Needing Val-ium
There’s a lot of concern from friends and in the media at the moment that the machines are taking over the world and we will soon to be doomed to a life of servility to our artificially intelligent masters.
Perhaps we are already in The Matrix and now the ChatGPT bot can write undergraduate essays, human life on the planet is all but over.
Obviously those of us sitting off the grid in rural Portugal may not feel this as much in our day to day life of strimming land, calibrating our new irrigation system or trying to decide what tiles to choose for the new bathrooms.
We have a more clear and present danger.
It’s far from being an artificial fear and is felt both by us and by Simon & Garfunkel the dogs: that the cute small balls of mewling fur are becoming a gang who now control the valley.
Apologies to the collective noun pedants – I know that our litter of kittens is growing up to become a clowder – but they are behaving increasingly like a street gang.
I haven’t explicitly seen this, but I suspect they deal drugs from under the table grapes, are laundering tinned pet food and are in a territorial alliance with the wild boar (wild boar).
Cats can also be referred to as a clutter, a destruction or a nuisance – all of which are eerily accurate – and to be honest a “chowder” of cats is currently becoming a more attractive prospect.
Of course it’s Val Kilmer who’s the gang boss; I guess I should rename her Uncle Angelo (yes, Val Kilmer played a mob boss...in the 2021 film The Birthday Cake which I’ll be honest I’ve never watched and which received terrible reviews).
It began as simple stalking and basic-level begging (which Simon has mastered over many years), but it’s progressed to attritional meowing and complex attacks combining aggressive door-opening, with kittenburst deployment.
They scatter in a feline form of clutter bombing, distracting us by employing kitten chaff as covering fire as the boss advances straight for a direct assault on the kitchen to grab anything available SAS-style (get in, grab the chicken, get out).
As soon as we get one kitten back outside, another one sneaks in and it becomes a farcical round the table/over the sofa chase scene from an old Laurel & Hardy movie.
The destructive nuisance made off with most of Friday’s grilled chicken leftovers in a rare lapse of Saturday morning defences and retreated to the indoor/outdoor room to pee on our favourite leather chair and lick their paws while staring at us like serial killers.
The sooner we pack off BatCat and Doc Holliday to their new home the better...three cats might be more manageable than five (and cost us less in cat food).
And Simon was specifically targeted this week – courtesy, we suspect, of his new haircut.
Val, the hormone-loaded new mum, has previouly launched pre-emptive strikes on unknown dogs and their owners (I do hope the scars are healing Carlos?), and it seemed she didn’t recognise that Simon with short hair was the same dog as Simon with long hair.
It’s one of the reasons cats will never take over the world...and why machines won’t either...for a long time at least...but I’m coming to that.
In the best traditions of the Pink Panther, Simon was Inspector Clouseau while Val was Cat-o, ambushing him when we least expected it.
We had to intervene. Is there such a thing a cat valium...Val-ium? If there is, she needs some. And so do we.
Of course cats continue to play an important role in The Rise of the Machines.
I honestly believe that if people stopped posting cute cat videos online, the whole internet would break, the Matrix would fall and a great dystopian cat conspiracy would be revealed surrounding their role as a social media entry drug.
While we fawn over kittens trying to back out of a paper bag, our data is given to predatory internet companies to use by Russia to sway elections or to sell to advertisers while all the time reducing our concentration span so we forget how recently the cats were last fed.
Sneaky animals.
It all began with their role in the development of machine learning in the early 2010s – the early building blocks of today’s media hysteria-storm over ChatGPT and how clever computers have become.
Stay with me while we descend this week’s rabbit hole, which has once again played its part in what has been a particularly unproductive week stalked by kittens and failing technology.
Thankfully the builders were cracking on, pouring more concrete for the infinity pool and the third building, and cementing a roof ready for tiling.
Having just about finished the first half of our Portugal wine podcast, but not launched it yet (see below), I’ve been totally paralysed by the fear of how on earth to distribute or even gain some meagre compensation…
At one point I wanted to do little else than eat chocolate and binge-watch Scandi-Noir police dramas.
But thankfully I was saved by obsessive strimming (as opposed to streaming) which provide non-fattening endorphins and a sense of achievement: when you’ve been cutting things down for an hour, you can see that things have been cut down.
I met Stanford Professor Andrew Ng when I was making a fun half hour documentary for the BBC in California called What If? The Next Tech Billionaires in 2012, when Uber was a quaint new thing in San Francisco.
Machine learning was in its infancy, computers were being trained to recognise images by bottle-feeding them pictures associated with labels and Prof Ng’s development of artificial neural networks – loosely based on the human brain – had just made everything exponentially faster.
By making tens of thousands of computers watch YouTube videos 24/7 for a week they had developed clusters of similar looking things and recognised what a “cat” looked like...even if they had never had the pleasure of being stalked or assaulted by one.
It was a big deal – the beginning of machines learning by themselves.
Today we can go to the Photos App on our phones, type in “cat” and the algorithm brings them all up – try it and you’ll see.
But here’s the flaw – I did that and while quite a few cats do come up, so do pictures of an owl mask (don’t ask), quite a few dogs, including Simon the dog next to an owl mask, our neighbour Daniel (!) and a pot plant.
Your phone doesn’t know what a cat is...it’s never met one...it just knows that of all the billions of pictures it’s analysed and learned from, things we call “cats” look a bit like cats...or owls...or pot plants.
Terminator, it ain’t.
In fact, I asked ChatGPT to explain to me why when I search for cats, owls appear and its explanation didn’t explicitly say “we’re not that good at it yet,” but did suggest it was my fault for recently searching for owls (how very dare you, I did no such thing).
I was lucky enough to spend a bit of time learning about machine learning and artificial intelligence during my amazing journalism fellowship at Stanford, and the key take-away is: they might be coming for our jobs, but they won’t be replacing us for a while yet.
If this is a subject that interests or scares you, there are a few good people to look out for.
Andrew Ng posts various interviews with top computing brains like Fei-Fei Li who is another pioneer in the field and who runs Stanford’s Human-Centred Artificial Intelligence unit. They did a great chat recently you can see here.
Prof Ng also spoke to Meta’s Yann LeCun which is a great reality check against fearmongering.
Prof LeCun says we have a misplaced impression that human-level intelligence is closer because of things like Chat GPT: “It’s not going to happen tomorrow,” he says.
“We think that when something is fluent it is intelligent, but that’s not true. Those systems have a very superficial understanding of reality – they don’t have any experience of reality,” said Prof LeCun, talking about something called the Moravsky paradox.
“Things that appear sophisticated to us in terms of intelligence, like playing chess and writing text, turn out to be relatively simple for machines.”
Whereas things we take for granted, he argues, things a ten year old can do like clearing the dinner table and filling the dishwasher is still beyond any robot existing on the planet.
There’s a question I’m sure ChatGPT has been asked once or twice already: “Are you going to take over the world?”
And so I asked and this was the reply:
“No, I am an AI language model...and I don't have the capability or intention to take over the world.
“The responsibility for how AI is used lies with the individuals and organizations that utilize it. It's essential to ensure ethical and responsible use of AI technology for the benefit of society.”
Well, it would say that wouldn’t it?
Oh, and did I mention that we used an augmented reality app when choosing our bathroom tiles? Nowhere is safe...
But would the current AI ChatBots pass a Turing Test ?