It's interesting, isn't it, living in a time when anything can be true as long as you shut your eyes, plug your ears and just keep saying it? When you can just come up with what we call a 'pub reckon' and, instead of remembering that it's a half-baked, half-remembered idea you came up with after an Amstel too many, you can just decide to wholeheartedly believe in it?
In hindsight, it might've been a mistake to train computers to replicate that behaviour.
For example: hey, did you know that Lewis Hamilton is the godfather of Max Verstappen and Kelly Piquet's new baby?
He isn't, but Google saw a couple of AI-generated Facebook posts that said he is and provided comically unbelievable pictures, so now if you search 'Max Verstappen godfather' on the search engine so ubiquitous that 'to Google' is the de facto verb for looking something up online, it will confidently tell you: "Lewis Hamilton is the godfather to Max Verstappen's daughter."
It'll continue, confidently, to tell you about "a role he was chosen for after visiting Max and his partner, Kelly Piquet, following the birth of their daughter in 2025. The gesture highlights the personal relationship and mutual respect between the two Formula 1 drivers, which contrasts with their intense rivalry on the track."
Kelly Piquet and Max Verstappen welcomed their first child together earlier this year
AI hallucination sees Hamilton named godfather of Verstappen's child
It's not too much of an exaggeration to say that Google's shift from 'website you use to search terms on the internet' to 'AI-led lie machine, which will also spit out results for sponsored sites ahead of ones which are more helpful and correct but don't pay Google for the privilege' has been one of the most harmful tech developments of the 21st century. And that's a crowded field. Like, a 25-team grid going into Turn 1 at Mugello.
Man, can I just break the fourth wall for a moment? I wanted to write this piece as a light, fun spearing of Google's AI search results being more bugged than Richard Nixon’s White House, but instead I just got incredibly depressed about two words in.
Did you know that people are only half as likely to click a even single link on Google if they get served an AI summary with their search? It's just blind faith and slavish obedience to a deeply flawed technology, which makes everything around it worse.
Here's your task for this week. When you're about to pass on something that you've read uncredited on social media, or asked Grok to summarise, or done anything other than have the basic human curiosity to know whether what you're about to say hasn't just been made up for clout or hallucinated by AI – don't. Pause for 30 seconds. Go and find an actual primary source for it, whether that's a video, a news article from a website with some level of accountability and respect, whatever it is.
I did it just now, actually going to Google to check that the 'Max Verstappen godfather' search summary is a real thing it delivers and not just a doctored screenshot someone posted online. It's not hard, and it'll stop you looking like a bozo more often than you'd think.
Every time you see something on Twitter and just ask '@grok is this true', you are actively choosing to make yourself a little bit more stupid. You're abdicating your own thinking and handing it off to a computer program which is famous for getting things wrong. You are choosing to do this instead of making a couple of clicks and finding out for yourself, more reliably.
Just take half a second to think critically about something, and make the world a minutely better place.
Because unless more people remember how to do that, we are screwed.