When AI takes all of the jobs, who will pay our taxes?

When AI takes all of the jobs, who will pay our taxes?

About 80% of state taxes come from individuals and payroll taxes. Only about 7% comes from corporate tax. And, if there are no taxes to be scooped into the fiscus, how is the government going to provide services for its citizens?

Schindlers Attorneys is a prestigious South African law firm with a long history. Some of the partners are of a technological bent. They have enthusiastically embraced AI, building a legal AI engine to assist them with all manner of legal preparation. 

Their system recently dealt with an “ex parte” application on a real matter that was presented to a full court Bench. The AI was asked to draft every submission necessary to sustain the relief sought, for counsel to read to court. The submissions were produced in under 30 seconds, 120 pages worth, using a training canon of case law, law texts, opinions and precedents, as well as the entire record of the matter at hand. 

The document produced by AI was (of course) checked by a lawyer, but there were precious few changes to be made before the matter came up in court. The application was successful and, when the attorney revealed that every word presented was produced by AI, one of the judges and several of the advocates in court commented (only half in jest) that their careers were over. 

Here’s the rub. This matter required the highest level of legal skills. It would usually be prepared and argued by very well-remunerated legal practitioners, who would take weeks to prepare and be paid tens or even hundreds of thousands in fees. 

If AI can operate at this level of very starched white collar, what of the rest of us offering our more mediocre skills to employers?

And so, I ask you to consider the following scenario, the likelihood of which nestles in the great pot of AI predictions along with many others, any one of which may be borne out (our AI future is going to be nothing if not unpredictable). 

It goes like this: AI innovations continue to accelerate, including in the areas of computer vision and robotics. Jobs soon start disappearing at scale as corporations race to deploy cheaper and more productive alternatives. This includes “thinking” jobs and jobs which require physical dexterity. Corporations, businesses and factories become human-free dystopias of robots, computers and automated devices. 

How widespread will this employee replacement be? How high up the skills hierarchy will it reach? The answer is starting to become clear – very wide and very high.

If AI can produce the work of highly paid lawyers in no time at all and at very little cost, then there are few levels of expertise in any industry which will be immune. The question that arises is how soon this replacement will happen and how brutal it will be. The argument has been made by some that it will “free up” employees to do other highly skilled or more productive jobs within the commercial sector. This is an optimistic view. There is no evidence that this will happen at a speed which compensates for the misery caused. 

John Mauldin, the author of the widely read newsletter Thoughts from the Frontline, reflects that previous employment disruptions brought on by technology took decades or more to take root, which somewhat mitigated the employment pain, allowing some people the option of retraining and reassignment.

He wrote: “In 1800, many jobs were agriculture-related. It was still very high 50-100 years later when industrialisation really began to kick in. But we had four generations to transition from farm jobs to factories and other businesses.” The same was true within many other industries in our more recent history, including assembly line workers, telephone operators. typists, typesetters, etc, albeit with somewhat shorter timeframes. New jobs and industries were eventually created and the disruptions to overall employment were minor. 

But, as they say, this time is different. Most alarming is the speed with which the new intelligent agents are being developed — in areas such as healthcare, writing, graphic arts, agriculture, customer service, manufacturing, education, finance and software development. We won’t have decades to prepare, skill up and restart, and protective regulation cannot possibly be implemented at anywhere near the speed of AI innovation.

Displacement has already started in many industries (I have written about the Swedish fintech company Klarna, which laid off 700 employees in favour of an AI replacement) and it can only accelerate, along with the astonishing rate of innovation in AI. 

Employers are interested in two things above all else — profit and their ability to compete. They may say comforting words about their commitment to their staff but, if they start underperforming in relation to their competitors, they will do what is necessary. And what could be necessary, it seems, is the wholesale replacement of humans with AI. If companies don’t do it and their competitors do, they will be out of business. It will be a simple decision driven by the mathematics of capitalism. Venture capitalist Joe Lonsdale makes the point succinctly: “Any business that can triple its profit margins by implementing AI systems is obviously going to do it as fast as possible.” 

Read more in Daily Maverick: AI shaping up to become the greatest geopolitical weapon in history

So, back to our scenario in which most people lose their jobs, but let’s also assume that goods and services fuelled by AI become so cheap to make, produce and distribute that the unemployed are able to sustain themselves quite nicely on state grants, giving them time to write poetry, watch football or whatever. 

Which brings us to my final point. 

Who is going to pay taxes when the jobs go to AI? About 80% of state taxes come from individuals and payroll taxes. Only about 7% comes from corporate tax. And, if there are no taxes to be scooped into the fiscus, how is the government going to provide services for its citizens? 

The formalisation of national income tax is a fairly recent phenomenon, often traced back to the UK Income Tax Act of 1799. It has since become the core of government funding, its greatest source of revenue. Political systems cannot afford idle citizens. This may become the biggest problem for governments if there is little work left for humans to do. 

As I said, this is one possible outcome among many. Even so, it’s worth considering. DM 

Steven Boykey Sidley is a professor of practice at JBS, University of Johannesburg. His new book It’s Mine: How the Crypto Industry is Redefining Ownership is published by Maverick451 in South Africa and Legend Times Group in the UK/EU, available now.


Comments - Please in order to comment.

  • Steve Davidson says:

    “When AI takes all of the jobs, who will pay our taxes?”

    Who on earth says AI is going to do that? It’s just going to remove all the bulldust like Zuma’s Stalingrad tactics for a start. In many ways it’s going to free up a lot of human minds – that are way, way ahead of anything a computer could think up – to launch a whole new wave of our progress. It’ll especially shut up so-called ‘experts’ like you, for a start.

    • Dewald Snyman says:

      Hi Steve – Have a look at the projections done by AI experts such as Grace et al from earlier this year (arXiv; 2401.02843). They ran projections of when AI would be able to do something at the level of a trained human and start to replace people on various levels of education. See also the reference that Steven made to Klarna which one of the best examples to date. Every single one of us will be impacted, some earlier than others. It is a real issue that Steven is raising awareness of and at the root of this discussion is not only who will be left responsible for taxes, but also in which countries these taxes are to be paid.

  • Mark Pucjlowski says:

    My question, if everything goes to AI and nobody’s working, who will be able to afford what gets produced, eventually factories will stop producing with no sales and then what?

    • Thinker and Doer says:

      An extremely important question, that is not even discussed in thus article. There is tall about having some basic Universal Income that would be given to all of unemployed people, but that is certainly not feasible for a substantial proportion of countries, if any. The implications of the vast increase in unemployment have not been properly assessed, but the technological push to make everyone redundant on the basis of “cost savings” is gathering pace, unrestrained. The economic consequences will be devastating, but those driving the push just don’t care.

  • Stuart Hulley-Miller says:

    Ever since the Brits started income tax to finance wars/colony’s/etc, countries have been expanding its use and application, mostly annoying the ‘people’. That income for them is needed is not in doubt. Maybe it’s now time to re-vamp this crazy and unequal system by getting the money directly from the people who need whatever is being ‘paid’ for. No stupid VAT sums or the like. Keep it simple. If you want to use something, pay. If you want to buy something, pay. If you don’t spend or use …… you don’t pay.
    I’m a simple farmer so you clever guys can ask AI how to do this.

    • The idea of a user-pays system, where individuals pay directly for the services they use, has merit in terms of fairness and simplicity. However, implementing such a system comprehensively could be challenging. Essential services like public health and education might suffer as not everyone can afford to pay directly, leading to increased inequality and social stratification.

      The current tax system, including VAT, can indeed be complex and burdensome. However, VAT is designed to tax consumption and is easier to collect compared to income tax. While simplifying the tax code is a desirable goal, completely eliminating VAT might not be feasible without a robust alternative.

  • 1. Misconception of Tax Revenue Evaporation:

    Sidley posits that the mass displacement of human workers by AI will lead to a catastrophic loss of tax revenue, primarily because individual and payroll taxes constitute about 80% of state revenues. However, this perspective fails to account for the increase in corporate profits resulting from AI-driven efficiencies. If companies replace human labor with AI, their operational costs decrease, leading to higher profit margins. These increased profits should, in theory, result in higher corporate income tax (CIT) revenues.

    2. Corporate Profitability and CIT:

    While it is true that corporate taxes currently constitute only about 7% of total tax revenues, the significant increase in profitability due to AI could change this balance. Companies experiencing windfall profits from AI-driven productivity gains would pay more in CIT, potentially compensating for the reduction in individual and payroll taxes. The assumption that tax revenues would evaporate into thin air overlooks the natural economic shift towards higher corporate tax contributions in an AI-centric economy

  • Daniel Cohen says:

    If companies are going to make megaprofits at the expense of employment then the major tax burden will fall on companies, owners and shareholders which the state will have to redistribute to individuals (in return for community work or whatever)

  • Temba Morewa says:

    So what?
    Most of the tax money is stolen or wasted by our governments! As the tax spending will need to be more efficient they will be affected much more than anyone else!

  • Bob Dubery says:

    Increased company tax and/or a tax on computer time used. This can be audited! It already happens in F1 motor racing where the teams are restricted in their use of CFD computing and this is done on the basis of CPU useage across a season.

  • Patterson Alan John says:

    When AI replaces all the donkey work and those people are unemployed, the premise that they can be up-skilled is perhaps reaching too far. As the article states, the time-frame will be very short and will those people have the ability to do far more skilled and sophisticated work when given the opportunity?
    In consideration of the many millions of currently unemployed people, low education standards and ongoing demise of many manual jobs through, for example, warehousing robotics, AI will exacerbate this situation. There is evidence of AI replacing many activities already and people transitioning to new roles, but most of these are people already have good skills that can be transferred.
    I foresee enormous disruption and with a growing world population, predominantly in developing countries, the consequences will be devastating. The poor are about to be served a technology that will condemn them to permanent, miserable poverty and hardship, let alone those who are now employed, who will be replaced by AI and not be able to find any work in workplaces that have shrinking job numbers.
    We may be faced with sidelining mechanical equipment in favour of manual labour returning to digging trenches and holes, in the desperate effort to offer the unemployed masses an opportunity to earn a living.

  • Dana van der Merwe says:

    Interesting and fresh fiscal insight. New legal careers, as yet not conceived, will probably develop with scientific advances.

  • Chris Brand says:

    Guys and Gals, its time to be more creative, rather than just looking at the dark side of what a good focused release of really advanced AI (I have tested 7 different variants of AI myself with a few scenarios and so far all have failed to “pass” – yes I accept only 95% as sufficient, i.e. not 30% as the schools does in the RSA).
    Step 1 (or highest priority): focus the AI Improvement to replace repetitive tasks performed by all politicians currently, (with their staff complement included), to prioritize how much money and on what “prioritized projects” it must be spent – based upon a standardized formula (nearly a Maslow hierarchy) where Life/Health are the highest priority, followed by food production (agricultural farmers), followed by Housing (including electricity and water), followed by infrastructure (roads, rails, buses, LED Streetlights), followed by simplified laws replacing most if not all current laws, Safety (gated communities with security, etc) – funds for these to come from salary and perk savings for reduction of political staff; and
    Step 2 (at same time as Step 1): reduction of all redundant municipal staff, redundant SOE’s staff, by implementing online facilities at libraries (two per city and a per town [suggestion]) where citizens can log requests for services, mentioned in Step 1, to be prioritized, based upon a consensus of citizens (only ID Bearing) in a voting process annually [suggestion] to determine weight of each “service” to be supplied, etc.

Please peer review 3 community comments before your comment can be posted


This article is free to read.

Sign up for free or sign in to continue reading.

Unlike our competitors, we don’t force you to pay to read the news but we do need your email address to make your experience better.

Nearly there! Create a password to finish signing up with us:

Please enter your password or get a sign in link if you’ve forgotten

Open Sesame! Thanks for signing up.

A South African Hero: You

There’s a 99.7% chance that this isn’t for you. Only 0.3% of our readers have responded to this call for action.

Those 0.3% of our readers are our hidden heroes, who are fuelling our work and impacting the lives of every South African in doing so. They’re the people who contribute to keep Daily Maverick free for all, including you.

The equation is quite simple: the more members we have, the more reporting and investigations we can do, and the greater the impact on the country.

Be part of that 0.3%. Be a Maverick. Be a Maverick Insider.

Support Daily Maverick→
Payment options

It'Mine: How the Crypto Industry is Redefining Ownership

There must be more to blockchains than just Bitcoin.

There is. And it's coming to a future near you soon.

It's Mine is an entertaining and accessible look at how Bitcoin made its mark, how it all works and how it challenges our long-held beliefs, from renowned expert and frequent Daily Maverick contributor Steven Boykey Sidley.

MavericKids vol 3

How can a child learn to read if they don't have a book?

81% of South African children aged 10 can't read for meaning. You can help by pre-ordering a copy of MavericKids.

For every copy sold we will donate a copy to Gift of The Givers for children in need of reading support.