Il Nuovo Postiglione - AI's 18-month Job disruption

In February 2026, Microsoft’s newly appointed chief executive of artificial intelligence, Mustafa Suleyman, told the Financial Times that AI systems could soon perform “human‑level performance on most, if not all professional tasks”. He argued that the rapid growth of computational power would enable machines to automate any task performed by someone sitting at a computer — a lawyer drafting a contract, an accountant balancing a ledger or a marketing manager running a campaign. According to Suleyman, many such tasks would be fully automated within 12 to 18 months. The Microsoft executive cited the ability of large language models to write code better than most human coders and said that creating bespoke AI models would soon be as easy as starting a podcast or writing a blog.

His pronouncement was one of the most dramatic in a wave of tech‑executive warnings. Anthropic co‑founder Dario Amodei said last year that AI could eliminate half of all entry‑level white‑collar jobs within five years, while Ford chief executive Jim Farley suggested that the technology could drastically shrink white‑collar employment. AI researcher Matt Shumer compared the current moment to early 2020, when the pandemic’s economic shock had not yet fully registered. Critics, meanwhile, noted that similar predictions have been made repeatedly; some viewers of Suleyman’s interview remarked that they had heard the same 18‑month warning before, and others argued that if AI is truly so disruptive it should replace top executives first.

Evidence versus alarmism
Despite Suleyman’s dire timeline, research suggests only limited disruption so far. A 2025 Thomson Reuters report on professional services found that lawyers, accountants and auditors mainly use AI for targeted tasks such as document review and routine analysis, yielding only marginal productivity improvements. Some studies even report a negative impact: a Model Evaluation and Threat Research (METR) experiment on experienced software developers found that using a popular AI coding assistant increased task completion time by 19 %, because programmers spent additional time correcting the model’s suggestions. Other research has demonstrated speed‑ups in specific contexts, but the METR authors caution that these gains do not generalize to all code‑bases. In the broader economy, profits remain concentrated. Data from Apollo Global Management showed that Big Tech profit margins rose more than 20 % in late 2025, while the wider Bloomberg 500 index saw little change. Wall Street analysts thus doubt that AI will deliver higher earnings outside the tech sector.

Hiring data also temper the narrative. Employment consultancy Challenger, Gray & Christmas recorded about 55,000 job cuts attributed to AI in 2025. Microsoft itself eliminated 15,000 jobs last year, though it did not directly link those reductions to automation. Some industry observers believe executives are using AI hype to justify traditional cost‑cutting; user comments on social media argued that businesses often announce AI‑driven layoffs to distract from poor financial performance, and several commenters questioned who would purchase goods and services if most people were unemployed.

Economic and political reactions
Suleyman’s remarks provoked a fast response from policy‑makers. U.S. senator Bernie Sanders called the prediction an “economic earthquake” and urged a moratorium on new AI data centers so that the technology benefits workers rather than a handful of billionaires. Lawmakers in several states have already campaigned against the energy demands of AI facilities, and the issue has become politicised during the U.S. presidential race. Even Microsoft’s overall chief executive Satya Nadella has warned that the industry must earn the “social permission” to consume vast amounts of electricity. In an interview, Nadella said that AI companies need to show they are “doing good in the world” or risk a public backlash over energy use. He added that AI’s benefits must be widely shared and not confined to a few companies or regions.

Financial markets have reacted nervously. Concerns about automation drove a recent sell‑off in software stocks, dubbed the “SaaSpocalypse,” after Anthropic and OpenAI unveiled agentic AI systems capable of performing many software‑as‑a‑service functions. Analysts observed that the sell‑off reflected fear rather than current impact; AI products such as Microsoft’s Copilot are still in the early stages of adoption, and there are significant hurdles to full automation. Experts note that successful deployment requires training, redesigned workflows and reliable AI agents, and many organisations are far from achieving those prerequisites. Paul Roetzer, founder of the Marketing AI Institute, argued that displacement will be constrained by the difficulty of integrating AI into existing systems.

Social response and ethical questions
Public reaction to the 18‑month forecast has been mixed. Some see AI as a new industrial revolution that could free people from drudgery, while others fear widespread unemployment and social upheaval. Online comments on the interview reveal a deep scepticism: viewers joked that by the time AI automates marketing, it will also be cleaning toilets, and some called for a universal basic income to offset job losses. Others warned that if AI renders people jobless, the economy will collapse due to lack of consumers. A number of comments also highlighted that AI predictions often overlook who controls the technology; one observer noted that executive positions are rarely listed among the jobs that could be automated.

Ethical considerations extend beyond employment. AI’s energy appetite and the environmental costs of data centers have prompted demands for responsible innovation. Nadella’s plea for social licence underscores the need for transparent governance, equitable distribution of benefits and safeguards against monopolistic control. Advocates argue that if AI systems do not deliver tangible improvements in healthcare, education or climate resilience, the public may refuse to tolerate their resource consumption.

Looking forward
The gap between breathless forecasts and current reality suggests that the future of work will be more nuanced than a simple countdown to obsolescence. AI systems are undeniably accelerating, and many routine tasks will likely be automated. However, evidence points to augmentation rather than wholesale replacement. White‑collar roles that blend critical thinking, emotional intelligence and domain expertise are proving harder to replicate than anticipated. Meanwhile, new opportunities are emerging for workers who can supervise AI, curate data and integrate automated outputs into complex processes. Rather than fearing an AI takeover, experts advocate investment in education, reskilling and social safety nets so that labour markets can adapt.

The next 18 months will reveal whether Suleyman’s prediction was prescient or hyperbole. What is clear is that artificial intelligence has entered a phase of rapid experimentation. The challenge now is to ensure that the technology develops in a way that enhances human welfare, spreads prosperity and respects the planet’s finite resources.