Both Sides of the AI Table: What Business Owners and Knowledge Workers Need to Know

According to McKinsey's November 2025 analysis, 57% of U.S. work hours could be automated with technology that exists today. Not in five years. Not pending some breakthrough. Today.

Every automation wave in history went after physical labor first. Factory floors. Warehouses. Manual assembly lines. The knowledge worker watched from a safe distance, degree in hand, convinced that cognitive work was the durable hedge.

That assumption is now wrong.

This is not a post about whether AI is good or bad for the economy. It is about what is happening right now, what it means depending on which side of the table you sit on, and what either party can actually do about it. Business owners and knowledge workers are staring at the same shift from opposite directions. The view is different. The urgency is the same.

What Is Actually Happening

The World Economic Forum's 2025 Future of Jobs Report projects 92 million jobs displaced and 170 million new roles created by 2030. The net number is positive. But the net number does not tell the story of transition. It does not tell you what happens to the analyst whose role disappears in 2027 while the new role requires skills she does not yet have.

Forty-one percent of employers surveyed by WEF are already planning to reduce headcount in areas where AI automates tasks. In 2025, 55,000 job cuts were directly attributed to AI. Amazon eliminated 14,000 corporate roles in the second half of last year, followed by 16,000 more in early 2026. Salesforce cut 4,000 support positions after AI handled half of all customer queries.

Here is something worth naming clearly. Research published in HBR and confirmed by workforce analysts at Metaintro suggests that nearly 6 in 10 employers citing AI as the reason for cuts were actually responding to budget pressures or post-pandemic overhiring corrections. AI becomes an acceptable narrative for decisions that were already coming. That matters, because honest employers and honest workers need to separate real AI displacement from AI as a cover story.

The real displacement is coming. It does not need embellishment.

If You Are the Employer

You are about to receive a productivity windfall. The cognitive overhead your team carries today, the research, the reporting, the drafting, the data entry, the pipeline hygiene, AI absorbs a meaningful portion of it. The efficiency argument is real.

The temptation is to treat that as a pure cost reduction exercise. Headcount down, margins up, board pleased. Some companies will do exactly that. Some already have.

The smarter question is what you actually need from each person now that AI handles what it handles. Because the tasks that disappear first are not the tasks that made your best people valuable. They are the tasks that consumed time and kept people from being more valuable. Judgment. Client relationships. Accountability. The ability to read a room, navigate a difficult conversation, or make a call when the data is incomplete. AI does not do those things well. Your people do.

Before you model out the headcount reduction, model out the job redesign. What does the role look like when the low-level cognitive work is offloaded? In many cases, it becomes a better job. In some cases, it becomes fewer jobs. Both outcomes deserve a clear-eyed assessment rather than a reflexive cut.

The obligation question is harder, and most business leaders are avoiding it. The productivity gain your company captures came from tasks your people used to perform. They showed up. They did the work. They built the institutional knowledge your AI tools are now drawing on. What responsibility do you carry for their transition?

Harvard Law School's governance research is direct on this point: boards should weigh the human capital impact of AI adoption, not just the efficiency line. This is not soft language. It is a governance standard. Companies that treat AI deployment as pure optimization, without structured consideration for affected employees, are increasingly exposed.

In Colorado, the AI Act (SB 24-205) takes effect June 30, 2026, requiring employers to guard against algorithmic discrimination and imposing civil liability for violations. The obligation is no longer only ethical. It is legal.

The companies handling this well are not the ones cutting the least. They are the ones with a plan. Reskilling programs before the cuts. Transparent communication about where AI is being deployed. Internal mobility pathways for people whose current roles are contracting. Companies genuinely transforming are redeploying talent, not just optimizing headcount. That distinction will matter in every labor market they recruit from in 2027 and beyond.

If You Are the Knowledge Worker

The tasks AI displaces first are exactly the ones most job descriptions are built around. That sentence is worth sitting with for a moment.

Entry-level and early-career roles carry the highest near-term exposure. Microsoft Research found that workers aged 22 to 25 in highly AI-exposed roles experienced a 16% relative employment decline compared to peers in less-exposed positions. The first rung of the career ladder is getting shorter.

This is not a reason to panic. It is a reason to be honest about where your value actually lives.

The roles AI augments rather than replaces share a clear pattern. They require judgment that cannot be reduced to a prompt. They require relationships built over time. They require accountability that a model cannot carry. The question worth asking is not "can AI do my job?" It is "what part of my job would a client, a colleague, or a leadership team actually lose if I were gone?" The answer to that second question is where your protection lives.

The access problem is real and deserves acknowledgment. The WEF data projects that 59 out of every 100 workers will need reskilling or upskilling by 2030, and 11 of them are unlikely to receive it. Not because they lack ambition, but because the pathways are not equally distributed. Reskilling resources, AI literacy programs, and career transition support are more accessible in tech-forward metros and large enterprises than in mid-size companies and regional markets.

Colorado's technology workforce is navigating this shift in real time. AI roles in the state average $108,000 annually, and job seeker searches for AI positions have grown 11 times since the launch of ChatGPT. Workers with demonstrated AI fluency are commanding wage premiums of up to 56% over their peers. The path forward is real. It is just not equally lit.

The System Was Not Built for This

Unemployment insurance. SNAP. Workforce retraining programs. These systems were designed for a specific kind of disruption: localized, sectoral, and somewhat predictable. A factory closes in a specific town. A mine shuts down in a specific county. The system can find the displaced workers because they are concentrated, and it can retrain them because the new industries entering the area have identifiable skill requirements.

AI displacement does not look like that.

It is simultaneous across industries. It is concentrated in white-collar work in a way no previous automation wave has been. And critically, it does not require a formal layoff event to register. When AI absorbs 30% of an analyst's responsibilities and the company simply stops backfilling that function at the entry level, that displacement is invisible to every existing tracking mechanism.

Congress does not officially track AI-attributed layoffs yet. Senator Hawley introduced legislation in late 2025 to begin doing so. That bill is still pending. The fact that we do not have a federal system to count what is happening tells you everything about where public infrastructure stands relative to the pace of change.

This is not a partisan observation. It is a structural one.

Who Owns the Transition

This is the question most business owners and most policy discussions are still avoiding.

Companies capture the productivity gain from AI deployment. The cost of the transition, retraining, income disruption, eroded career pathways, falls on workers and on public programs funded by a tax base designed around labor income. When AI displaces cognitive work at scale, payroll tax revenue contracts. The programs dependent on that revenue face pressure at exactly the moment demand for them increases.

The automation tax concept being debated in policy circles is designed as a speed governor. Not a penalty on AI adoption, but a mechanism to ensure the pace of displacement does not permanently outrun the economy's ability to reabsorb workers. Whether that specific tool is the right one is a legitimate debate. The underlying question, about who bears the cost and in what proportion, is not going away.

The honest answer is that companies, government, and individuals each own a piece of this. The proportion is the debate. And that proportion is being negotiated right now, in congressional hearings, in state legislatures, and in corporate boardrooms. Business owners who are not in those conversations will have the terms set for them. Workers who have not found a way to make their perspective visible will find the policy written without their experience in it.

Where Your Voice Goes

Most people assume they cannot influence how this plays out. That assumption is worth challenging. The infrastructure for input exists. Most people just do not know where to find it.

For both employers and workers, the U.S. Department of Labor's Employment and Training Administration is actively accepting stakeholder input on AI workforce policy. The DOL's AI Apprenticeship initiative, launched in April 2026, is working with employers and industry associations to shape AI training pipelines nationwide. There is a seat at that table.

In Colorado, the Department of Labor and Employment's Office of the Future of Work is the dedicated state-level channel for exactly this conversation. The Colorado AI Act's June 30, 2026 effective date creates a narrow window for employers to engage on implementation guidance before it takes effect.

At the federal level, your congressional representative is more accessible than most people assume. A direct, specific letter from a constituent who runs a business or works in a sector being affected by AI carries real weight in staff briefings. Congress.gov has a member finder that takes thirty seconds to use.

SHRM represents employers in workforce policy at the federal level and is actively shaping AI workplace guidance. The AFL-CIO is doing the same on the worker side. Both organizations accept member and public input on their policy positions.

For those who want to engage at the research and governance level, the Harvard Center for Labor and a Just Economy is publishing frameworks for worker rights in AI governance that are being used in actual legislative drafting.

And yes, OpenAI published an industrial policy document in April 2026 and is actively soliciting feedback at newindustrialpolicy@openai.com, with research grants of up to $100,000 available for work that builds on the ideas in the document. You may have questions about their motives or their record — those questions are well-documented. The document promises auditing regimes and worker protections; OpenAI's lobbying record against California's SB 1047 — a bill requiring exactly those things — tells a different story. The feedback channel is real regardless.

The Conversation That Needs to Happen

The frame of "AI versus humans" is already outdated. That was a useful shorthand for explaining a capability shift. It is not useful for navigating what comes next.

The actual conversation is about obligation, proportion, and timing. Business owners who deploy AI and capture the efficiency gain carry a responsibility that is partly ethical, partly legal, and increasingly strategic. Workers whose roles are changing need honest information about where the exposure is real and where the path forward actually leads. Policymakers need the lived experience of both groups to build systems that fit the shape of the problem, not the shape of the last one.

The window to shape this is not closed. It is narrowing. The channels to engage exist across federal agencies, state departments, professional organizations, and open policy processes.

Most people just did not know where they were.

Now you do.

FAQ

Q: Is AI really going to replace my job, or is this overstated? A: The displacement is real but uneven. McKinsey's data shows 57% of U.S. work hours are automatable with current technology. That does not mean 57% of jobs disappear overnight. It means the tasks that make up most jobs are changing. Roles built primarily around routine cognitive work, standard reporting, templated writing, and basic data analysis carry the most near-term exposure. Roles requiring sustained judgment, client relationships, and complex decision-making are more likely to change than disappear.

Q: As an employer, what is the minimum responsible action when deploying AI that affects existing roles? A: Transparency and a plan, in that order. Employees should know when AI is being integrated into workflows that affect their positions. They should have visibility into what that means. And the company should have an honest answer about retraining and redeployment before deployment decisions are finalized. Colorado's AI Act makes some of this a legal requirement as of June 30, 2026.

Q: Where do I start if I want to protect my career from AI displacement? A: Start with an honest inventory. Which parts of your current role could be replicated by a well-prompted AI model today? Those are your exposure areas. Which parts require sustained relationships, judgment under uncertainty, or accountability that no model can carry? Those are your protection. From there, the DOL's AI Literacy Framework and Colorado's Career Trail Guide are free starting points for identifying skill development pathways.

Both Sides of the AI Table: What Business Owners and Knowledge Workers Need to Understand Right Now