of these roles, but you do need to understand how AI tools work, where they fail, and how to translate business needs into AI-assisted processes.
What’s Happening to Wages
The wage picture is more nuanced than a simple “AI raises pay” or “AI lowers pay” story. Since fall 2022, when generative AI tools became widely available, nominal average weekly wages across the U.S. have increased 7.5%. But the computer systems design sector, which sits at the center of AI development, saw wages jump 16.7% over that same period. Industries in the top 10% of AI exposure saw wage growth of 8.5%.
Look more broadly, though, and the relationship between AI exposure and wage growth is essentially flat. Research from the Federal Reserve Bank of Dallas found no meaningful correlation between how exposed an occupation is to AI and whether its wages grew faster or slower after 2022. Across 205 occupations, post-2022 wage growth averaged about 2.2 percentage points above prepandemic trends, regardless of AI exposure.
There is one important exception: experience matters. In occupations where senior workers earn significantly more than junior ones (a high “experience premium”), greater AI exposure is actually associated with slightly faster wage growth. For occupations where experience doesn’t command much of a pay bump, AI exposure correlates with a small drag on wages, roughly 0.28 percentage points. In practical terms, AI appears to reward workers who bring deep expertise and judgment, while putting downward pressure on roles where the work is more routine and interchangeable.
The Corporate Training Gap
Companies are spending aggressively on AI itself. Corporate AI spending is projected to rise 44% in 2026. But investment in training the humans who use these tools is not keeping pace. Training budgets are expected to grow just 5%, and average learning time per employee is actually falling, from 47 hours to 40 hours annually.
That imbalance creates a real problem. Competitive advantage from AI doesn’t come from the software alone. It comes from the judgment, adaptability, and institutional knowledge that employees bring when they use the tools effectively. Long-term performance depends on resilience, trust, and the capacity to adapt, all qualities that require investment in people, not just technology. Amazon has committed more than $1.2 billion to upskill hundreds of thousands of workers for technology-enabled roles, but that scale of commitment is rare.
If your employer isn’t offering AI training, the burden falls on you. Learning to use AI tools proficiently in your specific field, whether that’s using copilot tools for coding, AI-assisted analysis in finance, or generative tools in marketing, is quickly becoming as fundamental as learning spreadsheets was a generation ago.
How to Position Yourself
The workers best positioned for an AI-driven economy share a few traits. They combine domain expertise with AI fluency, meaning they understand their industry deeply and know how to use AI tools to work faster and smarter within it. They focus on skills that are hard to automate: complex problem-solving, managing ambiguity, building relationships, and exercising judgment in situations where the “right” answer isn’t obvious.
Practically, that means a few things. First, get hands-on with the AI tools relevant to your work. Don’t just read about them. Use them for real tasks and learn where they’re helpful and where they produce unreliable output. Second, invest in the human skills that complement automation: communication, leadership, critical thinking, and the ability to work across teams and disciplines. Third, stay flexible. The specific tools and platforms will keep changing, so the ability to learn new systems quickly matters more than mastering any single one.
The 10% to 15% of jobs projected for elimination won’t vanish overnight. They’ll erode gradually as companies automate tasks, consolidate roles, and restructure teams. Workers who see that shift coming and start adapting now will have a meaningful head start over those who wait until it directly affects their position.