As dystopian AI forecasts shake investor confidence, the real challenge lies in separating plausible risks from exaggerated economic fears
Two days after India concluded the AI Impact Summit, envisioning AI as a net good, a dystopian Substack post by Citrini Research on the future of AI shook stock markets. The capital management firm dives into scenario analysis and paints a rather gloomy picture of the future AI holds in store. It does not go far, only to 2028, to show that if the AI bull run is genuine, then there might exist a world that succumbs to automation without any augmentation. While this AI-dominant world seems far from Orwellian, the plight of workers might not be all too different. However, it is difficult to assign any merit to Citrini’s “thought exercise in financial history” without understanding the assumptions behind this A(I)pocalyptic world.
Although a scenario exercise and not a model-based prediction, a 2028 crisis caused by the rapid deployment of AI is presented, especially in white-collar work. It is envisaged as a left-tail macro risk that markets may be underpricing. The central macro idea is that AI will boost productivity, raise margins, and increase output, but the gains will flow to capital owners (or, in this case, compute owners) rather than households. Thus, consumer demand weakens even while headline output stays strong. Here, the economy keeps producing, but less income circulates through workers.
AI agents can reduce search and switching costs, undermining business models built on inertia, convenience, or information asymmetry.
This is propagated through a self-reinforcing displacement loop in which firms cut white-collar labour, use savings to buy more AI, which improves AI (or makes it cheaper), leading to more labour being replaced. Since AI expenditure is treated as a cost substitution (operating expenditure), AI investment can rise even as total spending falls. This can be corroborated through the SaaS (Software as a Service) repricing channel. Agentic coding makes in-house alternatives more credible, weakening SaaS pricing power and compressing margins. The apprehension is that this begins in software but then spreads to any business with a large white-collar cost structure.
Figure 1: The AI Feedback Loop
Source: Citrini Research
A second channel could be friction removal in consumer markets. Many consumer services today are based on the idea of eliminating market inefficiencies or information asymmetries (subscriptions, bookings, insurance renewals, some advisory functions, tolls, commissions, etc.). AI agents can reduce search and switching costs, undermining business models built on inertia, convenience, or information asymmetry.
Besides the usual focus on supply-side dynamics, there are repercussions for systemic macro demand risk. Since white-collar workers form a large share of discretionary spending, even modest income losses among earners can hit consumption disproportionately. This creates a delayed but deeper demand shock. The demand shock could extend to financial contagion via failures in private credit markets. It could mean that the financial system becomes essentially dysfunctional due to erroneous shifts in the underlying risk parameters.
Since white-collar workers form a large share of discretionary spending, even modest income losses among earners can hit consumption disproportionately.
The usual macroeconomic policy toolkit of rate cuts and quantitative easing may stabilise the financial system, but cannot solve the underlying problem if AI is structurally reducing labour’s income share. In line with international rhetoric, the default suggestion is for redistribution or transfer mechanisms tied to AI gains (including taxes on compute or public claims on AI infrastructure returns). The final conceptual takeaway is the unwinding of the intelligence premium. Modern institutions were built around scarce human intelligence, and AI is making that input less scarce. The central claim is that markets, labour systems, mortgages, and tax systems are all being repriced around that shift.
Although hypothetical and almost a worst-case scenario, this narrative did a tremendous job of unnerving investors. The S&P index dropped by 1 percent on February 23, the day the study was published. The software component of S&P experienced its worst performance since the Liberation Day tariff announcements of April 2025. In India, despite the optimism and jubilation following the AI Impact Summit, the NIFTY 50 dropped around 300 points on February 24. On the same day, the NIFTY IT index was the worst sectoral performer, losing 4.74 percent. Market volatility should not be a cause for concern; however, it is alarming that a speculative sketch can so easily undermine investor confidence in AI.
The usual macroeconomic policy toolkit of rate cuts and quantitative easing may stabilise the financial system but cannot solve the underlying problem if AI is structurally reducing labour’s income share.
Narrative economics, as posited by Nobel laureate Robert Shiller, is especially relevant in this context. Since markets, and in turn, the economy are functions of human sentiments and behaviour, popular narratives tend to play a significant role in determining structure and outcomes. Stock market fluctuations and global economic recessions can be traced back to the promulgation of certain narratives. This has manifested again in the immediate financial outflow from tech funds in response to the Citrini tale. Beyond causation, it also reflects the public apprehensions about the emergence of AI. It goes on to show that despite bullish trends, there is a degree of uncertainty driven by animal spirits in the AI boom.
There are multiple reasons why this scenario is far-fetched and should not have led to any mass hysteria. From a general equilibrium perspective, the scenario could emerge under very specific assumptions. First, labour income loss would have to exceed productivity gains, resulting in a weakening of demand. Second, as income shifts from labour to capital, consumption would weaken. Third, worker reallocation mechanisms would have to be rusty. Fourth, downward pressure on wages would eventually lead to a credit crisis. Finally, the Keynesian assumption of sticky wages and prices must hold.
In the absence of institutional reform and capacity-building, the displaced workers will have no option but to “eat the tech”.
Recent literature shows why these assumptions might be extreme or even empirically incorrect. Most importantly, while task displacement is real, AI's impact on labour demand remains limited. This has a cascading effect on the remaining assumptions. If there is no labour income loss, it is unlikely that capital income would rise disproportionately, thereby dampening consumption. Note that this does not mean the capital-labour income ratio remains static — it has been evolving for decades and is a characteristic of capitalism. If there is no income loss, the shock does not dissipate to the financial sector, and we avoid the non-cyclical disruption.
As a potential scenario, it is excessive. As a stress test, it is economically coherent. While some assumptions are manifesting through AI diffusion, task exposure, and distributional risk, the full cascade of mass unemployment, consumer collapse, mortgage contagion, and a systemic credit crisis is not yet supported by current macro data. The world will have to grapple with a future that lies somewhere between this nightmarish scenario and the buoyant global order envisioned at the AI Impact Summit. The study does not change anything but lays bare the deficiencies in the AI gold rush. Policymakers must treat new AI advances with a fistful of salt and stick to the established agenda of structural transformation that smoothens the shocks from tech disruptions. In the absence of institutional reform and capacity-building, displaced workers will have no option but to “eat the tech”.
Arya Roy Bardhan is a Junior Fellow with the Centre for New Economic Diplomacy at the Observer Research Foundation.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.
Arya Roy Bardhan is a Junior Fellow at the Centre for New Economic Diplomacy, Observer Research Foundation. His research interests lie in the fields of ...
Read More +
PREV
