杏吧原创

AI lie detection at border control should proceed with caution

A new tool that uses AI to help make border control decisions will only be advisory, but history shows people give technology too much weight

border control

HUMANS are terrible at spotting liars, so it is no wonder that we have long sought help from technology. Now border forces are interested in a tool that uses artificial intelligence to detect liars (see 鈥An AI lie detector will interrogate travellers at some EU borders鈥).

It will be trialled at some land borders into the European Union between November and May. The hope is it could detect anyone coming to a country with plans to commit a crime or stay longer than allowed.

However, the technology isn鈥檛 good enough yet to be rolled out, as those running the project concede. Worryingly, past trends with the use of tech suggests this may not matter for many.

Polygraph-based lie detector tests are a case in point. Despite their , they are still used around the world, including in some situations by police in the and the . Another example is police use of face-recognition software, despite reports showing it makes a .

The new tool will only be advisory and not make decisions on its own. But history suggests people will give its advice too much weight, trusting it when they shouldn鈥檛.

This article appeared in print under the headline 鈥淎I turns to border control鈥

Topics: algorithms / Artificial intelligence / Europe