Joshua Bengio and LawZero: why fear of future AI distracts from today's threats

The debate over the future of artificial intelligence increasingly revolves around scenarios in which machines one day become intelligent and slip out of control. But the main thesis of this text is different: there is no need to wait for hypothetical superintelligence, because the key risks of AI have already manifested in real economics and politics. This is not about machine uprising, but about concentration of power, surveillance, pressure on workers, and the use of technologies in the interests of large structures.

The author directly states that he does not consider current AI systems intelligent and sees no grounds for believing that impressive statistical models will automatically lead to genuine intelligence. Therefore, he perceives conversations in the spirit of "what to do if AI becomes conscious" either as convenient distraction or as a marketing frame that allows discussing impressive future catastrophes instead of uncomfortable problems of the present. Hence the reference to "Pascal's wager": fear of an unlikely but grandiose event begins to dictate the agenda more powerfully than already observed consequences.

At the same time, the author's position does not boil down to unconditional techno-optimism. On the contrary, he acknowledges sharing part of the concerns of so-called AI doomers, but sees the source of threat not in the awakening of machine intelligence, but in people and institutions that already manage these systems. He is concerned that digital tools are concentrated in the hands of corporations whose scale makes public oversight difficult.

He is concerned that technologies are used against users and workers: to intensify surveillance, automate pressure, tighten productivity metrics, and further skew the balance of power in favor of employers and the state. This is where the main conflict with the more futuristic school of AI safety emerges. The occasion for it was a public discussion in Montreal, where the author spoke alongside Astra Taylor and Joshua Bengio.

Bengio, one of the most authoritative researchers in deep learning, Turing Prize laureate, and scientist whose work helped shape the current AI boom. He is now participating in the LawZero initiative, which proposes creating an international consortium and developing AI as a public digital good: open, transparent, verifiable, and safe. For supporters of this approach, long-term risks are serious enough to build protective institutions in advance.

The author, however, disputes not so much the idea of taking precautions as the order of priorities. By his logic, discussion of hypothetical superintelligence too easily becomes a convenient abstraction, while existing systems of recognition, recommendation, generation, and labor evaluation are already embedded in mechanisms of control here and now. They influence hiring, productivity, moderation, access to information, and relations between state, business, and citizens.

When power is concentrated among technological giants, and the state gladly relies on their tools, the problem ceases to be science-fictional: it becomes political, social, and labor-related. This perspective is important also because it changes the language of the discussion itself. Instead of the question "how to save humanity from an autonomous supermachine," a more grounded set of topics emerges: who owns computational resources, who sets rules for data access, what audit mechanisms actually work, and who is held responsible for harm from automated decisions.

This is a less spectacular agenda than apocalyptic forecasts, but it is precisely this that determines whether technologies will intensify inequality or serve the public interest. And this is the practical conclusion. The debate over AI's future is important, but it should not substitute for discussion of model transparency, platform accountability, worker rights, and the limits of digital surveillance.

If public attention is focused only on the rare scenario of future catastrophe, corporations and authorities will gain even more space for solutions that worsen lives already today. The text offers a simple but uncomfortable optic: the main question is not whether AI will ever become too intelligent, but who uses current AI, in whose interests, and with what consequences for society.