Will AI put us out of work? The 4 most likely scenarios

[ad_1]

newsletter
Newsletter

An analysis of the opportunities and challenges that artificial intelligence will bring to the workplace, by guest contributor Benjamin Eidam.

The ongoing development of artificial intelligence raises many questions about the future of work and our place in it.

The central question for many at the moment is: will AI put us all out of work? Or even make us unemployable for (most) work?

Or will we be able to collaborate effectively with this technology, enhance our skills, and create an unequally better world and fulfilling, well-paid jobs for all?

Ad

Ad

To answer this question, in this article we explore

  • The four most likely or most discussed scenarios among experts,
  • how AI could affect the world of work, extrapolating from current trends,
  • and highlight the arguments and counterarguments for each scenario.

We conclude this article with actions, perspectives, and action plans on how to make the most of all the scenarios discussed.

Let’s start with an overview and the why behind these scenarios, behind the power of change that comes from artificial intelligence. At a glance, it looks like this in a simplified way:

Figure 1: Why can AI even become a possible replacement for all human jobs, and what might that look like? | Bild: Benjamin Eidam / benjamineidam.com

To understand the potential impact of AI, it is important to understand the dimensions of this technology. If you think of “man versus machine,” it’s easy to see that this contest is lost before it even begins (and doesn’t necessarily make sense to begin with).

Once you grasp the almost infinite possibilities of computerized systems, you can deal with the coming implications in a much more informed way.
(A detailed discussion of these possibilities is beyond the scope of this article, but without at least a brief overview, at least scenarios 3 and 4 make little sense.)

Recommendation

However, this scenario does not seem particularly realistic, as we are often already technically above the required level. For example, we have helped clients in a variety of industries replace thousands of dollars in monthly marketing spend with an internal position that was initially made much more effective by AI training. And then, in the time saved, to take over the marketing tasks themselves with the new AI knowledge. In less time, at a much lower cost, and with higher quality, as friction losses in coordination and simply a non-internal view, etc. are avoided.

We are now also building complete tools for departments that make the entire department more effective, because it has an AI co-pilot that saves you up to 90 % of the time during the first drafts and beyond, as well as in the event of writer’s block, etc.

In short, if you give the market enough time for widespread adoption, this scenario is already very likely with today’s technology. But development is far from standing still. And neither do its possibilities.

2. AI as a productivity megaphone (📢)

In the second scenario, AI acts like a “steroid” for individual work performance, where, to put it bluntly, novices with AI achieve results like experts, and experts with AI create brilliant masterpieces.

This video is a good example of what this effect can look like in practice.

The central question in this scenario is then a very simple one: Won’t novice results suffice for most business needs? After all, very few customers can judge excellence from the outside, while sufficient performance can.

So, is it enough to be able to formulate your question reasonably clearly and to be able to evaluate the result at least roughly to be able to produce the most results? (initially mainly digital, but rapidly increasing analog)

Or will the base level for performance simply continue to be raised, as has often been the case throughout history? Either way, this result will lead to major changes and recalibrations in almost every area.

Arguments in favor:

  • AI can enhance the knowledge and skills of workers, leading to improved quality and efficiency of products and services.
  • End customers often cannot distinguish between expertise and AI-enhanced results, leading to a potential democratization of expertise. (For example, it is often difficult to tell from the outside whether a web designer has spent 100 hours or 10 minutes on a website, as long as minimum standards are met.)

Counterarguments:

  • Relying on AI could undermine the development of human capabilities if basic skills are neglected by AI support. This is likely to be met with a strong backlash as soon as it becomes apparent. We are already seeing the first cases.
  • Expertise may no longer be adequately rewarded across the board as AI-enabled services dominate the market.

If this scenario were to become a reality in the next 10 years, the market dynamics would change massively. What was previously outsourced can now be produced much more cheaply locally, at least for digital services, and possibly also for manufacturing and similar areas. (Depending on how quickly multimodal robots like teslabots become available at low cost).

The quality of the results will probably remain similar or even increase slightly, only the production requirements will be significantly different.

We are already seeing the first signs of this scenario, from programming simple code and various texts to images, 3D animations, and more and more videos. So, with some limitations, the same applies as in Scenario 1: it is unlikely that this is the future because it is already here. Only, as so often quoted, not yet equally distributed everywhere.

For these and other reasons, we do not consider this scenario to be the future. More likely, however, is the next one: the market.

3. AI as an expertise filter (💯)

Essentially, in this scenario, humans are paid primarily for the quality of the results they achieve with AI. However, they can only achieve and take responsibility for these results because they have expertise and experience in this area.

In other words, the machine does 99% of the work, but the critical instance of approving the quality and taking responsibility for the result remains with the human. This means that only the top 0.1 % – 5 % of the current workforce in each affected industry will be needed, but they will also need to be trained with an excellent understanding of AI.

This is the first scenario with “serious” implications for the global economy, as AI enables the top 1 % to 10 % of experts in each niche to do incomparably better work in the same or less time than their entire field before the use of such tools.
This could result in the potential unemployment of > 90 % of the workforce in that field.

In this scenario, for example, only a few thousand system administrators are needed worldwide to solve complex cases and monitor the work of automated systems. Everyone else would have to find a new job.

As the output per time is much better and the speed is much higher than today, i.e. there are many more results with higher quality/complexity in production, the responsibility of these positions increases accordingly. If I manage 30 or 300 projects per month with AI instead of 3 today, I have to be better than before.

It’s a kind of techno-meritocracy: the technology filters out the best, who get everything. The rest are left with almost nothing. (In current economic terms)

Another mechanism comes into play here: greater expertise also means being able to ask the AI better questions. And to classify its answers better. We already see this discrepancy in our work with some large language models. More reasons for this scenario.

Arguments in favor:

  • This scenario could lead to everyone working in their “genius zone” and market forces automatically realizing the opposite of the Peter Principle.
  • The infinite variety of niches could create new opportunities and specializations for workers that are almost unimaginable today. Provided that local education systems make these opportunities accessible to most people.

Counterarguments:

  • The resulting mass unemployment could lead to social and economic problems if large numbers of people cannot find suitable work. Or are not adequately trained and educated for them.
  • The concentration of expertise in the hands of a small group could lead to an (extremely) unequal distribution of resources and power.

If this scenario becomes a reality in the next 10 years in economies that currently have a high proportion of services, the effects will be dramatic. In most affected industries, ~95% of workers in their current fields could be laid off. Figure 2 in this International Monetary Fund article shows this very well: First, strong productivity gains are achieved (which we are already seeing, see Scenarios 1 & 2). Immediately thereafter, most people’s wages collapse toward 0, as they are no longer needed to create economic value in their current form.

Image: IMF

We believe this is the most realistic scenario at this time for two reasons:

  1. Market forces: Whether we like it or not, morals, personal views, or political positions play little role in practice, at least over the long term. But as soon as technological possibilities have reached the status of mass availability, i.e. have fallen below the amortization threshold, they will be used. This can even be seen in vital issues such as the current ecosystem catastrophe: results are almost exclusively achieved when the market regulates them.
  2. Simplicity: This scenario also seems very realistic at the moment, because the formula [expert in field x] + [knowledge of how AI is best used in its field] can be put into practice very quickly and is already multiplying productivity in companies in a wide range of sectors. With the right setup, you can quickly become familiar with a tool that you can ask directly when in doubt.

We have helped customers save up to 99 % of the time and cost of a task by optimizing their AI processes. However, this is only possible in a few hundred application areas today. As AI capabilities increase, this number is expected to grow dramatically.

Will this be the end of work for most people, while a handful of select individuals will continue to have well-paid jobs in the abstract? Possibly. In the final section, we take a closer look at the fact that this is not a disaster, but, on the contrary, maybe the best thing that could happen to us.

4. AI as the literal end of work (🤖)

In the final and most extreme scenario, artificial intelligence replaces humans in almost every conceivable job and field of work. This could lead to the collapse of the current economic principle for several reasons.

Some of the key mechanisms behind this are well explained in this video.

Even this case does not automatically lead to unrest and social upheaval, although such events are becoming more likely.

There are many conceivable outcomes when AI possesses such capabilities; OpenAI CEO Sam Altman himself has explained the better and more desirable scenarios here.

Arguments in favor:

  • AI can work faster, more efficiently, and with fewer errors than humans, making it superior in many areas.
  • The automation of jobs could lead to a massive increase in productivity and a higher standard of living, so it is economically extremely tempting to “run” towards this scenario. And then end up in a dead end.

Counterarguments:

  • The far-reaching social and economic consequences of the resulting mass unemployment could lead to unprecedented forms of instability and conflict.
  • The collapse of the current economic system would require significant adjustments and restructuring to find a new equilibrium. Without good solutions before this happens, extreme challenges are inevitable.

If this scenario materializes in the next 10 years, we will need a fundamental renegotiation of work, meaning, and value in the affected societies.

By then, at the latest, major political decisions will have to be made, at least in the areas of finance and education. In this case, one of the current central sources of meaning, identification, and self-worth for many people will be fundamentally called into question.

Because this scenario has been carried around as a threat for decades, often with a great deal of hype, we are currently cautious about judging its likelihood. It cannot be ruled out that “this time” everything really will be different, and there is even much to suggest that it will be. Nevertheless, this is a very big leap, and even if it happens, we are following the Sagan standard here and taking a cautious approach.

There is no limit to “better”: what all scenarios may overlook

Whether the prospects outlined here look dysfunctional or utopian is up to you. It is simply the spectrum of the most discussed scenarios.

However, one thing that is often overlooked must never be forgotten: In an entropy-based universe, solutions are always needed, no matter how much AI can decrease. There is no limit to “better,” which means that AI and human work in various combinations will continue to provide opportunities for improvement.

No matter how high one’s standard of living is, there is always a drive for more. That is part of human nature.

Following this approach, it is also possible that we will experience all 4 scenarios. In chronological order. Exactly as it looks at the moment, as described.

In any case, this classification will hopefully help to improve our understanding and our ability to prepare in practice.

Which brings us to the last part of this article: how can we as a society, companies, politicians, and individuals best respond to these scenarios? Is there perhaps even a “silver bullet” that will work in all scenarios?

One solution for all scenarios: 7 billion (automated) companies

Many approaches are being discussed as possible responses to the occurrence of the above scenarios; from an unconditional basic income to a robotics- or negative income tax to “fully automated luxury communism”.

And even if an exchangeable intermediate medium in the form of money always makes sense as a supplement to value, it is only suitable as a central mechanism to a limited extent if almost every economic value is produced automatically.

So why not utilize the fundamental economic mechanisms of supply and demand? Even with an almost infinite number of AIs, there will never be enough solutions for every problem; the space of productive possibilities through the combination of atoms is too large for that.

So why not, as mentioned, enable everyone to have their own company via AI as an “extended arm of themselves”?

Raval Navikant builds this approach on the concept of “productized authenticity” – the utilization of one’s characteristics and talents independently of oneself.

In a nutshell, it goes like this:

  1. I find out what is “really me”, what I do, what I like. (For example, cooking all kinds of pasta, refining it etc.).
  2. I validate these results. (For example, I cook for friends, relatives, etc. If it still feels fulfilling, my intuition seems to be right)
  3. I look for a way to bring this now consciously realized part of myself into a form that can be consumed independently of me. (e.g. by recording pasta tutorials, writing recipes etc.).
  4. I bring this part of my “self” in product form to a target group that I can help in the best possible way. (e.g. someone wants to cook the perfect pasta for their fiancée, what is it and how does it work?)
  5. In most cases, I have built up a business that deeply fulfills at least one part of my personality and helps people with problems in this field to solve them. An almost perfect economic win-win situation. The great thing about AI is that if the business side of this approach doesn’t bring me joy / is too challenging or resource-intensive, artificial intelligence can take over more and more of it for me. From website construction to the business plan, etc.

The bottom line is that you have better solutions for everyone, while everyone can develop and realize their full potential and is still / probably more relevant than ever before instead of being “technologically useless”.

This does not automatically mean that everyone has to go from being permanently employed to self-employed. This form of enterprise also leaves room for more than one pasta enthusiast per “authentic” company created in this way.

However, there is far less “must” in this approach than in almost everything we currently have and probably most of what is to come.

There are still many people today who love their job and do it out of pure passion. But unfortunately, this is not the norm and even these workers need a point of orientation in an AI-revolutionized economy.

Making this approach available to the masses will undoubtedly result in massive changes in areas such as trade unions, insurance, education, etc. But these are much smaller than chasing an exponential trend.

Bild: Benjamin Eidam / benjamineidam.com

The central question here is: Can we as a society imagine people being paid for helping others by living out their fulfillment?

This is a completely different approach to “work puts food on the table, be grateful that you have some”. Until this change in individual and collective mindset is at least possible, realizing it is likely to be challenging.

However, if you follow this idea, you end up with the scenario Ravikant calls “7 billion companies“:

“There are almost 7 billion people on this planet. Someday, I hope, there will be almost 7 billion companies.”

As mentioned above, there are, in principle, an infinite number of niches for better products and services. AI will be able to serve many of them. However, humans are social beings and will therefore always want to interact with other people. We therefore see the “7 billion companies” scenario as the best answer to all of the possible 4 scenarios.

This is because, in principle, it can already be realized today. And it will only become easier than the alternatives as AI capabilities advance.

If, through far-sighted political and social decisions, the AI revolution leads to almost everyone creating value by being fulfilled and helping others, then this is one of the best outcomes that this technological upheaval can bring.

Summary: Where do we go from here?

The four scenarios above show different effects of AI on the world of work, from the expansion of everyday working life to the far-reaching replacement of human labor. There are opportunities and challenges in each scenario.

To prepare for the future of work, both employees and companies should be open to change, learn continuously and adapt in order to benefit from the advantages of AI and mitigate possible disadvantages.

After all, this revolution in the world of work is already the third that we as a civilisation have undergone. By far the fastest after the Neolithic and the Industrial Revolution.

But like the two before it, there is only one good way to emerge from it better than we went in: Understanding the underlying technology and using it to best effect. To transform this specific knowledge into concrete results, we consult IT service providers on the optimal use of artificial intelligence. To achieve massive time savings in all areas.

Let us therefore conclude this essay with a quote from Abraham Lincoln:

“The only way you can predict the future is to build it.”

[ad_2]

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top