From many reports, automation is increasing in sophistication so that every possible full-time occupation can be performed by automated machinery.   Many have taken this to mean that automation will eventually replace humans in every employment type job.

An employment-type job involves some steady working hours over an open-ended indefinite period of time.  This provides structure in a person’s work week in that free time is scheduled around committed working hours.  More importantly of course, employment provides a steady flow of cash for being consumers of goods where most of the goods are beyond what is needed for basic survival.

Once automation takes over, there will massive unemployment.   People will be idle with no committed hours of work on the horizon, and these people will not have any money to consume anything.

There are numerous scenarios proposed for what will happen when automation takes over all jobs.   I tend to imagine the most likely scenarios are pretty ugly in terms of social unrest and the inevitable backlash of an increasingly authoritarian police state.

The state itself will have automated rulers, judges, and police forces.  We are already seeing this happening with the development and use of policing drones with increasing access to force, and with automated judgments and automated rule making.    The primary impediment to automating government is the political power of the humans currently occupying positions in government.   This is also evident inside government where automation decisions are routinely postponed to protect jobs.   Eventually, the government will be fully automated because it is pretty easy to do.

The scary prospect is a large, restless, and idle population facing an increasingly brutal authoritarian government by robots.

One proposal is to copy the Roman Empire’s bread and circuses approach, providing the idle population a basic income and free entertainment.   The latter is readily available with modern electronics and Internet.  We already have largely free access to music, videos, immersive games, and other forms of entertainment.  Even among outside world activities, many of those experiences can be delivered using similar economic models where things keep working so long as people keep using them.   Such activities such as travel are already fairly inexpensive and can be more inexpensive, though they probably always will be more expensive than the virtual forms of entertainment.   Let’s just say that free entertainment is feasible.   This goes a long way to solving the problem of the idle population.

The eventual automation of all jobs will occur gradually with people adjusting over time so they may not be offended if they find themselves spending all their time being entertained.   This is similar to how smart phones emerged as an essential accessory for the palm of the hand, face up with the forefinger of the other hand permanently kept no more than a couple inches away.   The coming introduction of mixed reality glasses will be similarly smoothly integrated into society where everyone will be walking with boxes over their eyes to engage in games that are integrated into a projection of the real world outside that box.

If we wanted to be clever, we can convince the person that playing the virtual reality game is his employment.   He earns money as a basic income doing virtual tasks within the game, or he gains wealth through game play.   This already exists in virtual games where the accrued wealth in a virtual world has hard cash value.

The inverse of virtual income is real income for doing something of virtual value.   An example of is the growing population of YouTube content creators whose primary income source is advertising from those videos.   This is really an extension of entertainment market that has existed for a long time, but YouTube content requires much lower level of investment in production equipment, editing, etc, to produce content that will produce a steady income stream.    Anyone can produce content that can create an income if they keep up the work to produce fresh content every few days or few hours.

However, I don’t think that model can scale for the entire population.   The YouTube content creators are spending a substantial part of their time producing a steady list of new videos.  In contrast, their revenue depends on a much larger population spending most of their time consuming that content.   This works now because most YouTube content consumers are attractive to advertisers because those consumers have disposable income.   When everyone is completely idle from unemployment, they can spend all their time watching YouTube, but they’ll have no money to buy anything advertised.   This model does not work in the end-state of total automation.

Going back to the bread and circuses model, we probably will need to solve the part of the bread, specifically in the form of providing a basic income to the unemployed.   I don’t see how the economics of this will work.   Automation is taking over most industries by producing products that can out-compete based on price.   These companies may be profitable, but they are not generally amassing massive cash reserves that even an automated government can tap to fund basic income for the masses of unemployed.

The companies that do amass fortunes are companies like Apple that can sell their products at premium prices due to some unique market position.   Such companies appear to be the exception for most highly automated firms.   I don’t see how we can fund a basic income for the unemployed, but to be fair I haven’t thought about it too deeply.   Most firms produce products where the revenue ultimately comes from consumers or through tax payers.   Both consumers and tax payers disappear in the fully automated world.

The fully automated world will be one with a population almost entirely far more impoverished than current standards though perhaps they have access to free virtue entertainment.

I can see evidence of increasing automation among the professions requiring higher education or certifications.   Increasingly automation is in the form of tools that we use for our primary job tasks.    Much of the work we do is repetitive, requiring similar operations on a recurring basis.  The justification for our position is that we are often uniquely capable of completing the task.   However, the same tools that allow us to complete the task are increasingly equipped with features to script the operations so that the task can be done by a single mouse click.    In many cases, it is optional that we use these automation features.  People like myself who hate doing the same thing twice readily adopt these and then point out that the task doesn’t need me any more.  There are other tasks to work on, but my automation just eliminated a major fraction of a full time job.

This example gives me the idea of what the real end-state will look like.   There will always be paying jobs for people to do.   What will change is that the concept of employment will disappear.   Employment being the idea of steady work of a predictable number of working hours each week and a predictable income at regular intervals.   The work of the future will be increasingly of short gigs, especially for the higher professions.   The task is to find a way to automate something new.  Once automated the job is over.

As mentioned, the automation technologies are making this easier by automating the infrastructure necessary to maintain a new automation that is introduced.   Basic features such as quality control, performance monitoring and remediation, and incorporation of updates are automated within the tool.   The only thing the employer needs from the human is the innovative addition of the new automated feature.

As we see in modern agile teams, the teams may in fact have steady work over long periods of time, but that is because there is a feature backlog in the pipeline.  The team gets re-hired for each short sprint to produce a new feature, where each sprint is a distinctly new job.   I see eventually the agile teams to be more dynamic where successive sprints will have different mix of participants.   I think this is inevitable because there will be competitive pressure from outsiders offering to do the same work for less money or with some specialty that is better suited for the next sprint’s feature.

The future of professional work is an agile model with time-limited employment measured in units of individual sprints.   Agile came out of software development practice, but is applicable to other professional activities especially as those activities involve interacting with automated systems.

Consider the legal profession where there is some claim for human employment to argue cases through court processes.  Even in the current context, there is rapid automation of legal services.   But as regulatory and judicial systems are automated, the lawyers will be interacting with automation instead of humans.   In such a scenario there is an opportunity to automate the attorney’s role in handling certain types of cases presented before a specific automated system.  Something like the legal system may similarly turn out to be agile in nature, and I suspect it already is in many areas.

The professions will see steady employment converted into unsteady gigs to accomplish some goal in an agile-like sprint.

Many other jobs, I believe, never will be automated.   Even if all jobs can be automated, many jobs are likely be more cost effective for humans to do.   There will be jobs where humans will be cheaper than machines and they will get the job done more reliably and often more quickly.   In particular, I am thinking of the dirty, dangerous, and heavy jobs at the lowest level.   These are the jobs like cleaning up the Fukushima Daiichi mess that is going to employ a lot of people over many decades doing jobs that automation can not compete with humans.   This is counter-intuitive because these are precisely the kinds of jobs we most want to automate.  Yet the demands of such disasters are too dynamic and unpredictable to automate and even automated attempts end up failing.

Pretty much any disaster or emergency scenario will continue to employ humans to enter the dangerous areas and do unpredictable tasks.   Humans will be needed for search and rescue missions, or fighting fires, or managing some other emergency to prevent it causing catastrophic damage (such as the recent Oroville dam concerns).   They may be assisted with automation, but I doubt that automation will decrease the head count of humans needed to do the work.

In the fully automated economy, the best jobs in terms of prospects of income and steady employment may be the worst.

More generally, I see other job opportunities.

I believe all of the work under the CEO can be can be automated.   I suppose it is possible to automate the CEO, but I think it is inevitable there were remain many human CEOs.   Most of these CEOs will want people to work for them to give them sound advice and to give them assurance that the delegated tasks will be completed competently.   Many of the people working below the CEO will also want to have at least an assistant if not their their own level of sub-delegation.    This does not extend very far, however.   Eventually the job will be clear enough for a person to define the necessary agile sprints to accomplish the automation needed to meet the new requirements.

Many people who are innovative work better when they have others to work with.   As long as people have resources to invest in innovation many will employ others to help them realize the innovation.   These are more like limited partnerships or corporate officers than regular employers.   However there could be a lot of such jobs if the regulatory state operates in an way the encourages innovation.

I think the current state is over-regulated to the point of discouraging entrepreneurship, and this is depressing the number of such jobs that may be possible.  I also think that a more automated state can be more flexible by applying regulation more tactically to address only areas of immediate concern in order to free up other areas of the economy where there is less of a concern of misbehavior or incompetence.   I think the automated state would have a major incentive to open up as many of these job opportunities as possible in order to get more people employed and contributing the economy.

Outside of entrepreneur jobs, I believe there is a human instinct to want to exchange something in value for something that is perceived as a benefit.   People will do things that other people will appreciate.  Those who appreciate it will want to reciprocate with something of value in return.

I think the idea that everything will be automated forgets why we have an economy in the first place.   Modern society measures its economy in terms of consumer goods, material constructions, and state infrastructure.   All of this can be automated.   But these are consequences of an innate human behavior to engage in an economy.   Even when all consumer goods are fully automated from mineral extraction to retail delivery, people will still end up bartering among each other for their handiwork.   People will pay for a neighbor’s garden vegetables, or a friend’s chicken eggs, even though comparable or even superior products for lower cost will be available at the store.   They do so because the economy is partly an expression of community and of friendship even if that friendship doesn’t extend beyond that occasional transaction.

An example comes to mind about why some people end up frequenting a favorite restaurant.   The restaurant may have good food and good prices, but a large part of why they return is because they like doing business with the people working in the restaurant.  They get to know the head chef, or they know the owner, or may just like the wait staff.   They continue to return because of the people who are there.

There is an economic value to human personalities at the other end of the transaction.   This human connection drives the economy as we return the same businesses to express our desire to do business with them.   This may not be a huge component of the economy, but once we automate everything else, this component will be more dominant.   We’ll buy things from sidewalk vendors offering their items from their personal truck or folding table.   We see this today with farmers markets or food trucks, both of which are getting increasingly popular and at least their point-of-sale presence is minimally automated.  If people have access to funds, they will use some of those funds on these kind of businesses even when comparable options are available at adjacent vending machines.

Another example that comes to mind is entertainment.  I described earlier that much of our entertainment is automated.  We listen to recorded music, we watch recorded movies, or we play programmed video games.   This is cheap entertainment at least in terms of the incremental cost of delivery.   Meanwhile, we still appreciate live entertainment and we especially like the performances of people we know or recognize.   We want them to perform live in front of us.  Maybe we will automate live entertainment when we create lifelike robots with human appearance and human-like personality so that we relate to the robots just like relate to fellow humans, but even then there will be living humans who will offer an act that we’ll prefer over the robotic act just for the fact we have a different connection to the human performer.  In this scenario, the robots and humans are peers but they still compete for our bonding with them at a personal level.   I think the humans will remain competitive in establishing that kind of loyalty bonding.

That brings me back to the idea of the basic income for the unemployed.   Dismissing my objection that I don’t see the economics of how this can be funded, I still think the idea of an income for unemployment will not work.   The government may go through very elaborate calculations about what minimum income is needed for basic needs of each individual.   Presumably, if a person spent less than this money, he will suffer too much.

People are very crafty.   Some may find ways to cut expenses enough to allow some of the money to accumulate into savings.  In an environment where most people have no savings, a little savings can quickly be valuable for informal transactions with this income.  Inevitably, there will arise an underground market of a form of employment that works off of savings instead of the present day income.

I think in other cases, people will extort income from others in their communities but in a way that is within the limits of the law at least in terms of avoiding the automated law enforcement.   I can see this happening in the automated entertainment of virtual reality games where the players bridge the virtual economy with the outside economy.   People can make virtual deals that may involve real-world pay offs.   People will be paying each other off with this basic income that supposes that people are unemployed.

Yet other cases may involve people gathering in communities to pool their incomes to share expenses, allowing them to collectively live at a higher standard or to collectively accumulate a savings.  In either case, they can use their elevated status to have power over others and this can lead to other underground economy schemes.

The automated government may be able to police this, although I have faith in the human ingenuity to outsmart the automation especially at these petty levels.   The concern is that the government may not detect this before the problem becomes out of hand with wars breaking out between competing cartels in different parts of town.   At this point, I doubt the automation in government will be able to quickly bring this to a peaceful solution.

This can be avoided by replacing the idea of an income with an expense account.   Here, each person is granted a per-diem for specific expenses such as meals, transportation, or housing.   The expense account can not exceed the daily limits, but if the person spends less then the limit, there is no left over to add to personal savings.   Similarly, the expenses have to be valid and this eliminated the possibility of using the funds in underground economies or pooling per diems to achieve a higher standard to set one group as more privileged as another.

I think this daily expense account is more realistic than a basic income to provide for the basic welfare of the unemployed in a fully automated economy.   It is also helpful to illustrate how dismal this state of affairs will be.  The expense account is a daily ration, and a cot in a shelter.  But the shelter will have plenty of available virtual-reality head consoles.