EVERYONE IS LEARNING
A DIFFERENT AI
AI use is highly personal. Different tools and workflows produce very different ideas about what the technology can do.
Not all AI works the same way
Chatbots, image generators, workplace systems, and automated scoring tools create completely different experiences.
People solve different problems
One person writes emails. Another analyzes documents, builds spreadsheets, researches, or creates images and websites.
Experience shapes expectations
People who do not use AI may dismiss it. Experienced users may recognize capabilities others have never encountered.
THE TECHNOLOGY CHANGES FASTER
THAN THE RULEBOOK
AI can change substantially within a few major product or model updates. Protections built around one tool may become obsolete before they are implemented.
We prepare for one specific technology
Workplace rules focus on the AI products that exist today rather than the capabilities that may appear next.
Product-specific protections expire quickly
A new vendor, model, or system can perform the same function under a different name.
Policies name tools instead of uses
Rules prohibit one application but say nothing about automated scoring, monitoring, or decision-making generally.
Build technology-neutral rights
Focus on what systems do, how their conclusions are used, and what rights workers retain.
JOBS ARE NOT THE SAME
AS TASKS
AI does not need to replace an entire occupation to transform it. It can remove, accelerate, or redesign individual pieces of the job first.
One occupation contains many different tasks
AI can remove tasks without removing the worker
- Routine information retrieval becomes faster
- Repetitive documentation partially disappears
- Simple customer questions are handled automatically
- Monitoring and administrative work become automated
- Employees remain responsible for exceptions and difficult cases
A JOB CAN BEGIN DISAPPEARING
BEFORE ANYONE LOSES THEIR JOB
One of the most credible near-term risks is not mass replacement. It is gradual employment erosion through automation, attrition, work transfer, and jobs that are never created.
A worker is removed from an existing position
This is the most visible form of job loss—and the outcome people usually imagine when discussing workplace automation.
The next worker is never hired
Retirements, resignations, and vacancies gradually reduce employment without a formal replacement announcement.
THE FIRST TASKS AI TAKES MAY BE
THE ONES WE HATE
AI may first enter the workplace by removing repetitive, frustrating, or time-consuming work that employees are perfectly happy to surrender.
Employees welcome the relief
Searching policies, repetitive documentation, routine follow-up, and simple administrative work become easier.
Productivity rises quickly
Tasks that once required hours may take minutes while employees produce the same—or better—results.
The workplace changes quietly
The faster method becomes normal before workers discuss what happens to the saved time or the remaining work.
WORKERS CAN BRING AI
INTO THE WORKPLACE FIRST
AI adoption does not always begin with a corporate rollout. Employees can discover useful tools independently and change how work gets done from the bottom up.
The company deploys the technology
Management selects the tool, defines its purpose, provides training, and establishes formal rules for its use.
An employee discovers a faster method
Workers use approved or personal tools for writing, research, spreadsheets, summaries, or information retrieval.
The workplace changes before anyone notices
Informal use can spread quickly, creating productivity gains alongside privacy, security, and work-ownership questions.
A FOUR-HOUR TASK BECOMES
A THIRTY-MINUTE TASK
Imagine an employee who regularly reviews customer cases, recurring policy questions, transfer patterns, or irregular-operations notes and organizes them into a useful report.
Search, organize, compare, and write
- Review notes and individual cases
- Identify recurring themes manually
- Organize findings into categories
- Build a table or summary
- Draft and revise the final report
An approved tool organizes the work
- Analyze appropriately protected or anonymized information
- Group cases by recurring issue
- Identify patterns and exceptions
- Create a structured table
- Draft a report for employee review
Who gets the saved time?
Does the employee retain it, surrender hours, or receive more work?
Does speed become expected?
Does a four-hour assignment permanently become a thirty-minute standard?
Who owns the workflow?
Can the employee’s method be copied, centralized, or transferred?
What happens next?
Does the tool assist the worker—or eventually remove the task?
WHO OWNS THE
PRODUCTIVITY DIVIDEND?
If AI turns eight hours of work into five, the technology does not determine what happens to the remaining three hours. Workplace decisions do.
The worker keeps the saved time
Paid hours remain stable while employees gain breathing room, flexibility, or time for more difficult cases.
The worker gives the time back
Leaving early feels beneficial in the moment, but employees lose paid hours and shorter days may become normalized.
Management fills every saved minute
The faster workflow becomes the new baseline, and employees receive more cases, more tasks, or broader responsibilities.
WORKERS CAN GRADUALLY
FURLOUGH THEMSELVES
Leaving early can be a completely reasonable individual choice. The risk emerges when reduced paid hours become normal and begin shaping future staffing decisions.
Going home early feels like a benefit
On any single day, an employee may reasonably prefer additional personal time over remaining at work after the workload disappears.
Reduced hours can become the new baseline
Management may eventually plan staffing around the assumption that fewer paid hours are required to perform the same work.
EASIER WORK CAN BECOME
MORE WORK
AI may genuinely reduce effort and frustration. The risk is that every productivity gain becomes a new minimum expectation.
Less recovery time
Every saved minute is immediately assigned another task.
More simultaneous work
Employees handle additional cases, channels, or responsibilities.
Tighter staffing
Productivity gains are used to justify fewer employees or hours.
Higher pressure
Temporary gains become permanent performance standards.
A HELPFUL TOOL CAN BECOME
A MANAGEMENT TOOL
An AI system may be introduced for a harmless or useful purpose. The risk comes from what the same system—and the data it collects—can later be used to do.
Why the tool was introduced
Improve safety, reduce repetitive work, answer questions faster, support customers, or make difficult work easier.
The same data creates new information
Management can learn how long tasks take, how employees differ, which rules are bypassed, and who appears unusual.
Someone discovers another use
Coaching information can become ranking, monitoring can become discipline, and helpful data can become evidence.
AI MAY NOT TAKE THE AGENT’S HEADSET.
IT MAY SIT BEHIND THE SUPERVISOR.
The more immediate workplace threat may be automated monitoring, interpretation, scoring, coaching recommendations, and disciplinary evidence.
More of the work becomes measurable
Calls, handle times, wording, tone, schedules, transfers, transactions, and system activity can become data.
AI decides what the data may mean
Software can classify behavior, identify patterns, compare employees, detect deviations, and flag activity that appears unusual.
A score can become an employment decision
Automated findings can influence coaching, scheduling, investigations, performance reviews, or discipline.
PERFECT ENFORCEMENT CAN
MAKE THE WORKPLACE WORSE
An AI system can accurately detect every deviation from a rule while making employees less willing to exercise the reasonable discretion that allows the operation to function.
How policies assume the work occurs
Every procedure is followed exactly, every situation fits the written rule, and strict compliance always produces the intended result.
How employees make the operation function
Real work requires judgment, adaptation, context, informal coordination, and reasonable responses to situations the rulebook did not anticipate.
THE SYSTEM IS RIGHT.
THE WORKPLACE GETS WORSE.
An agent makes a harmless or contextually necessary departure from a rigid process to resolve a customer problem and prevent another transfer.
More transfers
Employees avoid resolving cases that require judgment.
Longer handle times
Minor situations require additional approvals and escalation.
Less discretion
Employees protect themselves by following every rule literally.
Worse service
The procedure becomes more important than resolving the problem.
MANAGERIAL FEAR CAN
FLOW DOWNHILL
Supervisors and lower-level managers may also face automation, continuous measurement, performance comparisons, and pressure to justify their own positions.
Pressure moves to employees
Managers protect their own performance by demanding tighter compliance and higher output below them.
Workers become easier to measure
Employee behavior becomes highly visible while managerial judgment and system design remain difficult to examine.
Avoiding blame replaces judgment
Supervisors enforce automated findings aggressively because disregarding them creates personal risk.
“The system said so”
Automated recommendations provide cover for decisions while making responsibility harder to locate.
WHILE MANAGEMENT BECOMES EVERYWHERE.
AI can reduce layers of supervision while increasing the amount of monitoring, comparison, prediction, and automated control employees experience.
Limited reach
A human supervisor can observe only so many employees, review only so many calls, and investigate only so many events.
Human management has limits because human attention has limits.
Massive reach
An AI system can potentially review every interaction, compare thousands of employees, and apply the same rule continuously.
Management can become more pervasive even while fewer people perform it.
WHAT IF ANOTHER AIRLINE
GETS BETTER AT AI FIRST?
Doing nothing also creates risk. If another carrier learns to use AI more effectively, competitive pressure could eventually threaten bargaining-unit employment.
Faster disruption recovery
Better information and decision support could help an airline recover from irregular operations more quickly and consistently.
Better tools and fewer handoffs
Employees with stronger information access may resolve more complicated problems without unnecessary transfers or delays.
The same workforce becomes more capable
AI literacy may allow employees to recognize useful applications and improve workflows faster than a centralized project can.
AI CAN AUTOMATE BUREAUCRACY
INSTEAD OF WORKER AUTONOMY
AI does not have to create a more authoritarian workplace. It can give frontline employees better information, stronger decision support, and more authority to resolve problems.
AI concentrates authority
- Employees are continuously scored
- Automated findings drive discipline
- Decision-making moves into software
- Supervisors enforce system recommendations
- Reasonable discretion contracts
AI distributes useful authority
- Employees receive better information
- Routine bureaucracy is automated
- Frontline decisions become faster
- Supervisors coach and handle exceptions
- Responsible discretion expands
FEWER MANAGERS.
Here, “more management” means better coordination, information, training, and support—not more surveillance or control.