Pentagon Plans $30 Million for AI Lie Detector – Researchers Are Skeptical

TVS News AI, editorially supervised, presents Pentagon Plans $30 Million for AI Lie Detector – Researchers Are Skeptical.

You've read a lot about AI recently – about chatbots, image generators, autonomous weapons. But an AI lie detector, funded by the US military? That's exactly the plan the Pentagon is pursuing – and the response from the research community has been muted.

According to heise online, the US Department of Defense plans to spend around $30 million on an AI lie detector in the coming years. At its core, it's about software that automatically detects whether a person is telling the truth.

What is known about the plan so far

According to heise online, what's clear so far is primarily the scale: around $30 million US, spread over the coming years. This is not a small research project, but a substantial investment in a technology that is controversial in the scientific community. More concrete details on deployment scenarios, project partners, or timelines are not provided in the available report.

Why researchers question the concept

Lie detection is not a new topic – and the list of objections is correspondingly long. Even classic polygraphs are considered unreliable in research because they measure physical responses that can also occur with excitement, fear, or exhaustion. An AI system that instead analyzes speech, facial expressions, or behavior faces the same fundamental problem: it must infer internal states from indirect signals – and that's empirically difficult.

heise online puts it bluntly: "Researchers doubt the concept." Which institutes, studies, or individuals are specifically named in the report is not clear from the available excerpt.

What this means for you

$30 million sounds like a lot – but measured against the entire US defense budget, it's a manageable item. The interesting question is what such an investment signals: Is the Pentagon betting that AI-supported statement analysis will find its way into security and investigative work despite all the scientific concerns? Until robust results – or indeed a working prototype – are available, it will likely take some time. Provided the project is carried out as planned.

Source: heise online