On the Couch: New wave for tinfoil

We know what you’re thinking. A study has used a combination of magnetic resonance imaging and artificial intelligence to turn thoughts into complete texts.

We know what you’re thinking. A study has used a combination of magnetic resonance imaging and artificial intelligence to turn thoughts into complete texts.

Published May 6, 2023

Share

Tinfoil ‒ it’s the only solution.

Buy shares and take the Covid lockdown toilet paper route and stock up because it’s got a huge future.

Take long strips of foil and wrap them around your head. This used to be considered the action of someone who may not have a firm grip on reality, but the couch science council stumbled across some startling news this week which points to a renewed necessity for this safeguard for general consumption.

The news possibly came from a passing car. Or one of the neighbours.

University of Texas, Austin, researchers Jerry Tang, a doctoral computer science student, and Alex Huth, an assistant professor of neuroscience and computer science, published in “Nature Neuroscience” the results of a study that will help people who are conscious but unable to verbalise, or those who have suffered a stroke, to communicate.

That’s a powerful tool to help those who need it.

But what terrified the wits out of me is that part of the study uses computer stuff similar to the systems that drive OpenAI’s ChatGPT and Google’s Bard.

It’s complicated if you’re not a brain box, but basically it’s a brain decoding system that translates images into text. In layman’s terms, a functional magnetic resonance imaging scanner (fMRI) read’s brain activity and collaborates with hours of recorded vocals. It can then translate images or snatches of thought in the brain and spit out a text version. Apparently it’s not yet totally accurate and – I found this alarming – can be open to “interpretation”.

AI has exploded. No one really knows yet where it is going, how far it will go and the extent to which it will affect our daily lives.

My imagination ran wild when I read the results of the Texan study, for which subjects needed no implants. At the moment, people need to have their heads read by fMRI for this to work. But what if …?

Imagine, in a world of the AI of Things, you’re sitting in a meeting during which you have to find some delicate, diplomatic way of telling someone they’re an asshat and their plan spells doom all round.

My boss is one of the best, so there is no shadowy reference here. But what if your “superior” is really not and you mentally explore medieval means of torture while you smile and nod (to ignore later) the drivel being delivered to you.

Or, conversely, you stumble and say something truly embarrassing and “hear” all the kinds of idiot your audience is calling you. The self-flagellation of humiliation is bad enough without knowing the colourful glee of those celebrating your disgrace.

Even worse, what if AI learns to tap into whatever wi-fi is called in the near future, sort of 7G or something. Will that mean the person who wi-fi-phoned you and you really really don’t want to talk to will be able to “hear” you rolling your eyes and “saying” rude things?

Will your neighbour’s wi-fi be able to transfer the terrible names you call him or her when they start up with the leaf blower/power tool at 7 on a Sunday morning?

The possibilities, just like AI, are endless and privacy could become a quaint prehistoric notion.

Pardon, what did you say? You’re breaking up in my tinfoil wrapping.

  • Lindsay Slogrove is the news editor.

The Independent on Saturday