Add as a preferred source on Google

Human Misuse Poses a Greater AI Risk Than Rogue Machines

Powerful systems become dangerous when people surrender oversight and use artificial intelligence to expand surveillance, manipulation and cybercrime.

A Nov. 18 commentary by John and Nisha Whitehead argues that the greatest danger from artificial intelligence (AI) comes from how people deploy powerful systems, rather than from machines independently surpassing humans.

The authors connect that concern to a series of technology disruptions that exposed how dependent Americans are on systems they often do not understand or control. The incidents included a major Microsoft outage, a nationwide airline ground stop, problems affecting major banks and a United Airlines computer failure.

There were also reports that the National Security Agency’s classified internal communications network had gone down, while users reported problems accessing Apple services including iCloud, the App Store and Apple Music. The commentary presents the incidents as examples of how failures in interconnected systems can disrupt communications, transportation and financial services.

Speculation linked the nationwide airline shutdown to AI after the Federal Aviation Administration expanded an artificial intelligence system for air traffic control centers. Federal officials instead blamed a malfunction involving aging air traffic control infrastructure while equipment was being replaced and communications systems were being upgraded.

The Whiteheads argue that the question is not only whether AI caused a failure, but who remains responsible when automated or computerized systems break down. They say officials and corporate executives can adopt new technology without putting the necessary safeguards, infrastructure and human oversight in place.

Autonomous AI hacking tools such as ARTEX AI pose another risk. The tool was originally developed to help cybersecurity professionals identify weaknesses in computer networks, but the commentary says cybercriminals have used it to automate attacks against financial institutions, government agencies and private companies. Some attacks have progressed from initial targeting to privileged access in less than three hours, the authors wrote.

Malicious actors are using autonomous AI systems to identify vulnerabilities and conduct attacks with less human involvement. The Whiteheads say the technology itself did not develop criminal intent; people created the tools, chose the targets and put the systems to work.

Their broader argument is that AI amplifies existing human motives, including greed, arrogance and the pursuit of power. Tools designed to defend networks can be repurposed to exploit them, while systems built for efficiency can be used to collect personal information, manipulate behavior or reduce human safeguards.

The commentary also challenges the idea that computer-generated decisions are automatically objective. Government agencies may defer to computerized decisions because they appear impartial, but algorithms can contain bias and databases can hold errors. Automated systems can then magnify the mistakes, prejudices or political agendas of the people who build and operate them.

The authors point to surveillance systems, predictive algorithms, facial recognition and autonomous weapons as examples of technology whose capability does not by itself establish that its use is justified. They compare the concern with fictional depictions of machines that control people, arguing that a more immediate danger is people voluntarily surrendering independence for convenience.

The central warning is that technological progress does not guarantee human progress. Greater computing power can expand the reach and speed of harmful decisions when people fail to impose limits on those who control advanced systems.

The Whiteheads conclude that meaningful human oversight and restraints are needed as AI becomes more capable and more widely deployed.

Loading…