Diving overview:
- Rogue or weaponized artificial intelligence has a 10% chance of causing 1 million deaths, at least $100 billion in losses, or other “catastrophic harm” by 2030, with financial sectors being particularly vulnerable, according to a study based on estimates from 272 AI experts around the world.
- AI introduced in the financial sector could spark widespread adoption Losses due to turbocharger function A study by MIT FutureTech and the University of Queensland found that market manipulation, fraud, cyberattacks, computer sabotage and privacy violations were to blame.
- “Dangerous capabilities, weapons and cyberattacks, centralization of power, inequality and unemployment, and environmental damage have a greater than 10% chance of having catastrophic outcomes, even with realistic mitigation measures in place,” the researchers said. Even with mitigation measures in place, experts have determined that there is at least a 5% chance of a catastrophic outcome across 24 “risk areas.”
Dive Insight:
TThe MIT/Queensland study’s findings echo other studies in recent years that have raised alarms about the risks of AI to humanity, sparking public opposition and prompting steps toward regulation.
For example, AI experts who participated in the 2022 Existential Risk Persuasion Tournament estimated a 3% chance that: AI will lead humanity to extinction.
The MIT/Queensland study says the biggest risk lies in the prospect of AI acquiring the capacity for large-scale deception, weapons development, public manipulation, political abuse and cyber-attacks by 2030.
Other significant risks also loom large, the study said, including the possibility that nations and companies in the “AI race” may be trying to gain an advantage by rushing to deploy systems that have not been adequately checked for errors or security threats.
Research shows that malicious actors could use AI to launch cyberattacks and create chemical, biological, radiological, and nuclear weapons.
According to the study, AI could concentrate power and resources, leading to an unfair distribution of benefits and social inequality, or could accidentally generate false or deceptive information, distort people’s perceptions, and cause physical, emotional, or material harm.
Efforts to reduce AI risks are facing obstacles. According to the researchers, AI users and stakeholders are more vulnerable to a variety of threats than the AI developers, regulators, legislators, and standard setters who have the power to reduce risk.
“Computing, cloud, and data infrastructure providers are largely insulated from many of the risks of AI,” they said.
The vulnerability gap creates “misaligned incentives in addressing the most important AI risks,” according to the study.
“Even well-understood risks can still persist if responsibility is dispersed, competencies are disconnected, or disconnected from those who bear the harm,” the researchers said.
It added: “Commentators argue that the new regulatory landscape resembles a patchwork rather than a coordinated framework, and it remains unclear which risks governments should prioritize.”
