Citrini’s dystopian AI narrative may be wrong, but it could catalyse deeper thinking, policy change

Citrini Research’s imaginary memo from the year 2028, which pretends to recount how the widespread adoption of artificial intelligence (AI) caused a surge in white-collar unemployment and economic turmoil in the United States, was met by a barrage of criticism last week.

While I found myself mostly agreeing with Citrini’s detractors, I wondered if the pushback would have been quite as aggressive if the little known research firm’s doomsday narrative from the future hadn’t sparked the unnerving sell-off that it did.

Just about every company mentioned in the fictional memo saw their shares wobble early last week – from software-as-a-service (SaaS) platforms such ServiceNow and Salesforce to payments giants Visa and Mastercard and even asset managers Blackstone and KKR.

For many investors, the AI theme has so far been largely about the massive demand for chips and data centre capacity, which have spurred economic activity as well as stock prices. The main perceived risk was that some of the US mega-caps might be overinvesting in the AI space.

Yet, even with the recent souring of sentiment towards some of these heavyweight counters – such as Amazon and Microsoft – the S&P 500 is up 0.5 per cent since the beginning of this year, and up 38.1 per cent since its post-Liberation Day low in April last year.

With no actual let-up in AI spending, investors are still following the money. Notably, the chipmakers as a group are still marching higher. The iShares Semiconductor ETF – which counts Micron Technology and Applied Materials among its biggest holdings – has gained 17 per cent since the beginning of this year, and 127.5 per cent from its April 2025 low.

Over the past several weeks, however, the AI theme has evolved. Instead of only worrying about over-investment, many investors are now also becoming anxious about AI adoption potentially destroying existing businesses. In particular, software companies have suffered big sell-offs on concerns that the emergence of AI agents may unravel their businesses.

The iShares Expanded Tech-Software Sector ETF – which counts Microsoft, Palantir Technologies, Salesforce and Oracle among its largest holdings – has dived 22.8 per cent since the beginning of the year. The ETF is now trading only 1.8 per cent above its post Liberation Day low.

Whatever criticisms one may have about the logic and reasoning behind Citrini’s imaginary warning from the future, it seems to have appeared at just the right moment to catalyse deeper thinking about what AI means for investors, businesses and governments.

“Negative feedback loop”

The main arc of Citrini’s story goes something like this: As AI became better and cheaper, companies were incentivised to lay off employees and use the savings to enhance their AI capabilities, which led to them laying off even more employees.

The displaced workers spent less, causing a slump in profitability at companies that sold things to them. Those companies, in turn, laid off workers and invested more in AI to lower their costs.

Citrini described this as “a feedback loop with no natural brake”.

In its memo from the future, Citrini said it all began in the software sector, where AI agents enabled companies to replicate the core functionality of SaaS products that they used quickly and cheaply. But AI agents were soon also helping consumers book their holidays, renew their insurance policies and order food – compressing margins of established players in all these sectors.

The AI agents also looked for faster and cheaper ways to settle these transactions than through traditional credit cards, and most eventually turned to cryptocurrencies.

Within a couple of years, reduced profitability at SaaS companies, rising unemployment, and falling wages began to pressure the financial sector.

Meanwhile, the US government struggled to manage the fallout. Its traditional monetary policy tools of rate cuts and quantitative easing weren’t suitable. The real economy wasn’t being pressured by tight financial conditions, but by AI making human intelligence less scarce and less valuable.

Opening the fiscal spigots was problematic too, because of falling income tax receipts.

In the end, the US government eventually began exploring ways to tax the use of AI, and establish a public claim on the returns from AI infrastructure.

The research firm rounded off its missive from the future by exhorting the government to move fast, and take some chances. “Nobody’s framework fits, because none were designed for a world where the scarce input (of human intelligence) became abundant. So we have to make new frameworks,” it said.

“Whether we build them in time is the only question that matters,” it added.

Higher output, lower consumption?

Among the counter-arguments that have been put forward over the past week are that the displacement of white collar workers by AI would probably be constrained by limited data centre and energy capacity, as well as regulatory intervention.

Then, there is the difficulty of reconciling a positive productivity shock from AI with the fall in consumption that Citrini envisions. “If AI delivers big productivity gains, it will reduce prices and raise real income in sectors that aren’t displaced, causing other Americans to spend more,” said famed economist Paul Krugman, in a Substack post last week.

He added, “There’s no reason to believe that disrupting part of the economy will reduce overall demand.”

None of this means that AI will not wipe out certain sectors over time, of course. “When an industry suddenly collapses, people get hurt. Investors lose their money, workers lose their jobs and in some cases their whole careers,” said Krugman, in his Substack post.

“But does technology-driven industrial disruption cause financial and economic crises, the way Citrini Research predicts? I can’t come up with any examples.”

Wealth taxes in Singapore?

The way I see it, Singapore may well turn out to be an interesting case study on the macroeconomic impact of AI.

On the one hand, AI-driven automation in the services sector could well enable the Singapore economy to expand at a faster pace while relying less on foreign white collar workers. On the other hand, the same AI technologies that reduce the need for foreign talent may well put downward pressure on the pay scales of local white collar workers.

Last week, during his Budget round up speech, Prime Minister Lawrence Wong noted that Singapore’s economic growth does not create as many jobs as it once did, in part because its fastest growing sectors increasingly rely on automation.

This calls for policies to raise productivity, wages and career pathways across the services sector, he added. “And that’s why we are taking sustained action in areas like education, healthcare, and social services.”

This approach could help Singapore’s labour force cope with potential displacement and wage pressure as AI adoption grows, in my view.

What about people at the top of fields such as finance, law and medicine? Do the high-value, bespoke services they provide put them beyond the reach of AI disruption?

Some of them may well like to think so. Yet, if AI does end up depressing incomes from white collar jobs, it seems likely to me that the government will shift its approach to harvesting tax revenue. The release last month of household wealth data, and the inclusion of non-employment income in measuring inequality, suggest that wealth taxes may rise over time.

The bottom line is that we are all about to feel the impact of AI, one way or another.