The developer argues that the latest AI models are no longer merely useful assistants, but are capable of carrying out complex tasks autonomously and even refining their own performance. From this, he infers that most cognitive and office-based professions across a wide range of sectors face either extinction or radical transformation in the coming years.

Predictions of sweeping job losses have multiplied of late. Mustafa Suleyman, executive vice-president for AI at Microsoft, has suggested that artificial intelligence could replace the majority of white-collar roles within 12 to 18 months. He is far from alone.

Dario Amodei, chief executive of Anthropic, has stated that AI will eliminate half of all entry-level jobs within five years. The list of those advancing similarly apocalyptic forecasts is long.

It is hardly incidental that many of the loudest voices are employed by companies for which the rapid diffusion of artificial intelligence is commercially decisive. The pattern recalls the period some two decades ago when numerous figures in the oil industry warned that global reserves were on the verge of depletion. At the time, ‘peak oil’ – the point at which extraction of the so-called black gold would reach its historic maximum before entering irreversible decline – was said to be imminent.

Shumer’s text is carefully constructed. It unsettles the reader, only to present a proposed remedy. The answer, however, is not restraint or tighter regulation of artificial intelligence, as some might assume, but rather its fullest possible deployment. Those wishing to remain competitive in the new environment are urged to pay $20 a month for a subscription and to master the tools.

Relying on the basic free version, he suggests, will not suffice. Nor is that the end of the argument. Artificial intelligence, in this view, must be integrated comprehensively into daily work rather than used merely as an enhanced search engine. In practice, that means supplying models with ever larger volumes of data so that their capabilities can continue to expand.

Emotions instead of rational argumentation

The debate surrounding artificial intelligence is beginning to resemble earlier moments of collective tension, such as the Covid pandemic or the war in Ukraine. The phenomena differ profoundly in substance, yet their social reception displays striking parallels. In each case, public discourse tends to present a single allegedly rational course of action – here, the unconditional and rapid expansion of artificial intelligence. Nuance gives way to a stark either–or.

Some conspiracy-minded observers interpret AI as yet another instrument in a grand design by elites to tighten their grip on society. There is, however, a far more prosaic explanation rooted in social psychology and marketing.

Thinkers associated with the Palo Alto school of communication theory formulated a basic principle of human interaction: every act of communication operates on two levels – content and relationship. Facts belong to the first, emotions to the second. In practice, the emotional dimension tends to override the factual. Modern mass marketing has internalised this insight. The decisive factor in selling a product is not rational argument but the evocation and management of feeling.

Regrettably, that commercial logic has increasingly shaped political communication. In recent years the trend has intensified. Detailed policy programmes have largely given way to clusters of resonant slogans designed to stir voters’ emotions. Concrete pathways for achieving declared aims are often conspicuously absent.

Society has absorbed that style. Artificial intelligence is now discussed less in analytical terms than in emotional ones. On one side stands a paralysing fear of job losses, while on the other there is a buoyant euphoria promising rapid gains and a swift ascent up the social ladder through AI.

The difficulty is that many of the loudest advocates are themselves products of this system. Matt Shumer provides an instructive example. By invoking comparisons with a pandemic, presenting himself as an insider and playing on the fear of missing out, he amplifies a sense of urgency and presses for immediate action. In time, however, he too will be compelled to adjust to the very forces he promotes.

AI as the final nail in the coffin of liberalism

The social impact of artificial intelligence may prove very different from what many now assume. Rather than heralding a triumphant new order, it could represent the terminal phase of the prevailing system commonly described as neoliberal capitalism. The French philosopher Barbara Stiegler distils the logic of that system into a single imperative: ‘we must adapt’. The injunction presents itself as a neutral description of reality, yet it functions as a norm prescribing how individuals ought to live and perform.

Artificial intelligence can be seen as the most complete realisation of that demand, because the contest it implies cannot be won. No individual can compete with humanoid machines endowed with cognitive capacities rivalled only by multiple doctorates, capable of composing ancient verse while repairing a heating system. The image is that of a Sisyphean struggle.

The issue is not that human beings are doomed in every respect, but that responsibility is progressively transferred to the individual. The central question ceases to be which institutions, corporations and regulatory frameworks shape work and life. Instead, it becomes whether you can personally keep pace with perpetual upgrades – new workflows, new routines, new disciplines.

For that reason the mantra ‘you can’t beat AI’ is so potent. It seeks to close the argument before it even begins and to replace politics with a psychology of adaptability. The issue is not ‘let us halt technology’. It is to question the imperative that presents adaptation as unavoidable. Under that logic, innovation becomes a lash. The message is clear: automate or vanish, optimise or fail, adjust to the system’s tempo rather than require the system to answer to human needs.

The decisive question, then, is not whether artificial intelligence will be adopted, but who defines its purpose, its speed and its limits. Will it continue to generate compliant and flexible actors for an intensified competitive order, or can it be subordinated to the public good? For Stiegler, the entire doctrine of ‘adaptation’ reduces to a single proposition: not man for the environment, but the environment for man.

Thus we return to the apparent dissonance at the start of the year. While sections of the media cast AI as an inexorable force destined to overwhelm society, financial markets tell a quieter story of costs, borrowing and uncertain returns.

It is precisely in that tension that the meaning becomes clear. When we hear that ‘artificial intelligence is everywhere’, it does not mean that AI is unavoidable. It means that the command to adapt is being treated as a law of nature, so that a political decision is transformed into a purely individual obligation.