> ## Content Index
> Fetch the complete content index at: https://modalpathethics.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Applied Case: The AI Layoffs
- URL: https://modalpathethics.com/applied-case-the-ai-layoffs/
- Published: 2026-11-10T12:00:00.000Z
- Updated: 2026-11-10T17:16:21.000Z
- Description: Less work should mean more life available to live. That requires an actual arrangement connecting the saved labor to the people released from it.
- Author: Aidan Lawson
- Tags: Applied Case, Modal Systems, Business, Modal Path Ethics

## The Layoffs Can Work.

On February 26, 2026, Jack Dorsey announced that Block would cut more than 4,000 people, taking its workforce from more than 10,000 to fewer than 6,000\. He attributed the change to the combination of increasingly capable intelligence tools and smaller, flatter teams. [He also said the business was healthy.](https://www.sec.gov/Archives/edgar/data/1512673/000119312526076557/d108590dex991.htm?ref=modalpathethics.com) 

By November, this was no longer just a prediction about what artificial intelligence might eventually accomplish.

Block’s [third-quarter filing says](https://www.sec.gov/Archives/edgar/data/1512673/000162828026069524/xyz-20260930.htm?ref=modalpathethics.com) the workforce plan concluded in the second quarter. It recorded roughly $495 million in restructuring charges and continued to expect $800 million to $900 million in annualized employee-compensation savings. Savings realized through September were consistent with those expectations. 

The [accompanying shareholder letter](https://www.sec.gov/Archives/edgar/data/1512673/000162828026069513/q32026shareholderletter.htm?ref=modalpathethics.com) reported third-quarter gross profit of around $3.24 billion, up 22 percent year over year. 

The results do not isolate how much improvement came from AI, layoffs, product growth, or other changes. They do establish something the comforting story about incompetent executives cannot accommodate: a company can dismiss thousands of people, continue functioning, and report substantial savings.

On November 4, Block notified more employees of job cuts. [Business Insider’s account](https://www.businessinsider.com/jack-dorsey-block-layoffs-second-round-job-cuts-cash-app-2026-11?ref=modalpathethics.com) of the internal message identified customer operations, risk operations, and learning and content teams. The number affected was not established. This message referred to *better tools* and *changing work* without explicitly attributing the new cuts to AI. 

Keep the two rounds distinct and the company’s gains in view. Then examine what changed on the other side of them.

Before termination, an employment relationship repeatedly connected a person’s work to compensation, colleagues, equipment, organizational knowledge, and further work. A separation package carries some support beyond the ending, but does not continue the same relationship indefinitely. The person must establish another arrangement, draw on other resources, or reduce what they can maintain.

Some will move rapidly into better positions. Others will not. Block’s financial statements do not follow them far enough to tell us which happened to whom.

This is where an analysis of the AI layoffs has to begin. With the actual change, including the possibility that the technology works.

---

## A Smaller Payroll: Not Necessarily Less Work.

Commonwealth Bank gives us a different sequence.

In July 2025, it announced the elimination of 45 customer-service roles as it introduced an AI voice bot. Workers subsequently reported rising call volumes, overtime, and team leaders returning to the phones. The Finance Sector Union challenged the redundancies. In August, [the bank admitted](https://www.abc.net.au/news/2025-08-21/cba-backtracks-on-ai-job-cuts-as-chatbot-lifts-call-volumes/105679492?ref=modalpathethics.com) that its assessment had been inadequate and withdrew the decisions. Affected employees were offered the choice of remaining, redeployment, or departure. 

Here, the work had not disappeared in the way the staffing decision assumed.

The disputed employment decisions changed. But an offer to stay does not make the preceding uncertainty or decisions about departure unhappen. Nor does it establish that every affected employee returned to the same situation.

Syndio supplies another direct account, this time from its chief executive.

In May 2026, the company eliminated the position of Jonathan Vidales, a labor economist who had worked there for five years. It also expanded other employees’ responsibilities and introduced quotas. Maria Colacurcio [later explained](https://mariacolacurcio210230.substack.com/p/we-laid-him-off-then-we-hired-him) that management expected people to accept the changes because employment conditions favored employers. Several resigned. In August, Vidales applied for a newly opened role and was rehired. 

In [*Field Instruments: The Firm*](https://modalpathethics.com/field-instruments-the-firm/), Modal Path Ethics examined why management’s account of an organization cannot substitute for everything its members know. A role description records assigned responsibilities. It does not inventory every working relationship, practical judgment, accumulated workaround, or ability to recognize a problem before it becomes expensive. 

This does not make experienced employees irreplaceable. 

It establishes what replacement must actually replace.

A competent transition can remove obsolete tasks while carrying useful knowledge into different work. An incompetent one dismisses the people, discovers which functions had been lodged in their activity, and pays to reconstruct the connection. Customers and remaining employees carry the interval.

CBA demonstrates a staffing assessment contradicted by the workload. Syndio demonstrates a reorganization that underestimated existing people and their willingness to remain. Neither establishes that every AI-related layoff will fail.

---

## Follow the Payroll Outside the Company.

Klarna’s famous **700-worker** figure requires particular care because it has become a portable story about almost anything people want AI to mean.

Its S[eptember 2025 prospectus](https://www.sec.gov/Archives/edgar/data/2003292/000200329225000052/klarnagroupplc424b4.htm?ref=modalpathethics.com) describes the figure as a full-time-equivalent workload estimate, calculated from reductions in conversations handled by human agents after its assistant launched. The same filing reports roughly $39 million in 2024 savings, an AI share of 69 percent of customer-service chats over the twelve months to June 2025, and continued human support. Customer-satisfaction parity is the company’s reported result, not an independent finding of this article. 

In a [2024 CBS interview](https://www.cbsnews.com/news/klarna-ceo-ai-chatbot-replacing-workers-sebastian-siemiatkowski/?ref=modalpathethics.com), Sebastian Siemiatkowski identified the affected service labor as outsourced and distinguished it from earlier employee layoffs. 

So there is no basis for converting the workload estimate into a roster of 700 Klarna employees fired by a chatbot.

There is equally no basis for deciding that no workers could have been affected because they were outside Klarna’s payroll.

Klarna buys customer-service labor from another organization. Its need for that purchased labor declines. The next questions concern that supplier’s staffing, paid hours, contracts, recruitment, and redeployment. Klarna’s internal headcount cannot answer them.

The company boundary is not the boundary of the people doing the work.

The subsequent human-service story also needs correcting. When the [Securities and Exchange Commission asked](https://www.sec.gov/Archives/edgar/data/2003292/000162828025034998/filename1.htm?ref=modalpathethics.com) about reporting that the chief executive had acknowledged lower-quality AI service, Klarna’s lawyers disputed that interpretation. They said his comments concerned improving human service through a skilled freelance-consultant pilot, rather than an admission that AI service had deteriorated. 

That is the company’s rebuttal, not a reason to declare every criticism false. 

It does mean that the familiar story:

- **fire 700 humans,**  
  - *discover the chatbot cannot work,*  
    - **hire the humans back**

is not established by this record.

A mixed service can retain human specialists because some situations require them while automating a large volume of other work successfully. Its economic effects could still be substantial. The right investigation follows the actual changes in purchased labor and service quality.

Atlassian illustrates another distinction. In March 2026, [it announced](https://www.sec.gov/Archives/edgar/data/1650372/000165037226000021/ex991blogpostdatedmarch112.htm?ref=modalpathethics.com) around 1,600 job cuts, describing the restructuring as a way to fund further AI and enterprise-sales investment, accelerate profitability, and change its skill mix. Its announced support included at least sixteen weeks of separation pay and extended health coverage for eligible employees. 

That is partly an investment decision. A worker can lose a position because money is being redirected toward an AI strategy before any model performs that worker’s job.

- Calling this *“not really AI”* would conceal the causal role of the investment program.
- Calling it demonstrated technological replacement would claim a substitution the announcement does not establish.

The practical question remains concrete: 

> Which existing activity was stopped, where did its resources go, what new activity became possible, and what happened to the people whose support was withdrawn?

---

## You Can Keep the Job and Lose the Raise.

Gizmodo’s [November 6 wage argument](https://gizmodo.com/the-new-york-times-doesnt-understand-ais-role-in-suppressing-wages-2000836392/?ref=modalpathethics.com) concerns **suppression before replacement**. It highlighted David Kelly’s CNBC remarks in September. 

Kelly was speaking as JPMorgan Asset Management’s chief global strategist, not as a chief executive. His own published analysis supplies a stronger source than an argument about a television clip.

In April, he identified fear of AI-related job loss as a possible restraint on wage demands. In July, he argued that the threat of AI replacement could succeed the threat of foreign-worker replacement in weakening compensation’s share of national income, [thereby boosting corporate profits.](https://am.jpmorgan.com/us/en/asset-management/institutional/insights/market-insights/market-updates/notes-on-the-week-ahead/ai-inflation-and-interest-rates/?ref=modalpathethics.com)

This mechanism does not require a machine capable of performing an entire occupation.

Consider the actual negotiation being changed. 

- A worker previously expected that leaving would cost the employer enough to make a raise worth considering.  
  - The employer now has more ways to reorganize the work: partial automation, redistributed tasks, a smaller team, or a different combination of skills.
  - The worker also sees fewer attractive alternatives elsewhere.  
    - Refusing an offer becomes more dangerous;
    - losing the worker becomes less dangerous.

The job can remain occupied while the terms degrade.

This is why *“AI has not caused mass unemployment”* cannot settle it. The employment count does not tell us how much work the employee must supply, how much income the relationship carries, or how costly refusal has become.

There is evidence against treating this as one uniform effect. A Boston Federal Reserve study [published in September](https://www.bostonfed.org/publications/current-policy-perspectives/2026/workers-perspectives-on-ai.aspx?ref=modalpathethics.com), using worker surveys from December 2024 and December 2025, found that respondents reporting the strongest AI-related productivity gains also felt more secure and were more willing to seek raises. Workers with middling perceived gains were more exposed to job anxiety and reported weaker negotiating positions. These are survey associations, not a measured national wage effect.

That difference is central.

AI can make an employee more useful and less dependent on one employer. It can also make the employer less dependent on that employee. Sometimes it does both. The direction of bargaining power depends on which alternatives become practical for each side.

A tool that helps someone complete valuable work, demonstrate competence, or start an independent operation can improve their position. The same tool, made available to every employer while access to customers and income remains concentrated, can weaken the price they obtain for doing that work.

Greater capability does not automatically produce greater bargaining power.

Nor is every disappointing raise proof of intimidation. To establish wage suppression in a particular workplace, examine the pay changes, workload, rejected demands, available jobs, and actual use of replacement threats. Kelly’s forecast identifies a real mechanism to investigate. It does not provide a confession from every employer in America.

But it is real enough to defeat another shortcut: **employment can survive after the employee’s ability to negotiate its terms has weakened**.

---

## Wage Stagnation Is a Change in What Work Will Carry.

The national figures also need their correct objects.

[The Bureau of Labor Statistics reported](https://www.bls.gov/news.release/archives/realer%5F10142026.htm?ref=modalpathethics.com) that real average hourly earnings fell 0.7 percent between September 2025 and September 2026\. Real weekly earnings fell 0.1 percent; a longer average workweek offset part of the hourly decline. These are averages for employees on private nonfarm payrolls, not a history of every worker’s income, and they do not identify AI as the cause. 

- More dollars can buy less.
- More hours can keep weekly purchasing power nearly steady while consuming more of the worker’s life.

Neither change requires a layoff notice.

[Kelly’s July analysis](https://am.jpmorgan.com/us/en/asset-management/adv/insights/market-insights/market-updates/notes-on-the-week-ahead/a-top-down-view-of-the-earnings-surge/?ref=modalpathethics.com) adds a longer distributional comparison: compensation accounted for an average 55.7 percent of GDP in the decade ending in 1994, against 50.5 percent in the first quarter of 2026\. Compensation includes benefits. A declining share is not the same claim as “nobody’s real wage increased for thirty years.” 

The question is what happened to the connection between productive activity and the lives it supports. If output expands while a worker’s purchasing power, time, or security deteriorates, the expanded output has not repaired that worker’s deterioration by existing elsewhere.

There may be other benefits: cheaper products, new investment, pension returns, useful services, or public revenue. Those should be followed too. An investor is a person; a customer paying less is too. Their gains do not disappear because the worker’s loss is severe.

But a higher share price does not pay a dismissed employee’s rent unless some actual arrangement connects that gain to the payment. A cheaper product may help that employee, but the size and timing of the benefit have to meet the size and timing of the loss. A promised economy full of new jobs does not carry an intervening month without income.

This is the error in letting national growth finish an individual transition.

The reverse error is to demand that every productivity gain become a raise within the old job. Reduced prices, shorter hours, safer work, new enterprises, and better services are also ways of improving people’s lives. Some productive tasks should disappear altogether.

**The relevant achievement is an improvement in what people can actually sustain and do, not preservation of one particular payment arrangement forever.**

That makes wage stagnation more serious than a disappointing number and more specific than a general accusation against technology. It concerns the continuing capacity of work to support housing, care, rest, participation, and the ability to refuse bad conditions.

---

## The First Job Disappears Before Anyone Can Lose It.

In [*Modal Path Ethics Apologizes to the Gamers*](https://modalpathethics.com/modal-path-ethics-apologizes-to-the-gamers/), I corrected an earlier demand that critics identify the worker who had been fired. That was too narrow. Hiring, commissions, and training routes can deteriorate without any existing employee receiving a termination notice. 

The correction matters directly here.

Stanford’s August 2026 revision of [*Canaries in the Coal Mine?*](https://digitaleconomy.stanford.edu/news/canariesaug26/?ref=modalpathethics.com) reported that employment among 22-to-25-year-olds in highly AI-exposed occupations stood about 19 percent below the level implied by keeping pace with less-exposed peers. The data extended through June. The adjustment appeared to operate primarily through reduced hiring, rather than increased separations. The authors explicitly describe this as a pattern, not an identified causal effect of AI; education, earlier trends, and sample differences remain important. 

This is not a finding that *AI fired 19 percent of young workers.* It is evidence of a narrowing entry route that an exclusive focus on **layoffs** would miss.

[The October dashboard](https://digitaleconomy.stanford.edu/project/indicators/canaries-dashboard/?ref=modalpathethics.com) continues to track the age differences. The finding remains descriptive; its importance is that the entry route must be examined separately from the jobs held by people already inside. 

The harmed object does not have to be an imaginary employee whom one small studio supposedly owed a salary. 

Actual people have trained, acquired qualifications, sought commissions, applied for positions, and organized their lives around an existing route into work. Evidence that this route is deteriorating concerns those people and those arrangements.

No individual employer owes every applicant a job. That does not make the collective disappearance of entry opportunities harmless.

There is a further loss when entry-level employment carried training. A first job can supply supervised practice, access to equipment, contact with customers, opportunities to make recoverable mistakes, and someone experienced enough to explain why a plausible answer is wrong. Removing low-level production tasks may also remove the arrangement through which beginners become capable of higher-level work.

The remaining senior employee can look extremely productive while the organization stops reproducing the competence it relies upon.

AI does not force that result. In the published [*Generative AI at Work*](https://oup.silverchair-cdn.com/article-minimal/7990658?ref=modalpathethics.com) study, an assistant increased customer-support productivity by about 15 percent on average, with larger gains among less experienced and lower-skilled workers. The most experienced workers saw little productivity benefit and some deterioration in quality. These were uneven effects of assistance during human work, not evidence that eliminating the beginners was necessary. 

The implication is practical: use the technology to make entry and learning cheaper where it can. Do not automatically convert every reduction in beginner task time into a reduction in beginner positions.

Nor should obsolete busywork be preserved as an initiation ritual. A training route can change radically. It still needs time, access, feedback, and real opportunities to exercise the new competence. An online course without a route into paid practice has not replaced a paid apprenticeship.

The answer to “where will the next experienced workers come from?” is not necessarily “the old junior jobs.” 

It is **an actual maintained process through which new people acquire, test, and use the relevant capacities**. 

If that process changes, show its replacement operating.

---

## BioWare Was Already Making Something.

BioWare’s [October 21 closure announcement](https://blog.bioware.com/2026/10/21/bioware-update/?ref=modalpathethics.com) said that early development on the next *Mass Effect* was ending. EA placed the decision within a wider concentration of investment and priorities. These announcements do not establish that generative AI replaced the development team. 

The case still belongs here because it exposes what a count of released products misses.

The canceled game was unfinished. We cannot announce that EA destroyed a guaranteed masterpiece, assign revenues to a sequel nobody had ever played, and use our imagined success to prove the decision irrational.

But the development organization was not imaginary.

People were already working together. The project already connected their skills, accumulated decisions, resources, and coordinated activity. Ending it changed what that particular organization could continue doing. The fact that the next game had not become a purchasable object does not mean nothing existed to be damaged.

[*Modal Path Ethics Gets a Paper Route*](https://modalpathethics.com/modal-path-ethics-gets-a-paper-route/) makes this distinction through an actual newsroom: an organization can lose the relations supporting its activity while its people, records, and earlier outputs remain. We do not need to invent tomorrow’s brilliant investigation to identify today’s broken publishing arrangement.

The same discipline applies to the studio. The injury is not the absence of any imagined game. It is the interruption of an existing capacity to make one.

EA’s continued control of *Mass Effect* does not itself preserve that capacity. A franchise right can survive after the team carrying the work has dispersed. Conversely, developers can continue valuable work elsewhere without carrying that franchise with them. These are distinct survivals. 

This also answers the proposal to save the project by licensing it to another studio, briefly discussed in the Gamers apology. A license could reopen something important. It would still need to connect to financing, a willing team, usable production material, development time, and a route to an audience. 

A worker-owned successor would face the same material requirements. Different ownership can change whose purposes direct the work, but sadly it does not summon a production budget or give workers rights to intellectual property they do not control.

And the developers are not assets belonging to the fans. They may want new projects. They may be exhausted by the old one. Their continued creative lives deserve protection without requiring them to spend those lives completing a preferred sequel.

There is a real conflict here. Continuing a project consumes resources that could support other work. A business can have good reasons to stop. Recognizing the harm does not establish an unlimited duty to fund the game.

It establishes what the decision must reckon with: an existing team and activity, the cost of continuation, the cost of dispersal, and whatever workable handoff remains. “The brand survives” does not answer that account. Neither does “the fans are upset.”

A company should be allowed to end a project. But it does not get to redefine the project as though it never contained anything worth losing.

---

## The Artists Are Here. That Does Not Finish the Labor Investigation.

[*The Artists Are Still Here, You Fools*](https://modalpathethics.com/the-artists-are-still-here-you-complete-idiots/) addressed a particularly bad habit: encountering AI-assisted work and announcing that somebody should have hired a human, in spite of the humans already working on it. 

The Ivy said its in-house designers made its disputed illustrations using AI. Christopher Fryant publicly identified himself as an artist on Coca-Cola’s 2025 holiday advertising. Silverside’s account of its own Coca-Cola production describes a five-person core team working with more than twenty additional professionals.

These are actual people performing actual work. Disliking the instrument cannot remove them from the account.

And their presence cannot settle the whole employment effect either.

A production can employ artists while purchasing fewer days of work from other artists. It can retain an art director while reducing commissions. It can create a new kind of job while making an older occupation harder to enter. It can also produce something that would not have been financed at all under a more expensive process.

Those possibilities require different comparisons. The relevant evidence is the actual production history: budgets, paid hours, commissions, rates, new purchases, canceled purchases, and who obtained control over the result. The existence of a generated image supplies none of those figures by itself.

There is also a wider substitution that the phrase **“AI can do your job”** conceals. A generated sequence can reduce demand for work surrounding a physical shoot without performing the many individual skills of every person who would work on that shoot. Employment changes through the choice of production process, not only through a machine imitating each occupation separately.

On the other side, consider a real life human designer using the new instrument. Tomasz Opasinski’s documented Kinoteka workflow involved hours of exploration, hundreds of generated images, client selection, and Photoshop production work. The presence of a generator did not eliminate his contribution.

If such tools let a small team complete work it could not previously undertake, an actual productive capacity has opened. If they let a creator earn enough from a small commission to continue making work, that matters too. It is no less real because the creator has entered somebody else’s professional market.

An established artist does not own every future commission in their preferred medium. 

An aspiring artist does not have to prove that established professionals approve of them before their own creative activity counts.

The impartial account must therefore retain both kinds of person: the worker losing a supported practice and the person gaining a practical means to begin one. 

We cannot defend the first by making the second disappear. We cannot celebrate the second by pretending the first’s damaged livelihood has already been repaired.

More people able to make images and fewer people able to earn a living from particular image-making services can be true at the same time.

There is no contradiction to dissolve. There is just a transition to arrange.

Recognizing the artists also leaves training-material disputes intact. An actual human creator can use a tool whose makers owe answers about access, permission, and the use of other people’s work. *Human participation* does not settle those claims. Conversely, a grievance about particular material does not give a profession ownership of every future image, or the power to prohibit newcomers from using an unfamiliar method. 

Protect the demonstrated claim without inventing an even broader monopoly around it.

## The Human Heart Still Needs a Production Budget.

The film articles opened the same problem from another direction.

In [*MARKY//MARK*](https://modalpathethics.com/mark-wahlberg-vs-mark-wahlberg/), James Gray’s defense of expressive filmmaking sat beside Ava DuVernay’s interest in alternative financing and distribution, and Mike Leigh’s concern about financiers controlling young filmmakers’ decisions. Those are three different requirements: 

- something to express,
- the means to make it,
- and enough authority to decide what it becomes.

[Leigh’s October interview](https://www.screendaily.com/news/mike-leigh-is-very-concerned-about-lack-of-creative-freedom-for-young-filmmakers-in-uk/5221539.article?ref=modalpathethics.com) concerned restrictions on creative freedom. DuVernay proposed more direct relationships between artists, financing, and audiences. Neither problem disappears when image generation becomes cheaper. 

The person may still be perfectly capable of writing a beautiful film and unable to obtain the resources to produce it. They may obtain those resources only by surrendering the choices that made the project worth pursuing. Or they may complete the film and have no effective way to reach an audience.

A claim that machines *lack the human heart* answers none of those obstacles.

A claim that AI *democratizes filmmaking* does not answer them either. 

Lowering production costs is a real opening when it brings a project within reach. But a completed file is not an audience, a payment, or durable access to the tools needed for the next project.

The imitation problem raised in [*SLIME WATCH IV*](https://modalpathethics.com/slime-watch-iv-i-just-apologized-to-the-gamers/) belongs here too. Making production cheaper can also make competing copies cheaper. Protecting a creator therefore requires distinguishing 

- **demonstrable copying or false attribution**

from 

- ***somebody less well-known making a work in the same genre.***

Neither flooding a market with appropriated work nor granting an incumbent control over everybody else’s creative methods preserves an open route into production.

The practical question is whether creators acquire a more complete route from 

- intention  
  - to  
    - finished work  
      - to  
        - circulation and continued activity.

Cheap generation controlled by a platform can open one part of that route while making the rest more dependent on the platform. 

Tools available to independent creators, combined with usable distribution and financing, can move in the opposite direction.

That distinction also answers the first-draft argument examined in [*Russell T Davies and the Writer Who Was Always There*](https://modalpathethics.com/russell-t-davies-and-the-writer-who-was-there-the-whole-time/). A writer does not disappear because drafting becomes faster or involves a new instrument. Selection, revision, dramatic judgment, collaboration, and responsibility for the finished work can remain substantial human activities.

But *“the writer is still there”* cannot establish that the employment arrangement remains intact. If a company commissions fewer writers, pays less, removes development time, or eliminates the beginning writer’s paid practice, those changes need their own judgment. Preserving authorship as a description does not preserve writing as an available livelihood.

The two claims fit together without compromise: 

- recognize the person who made the work,
- then investigate what the new production arrangement did to other people’s work.

[Wahlberg welcomed the possibility](https://techcrunch.com/2026/10/16/mark-wahlberg-on-using-ai-in-film-the-right-way-and-being-open-to-an-ai-clone-carrying-on-his-legacy/?ref=modalpathethics.com) of his children or grandchildren controlling a future digital likeness. 

His permission would certainly matter. It would not settle every other labor question. A licensed likeness might provide income to one family while changing demand for new performances, casting, or surrounding production work. Authorization to use an image and the distribution of the resulting gains are separate matters.

Nor does a famous performer’s negotiating position establish the conditions under which a less powerful performer agrees. Consent has to be examined through the actual uses permitted, their duration, compensation, alternatives, and what the agreement allows the performer to refuse later. An agreement to one use should not quietly become permanent control over every future use.

The fully autonomous and extremely broad **Mark 2** in *MARKY//MARK* was fictional. Its further question of what claims a genuinely independent synthetic performer might acquire cannot be settled by treating current digital replicas as demonstrated persons. It also cannot be settled forever by an original actor’s property claim. That prospect remains distinct from the present, inspectable agreements governing people’s likenesses and work. 

For the current labor argument, the answer is already substantial: **creative capability, recognition, consent, income, and control can move in different directions during the same production change**. 

Defending one does not excuse ignoring the others.

---

## Less Labor: an Achievement Worth Keeping.

At this point, a familiar escape route presents itself.

> Make the old production process compulsory. 

Keep every task. Preserve every position. Require the commission, the department, the shoot, and the first draft to be purchased in the old way.

That would protect some existing arrangements. 

It would also close real possibilities for the people who could not afford them.

The novice support workers in the productivity study and the designer in the documented production workflow were exercising actual capacities with assistance. Requiring them to use a slower process would not create additional humanity in their work. 

The same standard applies when a developer completes a previously unaffordable project or an assistive tool removes a disabling production obstacle. A demonstrated opening for that person belongs in the account. 

**Their earlier exclusion is not something to preserve so an incumbent profession can retain its preferred production method.** 

Customers who gain usable, affordable services belong in the account too.

The point of technological improvement cannot be to maintain the maximum possible amount of compulsory labor. Time spent overcoming an unnecessary obstacle is not automatically more dignified than time freed for something else. We should welcome the disappearance of dangerous, exhausting, repetitive, or unwanted tasks when their useful functions are genuinely preserved.

But 

- removing a task,
- removing a wage, and
- releasing a person into a more capable life

are three different changes.

Only the first follows directly from successful automation.

The second follows from an employment and ownership arrangement. The third requires that the person actually obtain usable time, resources, access, and opportunities after the first two changes.

Without that connection, “freeing workers” can describe people who have more uncommitted hours and less ability to use them. Their time has been released from the employer’s schedule while their livelihood has been released from the employer’s budget. Whether that is *liberation* depends on what now carries their lives.

- Keeping unnecessary work as the price of survival is a bad arrangement.
- Removing the work while retaining that price is another bad arrangement.

We do not need to choose between them as permanent civilizational destinations.

---

## A Transition Has to Reach Its Next Step.

[Salesforce](https://www.salesforce.com/news/stories/salesforce-reshaping-workforce-in-age-of-ai/?ref=modalpathethics.com) has published a useful account of a different route. It says hundreds of support engineers were redeployed while other vacancies were left unfilled. Its process included training during working hours, educational support, matching people to actual openings, and moving some groups together. Kristen Chuck, after more than eight years in technical support, moved into a success-architect role. 

This is a company account, not an independent evaluation of every employee’s experience. Its importance is the concrete sequence it describes.

The old role changes. Training occurs while employment still supports the person. A receiving role exists. Relevant experience travels into it. The move preserves an operative connection between the worker’s developing capability and paid activity.

*“Learn to use AI”* is not an equivalent sequence.

That instruction can arrive after income ends, with no funded study time, no confirmed receiving employer, and no evidence that the advertised skill will produce work. The person is being asked to finance both the transition and the risk that it leads nowhere.

Paid retraining is also insufficient if the work at the other end never exists. A promised job is insufficient if it requires relocation the person cannot manage or a schedule incompatible with care they must provide. The next step has to be reachable for the actual person, rather than suitable for an interchangeable applicant in a presentation.

The Salesforce example also has a limit that its success story cannot answer alone. Redeploying insiders while leaving vacancies unfilled can preserve existing employees and reduce entry opportunities for outsiders. An institution can improve one group’s transition while making another group’s beginning harder.

That means the assessment must continue past the employees already inside the building.

The practical alternatives include paid entry routes using the new tools, funded periods of supervised work, preserving teams where their coordination remains useful, and reducing hours where less labor really is required. Their suitability depends on the work and resources involved. A shorter week that simply imposes the old workload within fewer paid hours is not the same intervention as a genuinely reduced workload with maintained income.

This is also why a staged introduction can be better than dismissal followed by emergency recruitment. CBA reached that lesson after its staffing decision. Syndio reached part of it after losing people. The more competent transition does the work before the loss makes the missing dependency obvious.

---

## The Company Can Have to Close. The People Still Have to Live.

Some firms cannot finance a long handoff. Some products no longer attract enough demand. Some projects absorb resources that could preserve much more elsewhere. 

Keeping an organization alive can impose serious costs on its remaining workers, customers, suppliers, investors, or the public.

**“No layoffs”** cannot serve as a complete ethics.

Neither can *“we had to compete.”* Competitive pressure can be real while the chosen distribution of its costs remains avoidable. A modest return to shareholders and the destruction of a worker’s only workable income route are not interchangeable losses because both can be expressed in dollars. A profitable firm and an insolvent one do not face the same available choices. 

[Block’s filing reports](https://www.sec.gov/Archives/edgar/data/1512673/000162828026069524/xyz-20260930.htm?ref=modalpathethics.com) around $700.9 million of share repurchases during the first nine months of 2026\. That does not prove that its former staffing level should have been maintained, that all the money was available at every earlier decision point, or that one expenditure would fund a permanent alternative. It does establish that reducing employment costs occurred alongside other consequential allocations of resources, which belong in the comparison. So do investment needs and the company’s continued ability to provide useful services. The claim that a particular cut is *necessary* has to survive examination of the actual alternatives; it cannot be inferred from the attractiveness of a higher margin.

For profitable automation, the cost of carrying displaced people through a reasonable transition should be part of implementation. Income support, benefit continuity, usable training, and placement are costs of achieving the change without immediately breaking the lives organized around the old arrangement. They should not enter the discussion only after the savings have already been distributed.

For failing firms, pooled insurance and public provision are indispensable because the terminating employer cannot be the only source of support. A failed company cannot be ordered into supplying resources it no longer possesses. Its employees should not lose access to essential care or shelter because that is true.

This is where the earlier [*Post-Money*](https://modalpathethics.com/field-instruments-post-money/) argument becomes relevant. It did not propose abolishing exchange, making every project viable, or maintaining every previous salary. It proposes reducing dependence on immediate purchasing power where that dependence repeatedly breaks essential access, while preserving the financing and physical capacity required to supply the service. 

A clinic needs staff. Housing needs construction and maintenance. Training requires teachers, tools, and time. Cash without additional supply may leave people bidding against one another for the same bottleneck.

The financing therefore has to extend beyond the individual job: employer contributions, pooled transition insurance, public revenue, and actual service provision are instruments to construct and test. Where productivity gains enlarge profits or other income, the tax and contribution arrangements will determine whether any of those gains help carry people whose wages have weakened. Calling growth a *common benefit* does not create that connection.

No administrator should acquire unlimited authority over what a person may study, make, or become in exchange for support. The object is to preserve the ability to continue, including the ability to refuse one prescribed career. A support system that makes survival conditional on obedience to a permanent placement machine would reproduce the dependency in another office.

An income floor can keep an artist housed while commissions disappear. Livelihood security, productive access, and creative authority remain separate achievements. People need more than an allowance to watch an economy they can no longer affect.

Optional markets remain. Firms can compete, projects can fail, and consumers can choose other work. The social requirement is that losing one economic relationship should not automatically make a person’s basic life unmaintainable.

That changes the significance of automation. Less dependence on one wage makes it more possible for a person to welcome the disappearance of an unwanted task, bargain over a shorter week, retrain without panic, or attempt independent work. Material security makes technological change less coercive without requiring that technological change to stop.

---

## What the Numbers Leave Behind.

Block’s [November shareholder letter](https://www.sec.gov/Archives/edgar/data/1512673/000162828026069513/q32026shareholderletter.htm?ref=modalpathethics.com) contains a striking sentence:

> “Intelligence depends on signal, **and money is the most honest signal.”** 

**Money** can supply useful information. Payments reveal demand backed by purchasing power. Costs constrain what a company can continue doing. Profits can finance investment and register that customers value a service enough to sustain its production.

But a person losing purchasing power becomes quieter in ***precisely that signal.***

Their rent does not become less necessary. 

Their abilities do not become nonexistent. 

Their unfinished work does not become worthless because no budget currently supports it. 

The price system records what can be paid for under the arrangements in place; it cannot, by itself, tell us whether those arrangements are destroying something they should preserve.

- At CBA, the staffing decision exceeded the demonstrated removal of work.
- At Syndio, the reorganization lost people and knowledge the company needed.
- At Klarna, a workload estimate and a vendor relationship should not be turned into an invented employee roster or an equally invented absence of labor consequences.
- At Block, completed cuts and reported savings have to be acknowledged without pretending the transition ends at the company’s financial result.
- The wage evidence requires a further distinction: a person can retain employment while losing purchasing power or negotiating room.
- The entry-level evidence identifies another: the opening can disappear before anybody gets hired.
- BioWare shows that the coordinated activity itself can be interrupted before its next output exists.
- The artist and filmmaker cases establish that increased creative capability, continued human authorship, fewer purchased labor hours, and concentrated control can coexist.

None of these findings requires us to oppose useful automation. 

None permits us to treat useful automation as a complete account of what happened.

The practical standard is demanding and ordinary. 

- Preserve the improvements that reduce necessary work.
- Identify the existing support and capacity the transition removes.
- Carry what remains valuable through a funded, usable handoff where one is available.
- Where a job, project, or company must end, prevent that closure from needlessly taking the person’s basic life and future participation with it.
- Then examine whether new people can enter,  
  - whether customers actually receive the service,  
    - and whether the people using the new tools gain enough independence to make something of their own.

> **Less work should mean more life available to live.**

That outcome requires an actual arrangement connecting the saved labor to the people released from it.

Until that arrangement exists, a company making more money with fewer employees has demonstrated *a company making more money with fewer employees.*

We should neither deny this achievement nor pretend it has answered the rest of the question.