Michael Eriksson
A Swede in Germany
Home » Misc. | About me Impressum Contact Sitemap

Automatization, qualification, and unemployment


This article discusses some issues relating to the continual move of easy tasks from humans to machines (and similar developments), including employment issues.


Below I will uniformly use the word “wage” to refer to income from work, be it an actually wage, a salary, or something different. This is motivated by a combination of simplicity and the fact that most references will be to work that typically is wage based in English speaking countries.

The main issue

Since at least the industrialization there has been a clear tendency that the least qualified tasks done by humans are moved to machines or have disappeared altogether. While this is, in and by it self, something positive, it does pose an indirect threat: Human ability has not developed at the same rate, which leaves the replaced individuals without gainful employment.

It is important not to over-react to this: Similar concerns have been raised in the past, including by the violent Ludditew movement. While these concerns may have been warranted in the short term and for specific individuals, the mid- and long term result was more wealth for everyone—including the people who would have worked in the abolished position, had things not changed.

The modern situation differs in at least two regards, however: Firstly, there are fewer alternate jobs of a similarly low qualification available today, which makes it harder to find alternate positions in the long term. Secondly, the distribution of ability (most notably intelligence) between different types of work used to be much more random than today: People tended to follow in the steps of their parents, few of the lower class could afford educating their children (and education did not always make a difference), members of the middle and upper classes tended to find “appropriate positions” irrespective of qualification (if they at all needed one), etc. Today, in contrast, there is a filtering system that guides the more intelligent into more qualified positions and keeps them away from unqualified ones. Thus, in the past, a higher proportion of the affected work-force had the brains to take on more qualified tasks, once given the chance.


The above mentioned filtering systems is far from fool-proof, however. Most notably, promotions and the like tend to be more determined by superficialities, manipulation skills, networking, etc., than by competence, intelligence, and actual suitability (cf. e.g. my discussions of company life)—here, however, we are still above the rising limit of employmentability. A more relevant, but smaller, flaw in the filter is that working-class children often grow up in an anti-education environment, which can prevent some of them from reaching their potential.


Textile industry

One of the earlier cases, where hand-weaving and -sewing, etc., were replaced by machines. Partially, this was done with a near immediate replacement (e.g. automatic looms); partially, with an ongoing mechanization (manually operated sewing machines of greater and greater ability).

Even in the relatively short term, improvements of an order may have been possible—today, the productivity is so far ahead of the old manual work that a comparison is ridiculous. The drops in prices and increase of availability of various textile products is correspondingly high. (Although partially caused by a move from internal production to production in countries with lower wage levels.) It is worth remembering that clothes were once quit expensive—and the time spent e.g. darning socks was a big money saver.


Those who lack in price consciousness should additionally bear in mind that they often pay far more than they need, e.g. by choosing brands with artificially high prices or by going to stores with high mark-ups. That people, in particular women, pay several times too much is not uncommon. Even quality clothes are cheap when bought in the right store—and a sizable part of even that price goes towards advertising, taxes, staff costs, ..., rather than the actual production of the clothes.

Bank tellers

While there are still plenty of bank tellers around, their jobs have gradually changed, and keep changing. First, a number of routine tasks were moved to ATMs; second, most other tasks were moved to Internet banking. (At least in Germany and Sweden: I have the second-hand impression that some countries, e.g. the check-based US, are trailing.) What remains is counseling, trouble shooting, and non-standard tasks that require a higher ability (although, regrettably, this higher ability is not always present). Obviously, there is still a considerable amount of standard tasks being done, to accommodate people who do not have Internet access or do not trust Internet banking; however, this amount is steadily shrinking, and will likely be near gone in the not-too-far-away future.

Checkout clerks

Currently underway is a gradual switch to automatic checkouts or checkouts where the customer operates the machinery. This was preceded by a gradual development in cash registers and the introduction of bar codes and non-cash payments. Interestingly, one of Germany’s leading supermarket chains, Aldi, skipped on bar codes and had their clerks count sums in their heads (!) until sometime in the early 2000s.


This switch is not merely a question of the proficiency needed to handle the cash register, but also of trust: Such a system would have been impossible when the customer had the ability to grab a handful of cash payed by someone else from the register, instead of putting his own cash in.

Garbage men

This is an area where a future change seems highly likely, and where machines and better equipment already has made some changes when compared with the past.

The question is: Why has this role, which is both poorly paid and of extremely low status (compared to e.g. bank teller), not been affected to a higher degree?

The main reason is that being a garbage man can actually require more thinking and, importantly, thinking/acting that is harder to automatize. The job of a check-out clerk is comparatively uniform and (at least after the introduction of bar codes) requires very little skill outside of trouble shooting and customer interaction (both things that most clerks are poor at). The job of a garbage man, OTOH, has many complications even in normal cases: Alone driving the truck is a task that is non-trivial to automate; and which would meet resistance from at least some people, if the attempt was made. Finding the garbage cans and bringing them to the truck also requires considerably skill in terms of AI, because their locations are rarely sufficiently uniform, there are different cans for different types of garbage (which must be handled differently), etc. In my current apartment house, for instance, the garbage men have to get into the house, go through the house, enter a yard, find the right cans, bring them to the truck, empty them, and bring them back the same way.

Replacing this with an automatic system would require either strong AI and robots, massive investments in infrastructure (including changes to all houses), or a complete re-working of the system (say through costly standardizations or an underground pipe-system for garbage).

Note, however, that the complications vary from location to location. In a system where the tenants bring the garbage in sacks to the road-side, e.g., the task is much easier, and automatization much likelier. Even here problems remain, like driving or differing between garbage and non-garbage.

Qualification vs. automatization

The discussion of garbage men and the problems associated with automatization applies generally: Qualification does not always correlate with the difficulty to automatize.

Learning to operate a manual loom at a “professional” speed, e.g., can take noticeably longer than learning to handle a cash-register, due to the manual skills necessary. Arguably, this would make the loom operator more qualified; however, it is a type of qualification that does not rely on intelligence and which is very easy to automatize.

Similarly, the job of computer is no longer relevant: Before computing power became cheap (or at all available) “computer” was an actual job title. In university settings, in particular, it was common to have a few people around to do the boring leg-work arithmetic and algebra that is today done by computers, e.g. to invert matrices. While this may seem like a highly qualified position, certainly something that most people would fail at, doing the same job with a computer (in the modern sense) is very easy: Humans are awful at arithmetic, and the fact that it took a good head to be a computer reflects human limitations—not the toughness of the task. In fact, in absolute terms, the hardest part of this job may have been the input and output (e.g. deciphering whatever scrawlings the professor called a matrix)—and with the modern computer being fed unambiguous characters over a keyboard...

Downward pressure on wages

An unfortunate side-effect is that there is a downward pressure on wages for jobs that could/should be done by more qualified people. An example is computer support, in particular per telephone: Because this is an area where most corporations wish to spend as little money as possible, they tend to hire people with very low wage requirements. The effect is that unqualified people do this work, leaving the customer in the rain—or someone who is qualified has to do the job at a lower wage than is justifiable. If the unqualified were employed elsewhere, this would not be possible and the quality of the staff would rise (possibly at the cost of a reduced quantity).


Notably, there is no immediately obvious cost to providing poor customer service; but only indirectly through customers who move to other products, or service cases that come back half-a-dozen times, instead of being completed in one go.

Further, from the POV of the corporation, a satisfactory solution can often be reached in one go—with a major additional cost for the customer. Consider e.g. that the first advice given on many hot-lines is “re-install Windows”: This will solve most problems, true; however, it will also incur hours of work for the customer, bring complications with backing up and re-installing data, possibly cause data loss, and is generally something that should only be done as a last-ditch effort, when a thorough investigation has led to no other solution. The advantage rests solely with the hot-line, who sees its work done in thirty seconds and can take the next caller; the cost solely with the caller.

Future developments

It is plausible that these changes will continue, with the bar of where humans are needed continually rising. In the long term even reasonably qualified jobs will be moved to machines (from rumor, I have the impression that many modern physicians consult extensively with “Dr. Google”), and in an even longer term (at least several hundred years) humans may be marginalized by AI systems.

Possible (or not) solutions

Increased education

Unfortunately, many are under the faulty impression that merely sending someone through more education (most notably college) will lead to a corresponding qualification—and, therefore, this is often touted as the solution.

The truth is very different: While more education will increase qualification, it will not do so in a uniform manner, nor will it overcome the disadvantages caused by a lower intelligence. The people who should go into higher education almost always already do so, and additional efforts in this direction will do more harm than good: The change in real (as opposed to nominal) qualification will be small for most, the perceived value of a degree will be diminished, and so will the actual value (because the standards of the degree are lowered to allow the less intelligent to pass).

Keeping wages low

If wages for unqualified work are kept sufficiently low, this can give organisations sufficient incentives to prefer unqualified employees over automatization.

This, however, brings many complications, including that the wages may be too low for a reasonable standard of living and that it can hold the long-term progress back in a negative manner. (Cf. the discussion of Luddites above, or consider that slavery has been mentioned as highly detrimental factor to the industrial development of both the Roman Empire and the “South” of the pre-abolition US.

A particular issue is political viability: Many politicians, union leaders, and the like, are in strong favour of comparatively high minimum wages; and considering that a minimum wage is an easy opportunistic way to gather votes from naive people, this could be a major stumbling block.


A minimum wage, OTOH, is extremely detrimental to the least qualified: The typical result is not that they earn more doing what they did before, but that they end up being unemployed—or that they do earn more than before, but now have to do the work of several people because half their colleagues were laid off.

Basic income/Citizen’s wage

A basic income/citizen’s wagew is sometimes suggested, and could provide a reasonable protection against the economic consequences for the unqualified—with the added benefit that many qualified people who want to pursue an independent path as life-long students, philosophers, or artists can do so without the extra burden of a “day job”, by choosing a more restricted standard of living. (I stress that my understanding of the word “student” implies noticeably more studying than partying.)

Among the down-sides: The money must come from somewhere. There is arguably an element of injustice towards others. Slackers who bring no benefit for anyone or anything will have an undeserved easy living—and others will be encouraged to join them. It can be a negative incentive where studies and work are concerned.


Some of these points apply to “normal” social-security systems; however, the greater complexity and arbitrariness of these systems make them less suitable for a number of reasons.

Long-term changes of population structure

The probably best, but also least likely to be implemented, solution is to change the long-term population structure, e.g. by giving financial incentives for the highly intelligent to reproduce and corresponding negative incentives to the unintelligent. Unfortunately, even making the suggestion is risky, because of the irrational blind hysteria that tends to accompany the subject.

Notably, however, it would bring advantages in other areas too, in that it would reduce human stupidity. In particular, it would combat the current trend of the unintelligent having more children and shorter generations than the intelligent.