Photo credit: Getty Images
The development of artificial intelligence is fuelling fears about the future — according to surveys, more than half of humanity already fears it.
After AI ‘spun out of control’ first at Anthropic and then at OpenAI, the issue of safety has become paramount. Experts agree that the greatest threat in the event of an ‘AI apocalypse’ lies precisely in the devastating consequences of the uncontrolled use of artificial intelligence.
The probability of such a scenario is currently estimated at 10 per cent. If, however, the development of AI is not restricted or regulated, this figure rises to over 20 per cent.
This is evidenced by the results of a survey of 272 artificial intelligence experts conducted by the Massachusetts Institute of Technology (MIT) – one of the world’s leading technical universities.
The experts ranked 24 risks and determined for each the probability of a ‘catastrophic’ outcome by 2030. By catastrophic consequences, they meant, for example, the loss of over 1 million lives, economic damage exceeding $100 billion, or intangible consequences on a civilisational scale — such as the collapse of democratic norms or the right to privacy.
One of the most widely discussed fears – the threat of mass unemployment – did not feature among the top three risks and ranked fourth in terms of the severity of potential consequences. Instead, first place went to a ‘Terminator’-style scenario.
According to the experts, the greatest danger is artificial intelligence that spirals out of control and learns independently to inflict large-scale damage on humanity.
This could happen due to technical failures, malicious acts or because ‘the AI’s goals do not align with human interests’ – this is precisely how MIT describes the main risk.
What kind of harm could such AI cause?
We are talking about deception, the development of weapons, the manipulation of people, interference in politics, cyberattacks, as well as the creation of more advanced artificial intelligence systems — to the point where they become aware of their own operating environment and are capable of self-replication.
Who is responsible for this?
The second and third places in the ranking of apocalyptic risks are also linked to the first.
These concern the development of weapons using AI — both for cyberattacks and for the mass destruction of people — as well as the ‘AI power vertical’ — the concentration of resources in the hands of a small circle of people with access to cutting-edge technologies, and the deepening of social and economic inequality.
All of this could lead to social upheaval and political crises. Surveys conducted in various countries around the world indicate that fears of such consequences are growing year on year.
Photo credit: EPA
For example, according to the latest global study by the Stanford AI Index, by 2025 the proportion of respondents who find AI ‘alarming’ had risen to 52 per cent, compared with 39 per cent in 2022. At the same time, the proportion of those inspired by the development of artificial intelligence has remained virtually unchanged at 53–54 per cent.
However, growing public concern is not leading to the introduction of any significant restrictions on AI, note the authors of the MIT study.
After all, decisions are made by those who benefit from the development of artificial intelligence, whilst the people whose lives it is most likely to change or worsen have virtually no influence on this process.
‘The gap between vulnerability and responsibility is not a new phenomenon. For example, it is the general public who suffer most from aviation and industrial disasters or from substandard medicines, whilst the responsibility for preventing such consequences lies with developers, manufacturers and regulators,” write the authors of the study.
‘However, these sectors are governed by strict standards, control mechanisms and legal liability. In the field of artificial intelligence, similar mechanisms are either only just beginning to take shape or are entirely absent. As a result, the most vulnerable groups of the population remain virtually defenceless, whilst those responsible for their protection lack sufficient incentives.”
Where, then, should one work?
Not all sectors are equally vulnerable to the risks associated with artificial intelligence.
The same experts assessed the level of vulnerability of 14 economic sectors to each of 24 identified risks. The information technology sector and the national security sector fared the worst.
The IT sector suffers most from disinformation, the manipulation of public opinion and interference in private life. National security is particularly vulnerable to cyber-attacks and the emergence of new types of weaponry.
The financial sector received similar assessments. On the one hand, AI complicates the fight against fraud and market manipulation; on the other, it is being increasingly implemented at all levels, which heightens the risk of serious consequences should it malfunction.
According to the analysis, the sectors found to be least vulnerable are those where artificial intelligence is not yet used as extensively — the hospitality and catering sector, agriculture, the manufacturing industry, and the culture and entertainment sector.