The gap is widening between the speed of development of artificial intelligence and the ability of governments to build rules governing its use, at a time when the technology has become one of the largest areas of global investment and a potential engine of productivity and growth, which makes regulating it an economic equation that goes beyond the issue of safety alone.
Strict rules may raise the cost of developing prototypes, slow down the rollout of applications, and burden startups, but weak oversight also carries a price tag that may appear in fraud, cyberattacks, automated errors, loss of trust, labor market disruptions, and risks that can spread through the financial system.
The equation becomes more difficult as competition intensifies between the United States and China , and regulatory models vary globally, while the European Union tries to build a framework based on levels of risk, and other economies seek rules that protect users without losing investment and technology.
Risk bill
The debate gained new momentum after officials and researchers at artificial intelligence companies warned of the risks of more autonomous systems, coinciding with the disclosure of incidents related to the use of advanced models in cyberattacks and their increasing ability to perform multi-step tasks with minimal human intervention.
This prompted the United Nations High Commissioner for Human Rights, Volker Türk, to deem voluntary self-regulation by companies insufficient, calling for the reporting of serious incidents, verification of the capabilities of models, and enhanced oversight of companies that develop the most advanced systems.
But the risks facing the economy do not need a scenario in which artificial intelligence reaches a level that threatens humanity to become costly.
Ziad Youssef, a professor and consultant in digital and intelligent systems engineering, told Al Jazeera Net that the first of these risks is related to cybersecurity and fraud, as artificial intelligence can expand the ability to produce malware, impersonate voices and identities, and carry out attacks, and banks, insurance companies, consumers, and infrastructure facilities may bear the costs of these attacks.
It points to another risk: the widespread repetition of a wrong decision. If a bank adopts an uncontrolled model for assessing credit, or if a health institution or company uses it in decisions affecting large numbers of people, an individual error can turn into thousands or millions of automated decisions.
Tags:
technology
