Sponsored by
This Week's Big Story
The AI jobs debate usually asks which tasks software can do. A new lawsuit against Meta raises a different question: what happens when software helps judge the people doing the work?
Twenty-six Meta employees allege that AI-assisted systems helped select workers for the company’s May layoffs. Each plaintiff had taken protected leave or requested or received a disability accommodation, according to Associated Press reporting. Meta denies the claims, which have not been proven in court.
The dispute centers on context. Someone on parental or medical leave will naturally produce less activity while away. A scoring system may interpret that drop in production as lower performance unless the review accounts for why the numbers changed.
-Brandon S.
When the layoff list starts with a score, every missing piece of context counts.
The AI Agent You Can Trust
The best assistants don't multitask their attention across a hundred tools. Neither does Catch. It's an AI agent that focuses on one thing — the admin work you'd rather not touch — and does it exceptionally well.
Scheduling, flights, restaurants, follow-ups, vendors, clients. You hand it over; Catch handles the back-and-forth and comes back with it done.
No context-switching. No dropped balls. Just your admin, quietly cleared — so your focus stays on the work only you can do.
Meet the agent built for admin, and it'll be ready to work before your next meeting.
Get started at catchagent.ai — and give your attention back to what matters.
Notable Terms Used Here
Automated employment decision: A hiring, promotion, performance, or layoff decision that relies in part on software.
Protected leave: Time away from work that may be protected under laws covering family, medical, pregnancy, military, or disability-related needs.
Proxy: A measurable item used as a stand-in for something harder to measure. Keystrokes may be treated as a proxy for productivity even when they leave out much of the job.
Human review: A person checks the system’s recommendation and has the information and authority needed to change it.
📊 Key Numbers and Trends
26 employees: The number of plaintiffs in the Meta lawsuit.
8,000 planned layoffs: Meta’s May reduction covered about 10% of its workforce, according to Associated Press reporting.
4,800 Microsoft cuts: About 2.1% of Microsoft’s global workforce.
57,000 jobs added in June: Hiring remained positive, but the gain was small for an economy of this size.
4.2% unemployment: The overall labor market remains stable even as workers in particular companies and occupations face a harder search.
Microsoft framed its reductions as part of a broader transformation while continuing to invest in AI, leaving the direct role of AI in those cuts unclear. Taken together, the numbers describe a labor market where companies are cutting large groups of people while the rules for measuring work are changing, with no evidence of AI replacement across the American workforce as a whole.
How a Worker Becomes a Score
According to the Meta complaint, the company used activity data, productivity measures, performance rankings, and AI usage data in the layoff process. The allegation offers a useful example of how workplace data can move from a dashboard into a decision that affects a paycheck.
Companies already collect large amounts of employee data. AI gives management a faster way to sort it, rank it, and compare workers across a larger group.
Most workplace scoring systems follow a basic chain:
The company collects data. That may include sales, completed projects, customer ratings, attendance, software activity, manager reviews, or use of internal tools.
The system turns activity into signals. A model may identify patterns, assign weights, or compare one employee with a larger group.
The signals become a rank or recommendation. Managers may receive a score, a risk flag, or a suggested list of people for review.
A person makes the formal decision. Human approval can catch a bad recommendation, but only if the reviewer understands the data and has enough authority to challenge it.
A system may count what is convenient to measure while missing the reason behind it. Managers can make the same mistake, and AI allows it to move faster and reach more people if it isn’t scrutinized first.
The Four Layers
AI will reach each part of the economy differently. The assessments below describe the current direction of pressure rather than the specific odds of any one person losing a job.
L1: Natural Resources
This layer includes farms, fisheries, forestry, mining, water sources, and energy production. AI can help forecast demand, monitor crops and equipment, schedule crews, analyze geological data, and flag maintenance problems. Large farms, mines, forestry operations, and energy producers may need fewer people doing routine reporting, dispatch, and first-pass analysis as those systems improve.
Brace for: Smaller administrative teams, more centralized monitoring, and higher expectations for each worker to use sensor data and automated recommendations. Autonomous equipment may also reduce some driving and equipment roles at large, standardized sites.
Keep in perspective: Pumps still fail, weather changes, job sites move, and safety rules require people who understand the physical system. Repair work, field maintenance, emergency response, and local operating knowledge remain difficult to replace from a desk.
Best preparation: Add diagnostics, equipment maintenance, safety credentials, or the ability to check an automated recommendation against conditions in the field.
L2: Manufacturing & Construction
This layer covers processing, factory production, assembly, refining, enterprise production systems, and construction. Factories give AI and robotics a controlled environment. Scheduling, inventory planning, visual inspection, machine tending, drafting, and routine quality checks can all be compressed when the equipment and software work together. Construction is harder because every site has different conditions, crews, permits, and sequencing problems.
Brace for: Fewer routine planning and inspection roles, smaller production crews in highly automated plants, and less demand for workers who know only one repeatable task. Some reductions may arrive through attrition or smaller new-hire classes before they appear as a large layoff.
Keep in perspective: Retrofitting a plant costs money and creates downtime. Construction still requires people to fit, weld, wire, repair, and make judgment calls in changing conditions. A software demonstration does not remove those constraints.
Best preparation: Learn the whole process around the machine. Maintenance, controls, troubleshooting, quality ownership, and crew coordination offer more room than a single repetitive station.
L3: Retail, Services & Distribution
This layer includes retail, consumer services, healthcare delivery, finance and professional services, warehousing, trucking, finished-goods logistics, and last-mile delivery. Customer support, basic claims processing, scheduling, checkout, document handling, simple marketing work, and warehouse planning are easier to automate because the work already moves through software. Employers may also use AI to predict staffing needs and tighten schedules.
Brace for: Fewer entry-level office and professional-service openings, reduced call-center staffing, more self-service, and smaller support teams handling a larger number of customers. For hourly workers, the first effect may be fewer hours or unfilled vacancies rather than a formal layoff notice.
Keep in perspective: Customers still need help with unusual problems. Skilled sales, caregiving, repair, hospitality, licensed services, complex advice, and last-mile work depend on trust, physical presence, or judgment that a standard chatbot cannot provide consistently.
Best preparation: Move toward escalation, retention, complex customer problems, field service, or work tied directly to revenue. Learn the software well enough to supervise its output instead of competing with it on the easiest tasks.
L4: Management & Politics
This layer includes corporate strategy, management decisions, labor rules, regulation, tax and trade policy, and the political institutions that set the terms for the other layers. AI is inexpensive to deploy across reporting, presentations, scheduling, basic analysis, document review, and internal support. That gives employers a way to combine roles or ask a smaller management team to oversee more work. It also gives executives new tools for scoring workers when a reduction is already under consideration.
Brace for: Smaller corporate planning, HR, policy-support, and coordinator teams; wider spans of control for managers; and more measurement of response time, output, tool use, and other activity that is easy to count. New regulation may also require companies to document how automated employment decisions are made.
Keep in perspective: Executives, public officials, and regulated institutions still have to stand behind personnel decisions, legal approvals, policy choices, and major uses of capital. AI can prepare a recommendation without accepting the consequences.
Best preparation: Tie your work to judgment, accountability, risk reduction, or a decision you can defend. Keep a record of those results because an activity score may miss them.
What to Watch Next
The Meta case: Court filings may reveal what data was used, how much weight the software carried, and what human review occurred.
Employer disclosure: Watch whether companies begin telling workers when AI contributes to performance or layoff decisions.
Appeal rights: A real review process should let a worker correct bad data and add missing context before a decision becomes final.
Management language: Terms such as productivity, utilization, efficiency, and AI adoption can reveal which behaviors a company plans to measure more closely.
💡 Your Action Items
Even if you never see the model that helps evaluate your work, you can improve the record it reads and protect the context around it.
Keep your own record of results. Save performance reviews, completed projects, customer praise, sales results, and written feedback you are allowed to retain. Do not rely on a company dashboard to tell the whole story.
Document approved leave and accommodations. Keep copies of approvals, dates, and relevant communication somewhere you can access outside the company system.
Ask what went into the decision. If a review, warning, or layoff appears inconsistent with your work, request the criteria and correction process in writing. The company may limit what it shares, but the request creates a record.
Manage people by results. Small-business owners and managers should test whether the metric reflects the actual job. Do not use easy-to-count activity as a shortcut for useful work.
Look for the compliance opportunity. Entrepreneurs building for HR departments should pay attention to audit logs, appeals, bias testing, and documentation. Companies will need ways to explain how a recommendation was made.
If you believe protected leave, disability, pregnancy, or another protected status affected an employment decision, save the records and speak with a qualified employment attorney or the appropriate government agency.
Your Coalscoop-informed edge: AI layoff risk follows the structure of the work. Document the value you own which a dashboard misses, and move toward work where someone must own the result.
Thanks for reading. If you think others would find value in this perspective, please forward and help our community grow. And if you're someone who received this from a friend and would like to subscribe, visit coalscoop.com.
-Brandon S.
Sources
** Disclaimer **
Coalscoop is published by Firesteel Studios, LLC for informational and educational purposes only. I'm not a licensed financial advisor, investment professional, or attorney, and nothing here constitutes financial, investment, legal, or professional advice. By reading Coalscoop, you acknowledge that you're solely responsible for your own decisions and will not hold Coalscoop or Firesteel Studios, LLC liable for any losses or consequences arising from the use of this information.