37. Automation and Jobs: What’s Likely Next

Automation has been changing work for decades, but the next phase is likely to feel different. Not because machines suddenly become “smarter” in a vague way, but because the scope of what they can do keeps widening across routine office tasks, customer interactions, scheduling, reporting, and parts of decision support. The immediate question for workers and employers is not whether jobs will disappear, it’s how job content will change, where new openings will appear, and which skills will actually transfer across environments.

The uncomfortable truth is that automation rarely delivers a clean trade: fewer people, same pay, better services. More often, it creates a shuffle. Some tasks shrink, others expand, and a different set of people becomes responsible for the remaining work, including oversight, exceptions, and quality. If you’ve worked near operations, you already know this. A system can automate the happy path and still leave the messy path to humans, and “messy path” tends to grow when you connect more systems together.

What’s likely next is less about a single wave of job loss and more about a steady recalculation of who does what, how quickly, and under what incentives.

What automation is really replacing

When people talk about automation, they often picture factories or robots. That’s understandable, but the next wave is at least as much about information work. Many roles are composed of a mix of activities: collecting data, interpreting it, communicating it, coordinating with other people or systems, and handling edge cases. Automation tends to target the predictable slices of that mix.

There are a few common targets:

First, repetitive processing. If a task involves extracting the same fields from the same kind of documents, checking straightforward rules, or producing a standard output, it becomes a prime candidate for automation.

Second, standardized interaction. If a customer request follows common patterns, routing and response can be automated, and only the exceptions get human attention.

Third, time-consuming coordination. Scheduling, invoicing, status updates, and handoffs across departments can be automated by linking systems that were previously separated.

Fourth, “first drafts” of knowledge work. Drafting summaries, generating routine reports, translating formats, and suggesting next actions are ways automation can reduce the time humans spend getting started, even if humans still do the final review.

The key detail is that automation rarely removes the need for judgment entirely. Instead, it moves the judgment upstream, downstream, or sideways. Upstream looks like better triage before work begins. Downstream looks like review, escalation, and remediation after automated steps. Sideways looks like coordinating with other teams to correct systemic issues that automation exposes.

You can see this in organizations where new tooling arrives with promises of “efficiency.” The first months often involve surprising volumes of exception handling, because the automation is now surfacing problems that were previously hidden in manual steps. That’s not a failure of automation, it’s an audit of reality.

The job market effect: displacement and reallocation

The simplest story people tell is “automation destroys jobs.” The more realistic story is “automation reassigns tasks across roles and changes the composition of employment.” You can get displacement in specific occupations without overall collapse in employment, especially if organizations redirect savings into new service capacity, new products, or entirely new workflows.

But there is a middle layer that matters for workers: the time horizon and the friction of transitions.

When automation is introduced gradually, workers have time to adjust. When it is introduced aggressively, there is a broader risk that some people fall into gaps between old job requirements and new ones. Those gaps are often about skill mismatch, location, seniority, and access to training. Even when someone can learn the new tools quickly, they might not have the chance, because companies hire around the systems they already plan to use.

In practical terms, the job market effect usually shows up in three ways:

1) Narrower roles. Some jobs become more specialized, focusing on exception management, quality control, or client-specific nuance. 2) Higher productivity expectations. Output per worker rises, and “average” performance thresholds shift. 3) More layered workflows. Tasks that used to be handled by one person split into multiple roles: data prep, automated processing, review, and final delivery.

Those changes can be good for customers and sometimes even for workers, but only if the reallocation includes new opportunities, not just intensified pressure.

Where demand may grow

Automation can reduce demand for certain tasks, but it also tends to increase demand for people who can build, maintain, govern, and audit automated systems. That includes technical roles, but also roles that blend technical knowledge with operational responsibility.

Here are some job categories that commonly see stronger demand as automation expands, though titles vary widely by industry:

Automation and workflow engineering roles, especially where systems span multiple departments
Data operations, quality assurance, and model or rule governance work
Cybersecurity and privacy specialists focused on automation risk
Customer support and operations roles that handle exceptions and complex cases
Domain experts who translate requirements into measurable outcomes

Notice what’s not on that list: “robot operators” as a mass category. In many settings, the work is too distributed for that. The future looks more like oversight and system stewardship than operating a machine in a single place.

The overlooked factor: the exception rate

People underestimate exception rate. In early discussions about automation, exceptions are treated as edge cases. In practice, exceptions can dominate workload when systems touch messy data, diverse customer needs, or ambiguous instructions.

The exception rate depends on three things:

Data quality: If inputs are incomplete or inconsistent, automation needs more human corrections.
Process clarity: If the workflow is poorly defined, automation exposes the gaps.
Governance maturity: If there are no good policies for when automation should stop and escalate, humans get pulled into decision loops.

This is why two companies deploying similar tools can experience very different labor outcomes. One firm might automate a narrow, well-governed workflow with low exception rates and reap productivity quickly. Another might roll automation into a broad, inconsistently applied process and end up creating a heavy “review tax.” That review tax can still be worthwhile, but it changes how many people are needed and how their time is structured.

If you’ve ever watched an organization implement a new system that “should reduce manual work,” only to discover that the old process used to absorb unclear requirements, you understand the pattern. Automation doesn’t just remove work. It reveals where work was previously hidden inside ad hoc decisions.

The likely next phase is that exception management becomes a formal discipline, not an improvised skill. Companies will invest in better workflow definitions, stronger feedback loops, and clearer escalation rules. That, in turn, creates stable demand for roles focused on quality and process control.

Skills: what transfers and what doesn’t

A lot of advice about automation and jobs focuses on learning new technical tools. That can be helpful, but it misses the deeper point: automation changes the value of certain cognitive skills.

Some skills tend to transfer well:

Structured thinking: Breaking messy problems into steps and inputs
Communication: Translating requirements into language machines and teams can follow
Monitoring and verification: Knowing what “good” looks like and how to measure it
Domain knowledge: Understanding customer context, compliance constraints, and real operational trade-offs
Adaptability in workflows: Learning new tools and changing routines without losing judgment

Other skills are more vulnerable. Any skill that’s essentially “remembering procedures” without understanding the underlying logic becomes less valuable when software can retrieve and follow procedures. Similarly, tasks that rely on speed in routine document formatting can shrink as tools get better at it.

Yet it’s not always a direct replacement. Sometimes automation changes where speed matters. For instance, a role might shift from producing first drafts quickly to validating outputs quickly, or from manual data entry to anomaly investigation.

There’s also a social side. Skills do not become valuable in isolation. They become valuable when organizations redesign job structures around them. Two people with similar skills can end up in different outcomes based on how their employer rewires work and who gets access to new assignments.

A short reality check on “reskilling”

Reskilling sounds straightforward, but in employment terms it’s a negotiation between the worker’s time, the employer’s staffing plans, and the credibility of the training signal. Some training helps immediately because it reduces friction on a current job. Other training helps only when an organization actually opens a pathway to new roles.

One practical way to judge what training will matter is to look at the tasks that are already changing. If your day-to-day work includes new automated steps, the training most likely to pay off is training that helps you supervise, correct, and improve those steps.

If your work is untouched, training might still help, but it may not convert into better opportunities quickly, because the employer may not need your new skills yet.

The likely timeline: uneven, not instant

Automation adoption is rarely uniform. Regulatory constraints, legacy systems, and budget cycles slow some changes. Other changes happen faster where tooling is modular and the business case is clear.

So what’s likely next is not a sudden cliff. It’s a series of uneven transitions across industries and geographies:

customer-facing functions adopt automation in front-of-house first, then expand into back-office
internal reporting and compliance-adjacent workflows often get automated early, because they have clear outputs
maintenance and operations can lag, because processes depend on physical context and on-site decision-making

Even within a company, some teams move quickly while others wait for stabilization. That means workers will experience automation not as a single event, but as a sequence of incremental changes to what they do each week.

The most important effect for workers is the acceleration of change. When software updates monthly instead of yearly, the learning curve becomes continuous. People who can keep learning and who have some control over their workflow adapt better than people whose role is rigid and whose time is already consumed by compliance or production quotas.

Why wages and job quality may matter more than job count

Job loss headlines miss another outcome: job quality. Automation can increase efficiency, but if productivity gains flow to owners while job conditions do not improve, worker bargaining power weakens.

There are at least three job-quality dimensions to watch:

1) Schedule control. If automation makes demand “smooth” across the day, some jobs become more predictable. If it enables constant availability, other jobs become less predictable, with more monitoring and shift variation. 2) Work intensity. Even if headcount stays stable, automated systems can raise expectations for speed and throughput. 3) Autonomy. Some workers lose control when decisions become standardized. Others gain control when automation handles the tedious parts and humans focus on judgment.

In my experience, the biggest friction appears when automation is implemented without redesigning escalation and workload distribution. When the system is “supposed to handle it,” but it can’t reliably handle edge cases, humans get overloaded with unresolved tickets. That is how automation can create a quality problem, not just an efficiency change.

The likely next phase in many organizations is more formal accountability: clearer economy data metrics for model performance, better audit trails, and more explicit responsibility boundaries between humans and automated systems. That is a governance improvement, but it also changes how performance is measured, and that can feel like pressure even when it’s fair.

A practical way to think about your next move

Workers cannot control corporate strategy, but they can make choices that reduce risk and increase leverage. Not every move requires leaving your current industry. Often, it’s about becoming the person who can operate in the “hybrid” space: understanding the work and the automation that touches it.

If you’re trying to plan around automation, a useful approach is to map your role into three categories: tasks that are being automated, tasks that will likely remain human, and tasks that are unclear today but becoming more important as systems expand.

Then you decide where to invest. That might be technical literacy, process documentation, or client communication. It might also mean seeking work that touches exception paths, because those are the places where automation creates demand for human judgment.

Here is a short set of moves many people can make without waiting for a perfect opportunity:

Ask where your workflow is being automated, then learn the failure modes that require human intervention
Build credibility by documenting processes, creating checklists, or improving accuracy metrics even if tools change
Offer to own a small improvement project tied to measurable outcomes like cycle time, error rates, or rework
Develop communication skills for stakeholder alignment, because automated workflows often break at handoffs
Use portfolio-style proof, such as case studies or examples of reduced errors, rather than vague claims of “I can learn”

That list is intentionally modest. It assumes the next job shift will be incremental and that you need evidence of competence, not just attendance in training sessions.

The employer side: what companies get wrong

It’s easy to blame workers, but automation adoption often fails because organizations treat it as a procurement task instead of a redesign task.

A few common mistakes show up repeatedly:

When companies implement automation without clean inputs, they create constant exception loops and burn out reviewers. When they measure success only by throughput, they ignore accuracy and downstream rework. When they remove roles without considering escalation, the “last mile” ends up unstaffed.

Another recurring problem is governance. If there is no clear policy for when automation can act, when it must ask a human, and how errors are corrected, then the system’s outputs become unpredictable, even if the automation itself is statistically strong.

The likely next evolution is that governance becomes part of the product. Organizations will demand auditability, monitoring, and faster feedback mechanisms. That raises demand for people who can design controls, create evaluation frameworks, and translate business risk into operational rules.

It also means that some employment risk shifts away from simple task performance and toward accountability roles. People who can handle responsibility for automated outcomes become more valuable, while people who perform “pure execution” with minimal judgment may face more narrowing.

Equity and geography: who benefits first

Automation does not land evenly. It benefits early adopters with the capital, data infrastructure, and talent to implement systems well. It can also benefit workers who already sit close to workflows where automation is applied, because they can shape how the tools are used.

Meanwhile, workers in sectors with limited digitization, weaker data foundations, or strict labor regulations may experience slower adoption. That doesn’t guarantee better outcomes. It can also mean fewer opportunities to transition into automation-adjacent roles, because the tools and the investments never arrive locally.

Over time, though, the pattern often reverses when supply chains and platforms push standardization. When a major customer demands automated reporting or compliance evidence, even slow adopters get pulled into the change. The job consequences then show up as accelerated process standardization, not only as layoffs.

In practical terms, geography still matters. Local labor markets influence how quickly workers can move to new roles, and commuting patterns influence the set of opportunities that are realistic. Even when a new job exists, relocation can be the hidden barrier.

What to watch next: signs the labor impact is stabilizing

You can learn a lot by watching how organizations talk about automation, but more importantly, by watching what they do operationally. The healthiest environments tend to show signals like these:

Performance metrics that include quality, not only speed Clear escalation paths and workload balancing for exceptions Training tied to the actual workflow changes, not just general “digital skills” A willingness to redesign roles so humans and machines complement each other

The unhealthy environments look different. They talk about “leaning out” headcount while leaving exception handling to whoever is still staffed. They celebrate early demos but avoid measuring real error rates over time. They treat adoption as a one-off deployment rather than a continuous improvement cycle.

Over the next few years, the organizations that handle automation well are likely to become the ones that professionalize hybrid work. That professionalization is where stable job opportunities can emerge, even as some tasks disappear.

The bottom line: expect change, but not a single fate

The likely next chapter of automation and jobs is not a prophecy of universal job loss or universal job growth. It’s a rearrangement of tasks, with winners and losers in the details.

Tasks that are routine, standardized, and well-instrumented will keep shrinking in human time. Tasks that require judgment under ambiguity, exception handling, and accountability will keep growing in importance. The middle ground, where humans supervise automated systems and improve the workflows around them, is likely to expand.

If you want a practical stance, it’s this: treat automation as an evolution of work design, not a sudden replacement of workers. Then build your career around economy the parts of the workflow that remain valuable when the automation gets it wrong, because those parts are the ones where judgment, communication, and operational understanding matter most.

And if you’re an employer, the most reliable strategy is to design for complementarity: automate the repeatable steps, invest in governance, and create real pathways for people to move into the roles that automation makes more central. That approach reduces harm, strengthens quality, and makes the labor story more sustainable.