I'm a lawyer in Ohio. Most days, I'm not debating abstract policy in a boardroom—I'm dealing with real people in real stress: families in transition, business owners trying to keep the lights on, and clients in criminal cases whose futures can turn on a single hearing date. My brand is blunt because the truth is blunt: you have rights. Rights aren't theoretical. They're what keep the strongest from steamrolling everyone else.
That's why I paid attention when Anthropic CEO Dario Amodei spoke at Davos about the future that advanced AI could create—one with massive economic growth, but also massive displacement if we don't prepare. He floated a scenario where AI could drive 5–10% GDP growth while unemployment rises to around 10%—a combination he argued we almost never see—and said government will likely have to play a role in managing job displacement and ensuring the gains are shared.
From where I sit—running a growing firm, serving as a municipal law director, and building legal tech designed to make help more accessible—that warning is less about politics and more about the stability of the social contract. If AI becomes a tool that produces prosperity for a narrow slice of society while hollowing out opportunity for everyone else, the backlash won't be surprising. It'll be inevitable.
What I care about first: rights, dignity, and trust
AI is going to change the legal profession, the economy, and public institutions. But in law, you learn fast that systems matter—not slogans. The rules and processes we build determine whether people feel protected or powerless.
In the U.S., due process isn't optional. It's the core promise that government power must be exercised fairly, with meaningful notice and a chance to be heard. Courts have said that for decades in contexts like public benefits and employment. See, e.g., Goldberg v. Kelly, 397 U.S. 254 (1970) (requiring procedural protections before termination of certain welfare benefits); Mathews v. Eldridge, 424 U.S. 319 (1976) (balancing test for what process is due); Cleveland Bd. of Educ. v. Loudermill, 470 U.S. 532 (1985) (public employee entitled to notice and opportunity to respond before termination).
That matters in an AI economy because automation doesn't just "speed things up." It can also make decisions feel unchallengeable. If a system denies a benefit, flags a citizen, grades a student, screens a job applicant, or drives a case outcome, people deserve transparency and a path to contest errors. If we don't build that in, we won't get "efficiency." We'll get resentment and instability.
The headline that should make everyone pause
Amodei didn't just talk about growth and unemployment. He also raised the possibility of a society that "decouples"—where a relatively small group captures an outsized portion of the upside while the rest stagnates. He described a "nightmare" scenario where a small elite benefits from enormous growth while many others are left behind.
Even if you disagree with his numbers, the direction of the concern is hard to dismiss: AI amplifies leverage. When one tool makes one person or one company 10× more productive, the rewards can concentrate—fast—unless we intentionally create broad on-ramps.
The Data Is Already Here
Workforce disruption tied to restructuring and automation is already being tracked. Challenger, Gray & Christmas reported 153,074 job cuts in October 2025, and noted that AI was among the leading cited reasons, with 31,039 October job cuts attributed to AI and 48,414 AI-attributed cuts year-to-date through October.
Whether AI is the sole cause in each situation is less important than the pattern: companies are reorganizing around capability. That's what technological inflection points do.
What's important to me: building ladders, not moats
I'm not anti-AI. I've invested heavily in understanding it. I've integrated it into legal operations and education. I'm building systems that help people get oriented, get informed, and make better decisions—faster. That's not because I want to replace lawyers. It's because I want to reduce friction, lower cost, and increase access.
But here's the part that gets lost in the hype: if we deploy AI like a shortcut to cut labor without building pathways for people to adapt, we create a two-speed economy. And two-speed economies don't stay stable.
So when Amodei argues that government may need to help ensure AI's benefits are shared—and that displacement at macro scale demands a response—I hear that as a practical warning, not a partisan talking point.
The question isn't whether government "should" be involved. The question is whether our institutions will be ready to deal with the consequences when AI accelerates faster than workforce transitions historically do.
A lawyer's lens: the difference between innovation and exploitation
In business law, we talk a lot about incentives. In criminal law, we talk a lot about accountability. The AI era is going to force both conversations at once.
If AI enables huge productivity gains, the natural market impulse will be to funnel those gains to owners of capital, proprietary models, data, distribution, and compute. That's not "evil." That's how the system is wired. But it also means we need counterweights if we want broad prosperity.
This is where competition policy and guardrails matter. Our legal system has long recognized the danger of concentrated private power—and the need to protect competitive markets. See, e.g., Brown Shoe Co. v. United States, 370 U.S. 294 (1962) (describing antitrust's concern with concentration and preserving competition); United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001) (addressing unlawful maintenance of monopoly power).
I'm not saying every AI company is a monopoly. I'm saying the stakes are big enough that we should not sleepwalk into a world where a handful of entities control the models, the distribution, and the terms of participation.
The "shared upside" agenda that actually works
From my vantage point—running a firm and serving a community—"sharing the upside" can't just mean slogans. It has to look like specific, durable mechanisms:
1) Make AI skills a baseline, not a luxury
If AI boosts productivity, then AI fluency should be treated like basic literacy. We don't need everyone to become an engineer. We need most people to become capable operators and supervisors of AI tools in their domain—healthcare, law, construction, logistics, finance, education. That's community colleges, apprenticeships, employer partnerships, and rapid credentialing. It's also retraining that respects adults with mortgages and families—training that's paid, structured, and connected to real jobs.
2) Incentivize business adoption that creates opportunity
If AI is going to be a productivity multiplier, small and medium-sized businesses need access—tools, training, and support—so Main Street isn't permanently behind Big Tech. When small businesses can adopt responsibly, they don't just survive; they hire and grow.
3) Build safety, transparency, and contestability into high-stakes uses
In law and government, you don't get to hide the ball. If AI is used in high-stakes settings, the people affected should have: a clear explanation of what it did (at an appropriate level), a human point of contact, a process to challenge errors, and auditing that isn't just internal. That's not anti-innovation. That's how you preserve legitimacy.
4) Prepare for workforce displacement with real tools, not wishful thinking
If displacement becomes as large as leaders like Amodei fear, then the response has to be real: wage insurance, mobility support, rapid reemployment pathways, and (yes) potentially new ways of distributing gains when productivity explodes. Even the layoff data coming out of Challenger's reports shows how quickly "restructuring" can cascade through sectors—and how explicitly AI is now being cited as part of the story.
5) Use public procurement to push standards
Government buys a lot of software and services. That purchasing power can create market demand for transparency, security, and interoperability—without needing to micromanage innovation.
The legal profession is a test case for whether AI helps regular people
In my world, the "shared upside" question has a very practical translation: Will AI make legal help more accessible—or will it just make the biggest players bigger?
I'm optimistic because legal work has huge areas that are expensive largely due to time, friction, and information bottlenecks. AI can help with: client education and intake, organizing facts and timelines, drafting and revision workflows, and operational efficiency that lowers costs.
But the profession also has a non-negotiable core: judgment. Advocacy isn't autocomplete. A custody case isn't a spreadsheet. A criminal case isn't a template. People come to lawyers because they need strategy, courage, ethics, and someone who will stand next to them when things get ugly. That's why I build AI to augment humans, not erase them. The mission is the same as my law practice: protect rights and get people through hard moments.
The bottom line
If the AI economy becomes a machine that concentrates wealth and destabilizes work, we will pay for it—politically, socially, and personally.
And if we build the right ladders—skills, safeguards, access, competition, and fair participation—AI can become one of the most empowering technologies we've ever seen.
Amodei's warning isn't that AI is bad. It's that the default settings of the economy don't guarantee a fair distribution of gains, especially when disruption hits at a macro scale.
From my standpoint—rooted in law, community, and building practical tools—what matters is simple:
Progress is only progress if people can live in it.
Selected Citations
News / Reporting (Bluebook style)
- Anthropic CEO Says Government Should Help Ensure AI's Economic Upside Is Shared, WALL ST. J. (Jan. 2026).
- Anthropic CEO Dario Amodei Says AI Could Increase Both Economic Growth and Unemployment, PYMNTS (Jan. 20, 2026).
- Shashwat Chauhan, US Layoffs for October Surge to Two-Decade High, Challenger Data Shows, REUTERS (Nov. 6, 2025).
- Nov. 06 October Challenger Report: 153,074 Job Cuts on Cost-Cutting & AI, CHALLENGER, GRAY & CHRISTMAS, INC. (Nov. 6, 2025).
Cases (Bluebook)
- Goldberg v. Kelly, 397 U.S. 254 (1970).
- Mathews v. Eldridge, 424 U.S. 319 (1976).
- Cleveland Bd. of Educ. v. Loudermill, 470 U.S. 532 (1985).
- Brown Shoe Co. v. United States, 370 U.S. 294 (1962).
- United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001).
AI Disclaimer
This article was composed with the assistance of AI tools and has been reviewed by a human author for accuracy and coherence. Readers should conduct their own research and remain skeptical of potential factual errors.
Matthew A. Mishak, J.D.
Attorney, Municipal Law Director, and Founder of LegalTek.ai. Focused on making legal help more accessible through technology while protecting rights and dignity.