July 4, 2022
Lies are as old as language. We once thought the internet would make us smarter, by providing access to a world of information. But it has weaponized disinformation. We’re having an epistemology crisis, many people falling into quicksands of falsehoods, or not knowing what to believe. Threatening the underpinnings of democracy and civilization itself.
But imagine having a device you can rely on to sort truth from lies. Well, now it’s here! Three leading technology firms have joined in an unprecedented alliance to create the Goopplesoft AI Truthfulness Oracular Revealer (GATOR). Just point it at any information source — a TV, computer, smartphone screen, newspaper, etc. — or your Uncle Harry — for an instant reality check.
This is not virtual reality, or artificial reality, but real reality.Where two plus two equals four. If someone says five, the lie will be exposed. So will every other kind of untruth.
The GATOR comes in three attractive colors: baby-bottom pink, glaucous blue, and boiled-frog green. With an equally attractive price of just $99.95!
How does it work? Not like conventional “lie detectors,” which are so problematic as to be mis-named. GATOR instead uses Artificial Intelligence (AI). Such systems have been advancing for a long time. A basic approach has been “deep learning,” like where you feed a program a zillion cat pictures, from which it trains itself to recognize another picture as showing a cat. Of course such a system is limited; it can’t identify a dog.
But now researchers have been developing “foundation models,” with broader capabilities. One method is to feed in millions of words of text. The program hides a word from itself; tries to guess the word; then examines the result. Repeat millions of times and the system not only develops deep intuitions about language, but even what amounts to some understanding of the world.
The Economist magazine recently focused on this. The cover art (see picture) was created not by a human but by an AI system — instructed with nothing but the headline!
Such AI principles power the GATOR. Having access to a vast corpus of information about the real world, which has been uploaded to the Cloud. When asked to evaluate any assertion, the device will search through that gigantic database, finding every relevant fact, integrating them all with its own basic knowledge, and producing an answer that has a correctness probability of virtually 100%.
So just point your GATOR at some TV broadcast — Fox News, for example. Or some politician — Donald Trump, for example. Or at any piece of text, any social media post, any website, or at any person bloviating in real time. You’ll know instantly whether what you’re seeing or hearing is true or false.
Just think how useful this will be with regard to advertisements. You can even bring it to church; point it at the preacher.
Admittedly many questions don’t have factual true-or-false answers. GATOR can’t tell you whether abortion is moral. But a lot of important questions — despite being widely contested — do have actual answers. The 2020 presidential election was not stolen. Vaccines don’t cause autism. For those we shouldn’t need AI. But apparently many unassisted human brains do need such help.
Indeed, GATOR will change the world. Resolving that age-old conundrum, “Who are you gonna believe — me, or your lying eyes?”