[Chronique] A tax that will be unanimous

Death, taxes and… robots? We rarely come across a topic that everyone agrees on, but everyone seems to agree on the inevitability of these three. Is the impact of generative AIs like ChatGPT also inevitable, or can we already imagine governments taxing their work to benefit the population?

The formula is already found. It was also part of the Green Party of Canada’s election platform during the 2019 election campaign. Basically: companies that use AI to automate tasks should calculate the remuneration that would normally go to a human who accomplishes this work and then remit to the State the equivalent of the tax on this remuneration.

Four years later, AI experts who are calling for a moratorium of at least six months on the research and development of natural language understanding and production systems at least as sophisticated as California-based OpenAI’s GPT-4 are set to change. their speech to model that of the greens.

They would only have to quote the founder of Microsoft, Bill Gates, who these days looks like a prophet. In 2017, he proposed a tax on AI to governments that want to slow down the emergence of the technology a little to better absorb its transformative social impact.

Bill Gates said six years ago: “A worker whose job earns him, say, $50,000 pays taxes on that income, pays for social measures, like employment insurance, and all those things. If a robot manages to perform the same tasks, one can imagine that one should impose this robot at a similar level. »

Gates imagined at the time that the money raised in this way could be used to strengthen the education and health systems in particular, and to make society have more empathy for the people most likely to be left behind. by the anticipated automation of the labor market.

6650 billion good reasons

Because this automation could hit hard. It could also pay off big. The financial gains promised by the mass adoption of the next generations of AI are in the trillions: at least US$6.65 trillion by 2030, according to Goldman Sachs. It remains to be seen to whom all this money will go.

Workers may be the ones who will see the least benefit. The American firm Goldman Sachs calculated last month that the generative artificial intelligence and extended language models that drive applications like ChatGPT threaten the existence of some 300 million jobs worldwide. Jobs mainly located in advanced economies, such as Canada, the United States and Europe.

Historically, technological revolutions have ended up creating more jobs than they have eliminated. Since the industrial revolution, they have certainly wiped out entire professions, but over time people who previously might have opted for one of those jobs that ended up being automated ended up finding another specialization.

Living conditions have improved as technology has increased the overall productivity of workers and businesses. However, the most recent digital technologies could break with this tradition, and this is what worries some economists.

AI to the rescue of the tax authorities

In Silicon Valley, where much of the innovation in AI is emerging today, the idea of ​​a guaranteed minimum income is being pushed by the biggest promoters of technological innovation. This includes libertarian Elon Musk. One of the measures devised to finance this expenditure is a tax on robots.

The European Union has considered taxing AI, but is hesitant, saying it would hurt innovation. South Korea has already implemented such a tax and is not faring so badly.

One thing is sure, the relationship between the taxman and the robots will quickly become more and more intimate. How ? The very prestigious Johns Hopkins University, located in Maryland, has set up a research program whose sole task will be to create an AI capable of finding and then exploiting all the tax loopholes that allow American companies and taxpayers who have them ways to avoid paying income tax.

The University believes that such a system will eventually allow the Internal Revenue Service (IRS), the American tax authorities, to identify cases of tax evasion which, according to its own calculations, would cause it to lose more than US$500 billion. every year.

We do not know if this is good news, but the project would be more complicated than it seems. The reason ? Taxation as it exists is too complex for the most advanced AI on the market today.

Tax professionals can rest easy. The AI ​​is still unable to complete our tax returns. It does not seem close to replacing them. Maybe, if we tax her, she won’t be able to escape… unless she hires a good accountant!

To see in video


source site-41