Are you careful or cretinous? AI can be the former, freeing humans up to be the latter
Join my impossible mission to stop leaping off tall things and onto fast things
It’s time for ACTION! You know… gadgets and explosions and stuff. Hell yeah.
Exactly what form this action should take is not important right now. What matters is to maximise my human propensity for impulsiveness and improvisation, and JUST DO IT (whatever ‘IT’ is)! Like they always say in the films, let’s kick bottoms! Or, failing that, a donkey! Don’t ask why, JUST DO IT.
The trigger for my impulsiveness is the card payment machine at another auto-checkout. Here it is:
COMMUNICATION IMPOSSIBLE? It’s a new action-movie franchise! But instead of Tom ‘I lost Dabbsy a lucrative contract as a result of my cardboard acting in a horror film’ Cruise [*], Communication Impossible will star your favourite Friday afternoon tech chroniclist. Payback time!
One tip I learnt from the Mission Impossible films is that I must immediately leap off a tall thing and land on a fast thing. Unlikely though it sounds, this will somehow bring about a solution to my problem. Unfortunately, I am on the ground floor of a two-storey shop.
So I dash outside, forgetting about my shopping (an electronic voice assures me that it is still in the bagging area) and race across town to find a building tall enough for my purposes. I tear through reception and into the lift and wait a few minutes listening to this year’s winning Eurovision Song Contest entry until it reaches the top floor, where I tear out again and shoulder open the emergency exit onto the roof.
From here I sprint to the edge and hurl myself into the void …
… landing neatly on the seat of a motorcycle that someone implausibly parked on the street below having left its engine still running. I rodeo it across a packed town square to make the pigeons go wild, then ride though a candlelit church for no good reason and finally up a conveniently located but curiously incongruous ramp in a dockyard onto the roof of a passing train.
OK, in hindsight it is not clear how this might fix the error on the card payment machine. It just seemed the right thing to do on the spur of the moment.
I have a habit of doing things throughtlessly that I soon regret, before going on to repeat the same mistake again. I wrote about this a couple of years ago, mentioning a guided beer-tasting session that involved me forgetting that drinking my sample beers AND those of Mme D might make me feel a bit squiffy.
It’s not that my memory is at fault. In fact, I remember quite clearly on that beer-tasting event that a couple of the beers smelt like shit or sick, yet I went ahead and drank them anyway.
On that theme, a local bar to me here in France called ‘My Beers’ regularly hosts evenings in which you can play the infantile American frat pub game ‘beer pong’. You can only admire their honesty in the social media publicity, as you can see below:
If you say so.
While we’re on the topic of drinking beer, here’s something I’d not seen before: a ‘uritrottoir’. No, not a French performing street astronaut, but a miniature public lavatory.
In France of old times, towns would notoriously erect pissoirs – a kind of outdoor public toilet for men, situated on main streets and used by those caught short between the bar and their own front gate. Some pissoirs remain standing as historical curiosities but nobody in their right mind would actually use them, at least not during the hours of daylight.
An unfortunate side-effect of dismantling the stinky pissoirs in the modern era is that drunks stumbling out of bars just nip into a side-street and piss in the gutter. So a uritrottoir is a one-person pissoir situated in those same side-streets where a drunk might slip into for a waz. Here’s a photo I took of one in Toulouse:
Apparently this is healthy for the bladder, cleaner for the environment and keeps locals happy by stopping drunks excreting in the gutter. I understand the council arranges to collect the urine so it can be used for compost or fertiliser or something. I’m not quite sure of the details. It was all explained on the uritrottoir itself but I couldn’t properly read the notice as my eyes were streaming with tears due to the overwhelming stench of acrid piss that lingered within a radius of half a kilometre of the unit.
How desperate would you have to be to use one of these ultra-public toilets? Not desperate at all, probably. Rather, you’d do it on impulse. Imagine you’re walking back to the bus stop, wishing you’d nipped to the gents one last time before heading out of the pub but thinking it’ll be OK as you can relieve yourself in comfort and safety once you get home. Then… you spot a uritrottoir. Impulse takes over.
Now, despite observations that suggest the contrary, artificial intelligence (oh here we go again) is not actually very good at replicating this kind of impulsive action.
Chatbots, for example, spout out all kinds of nonsense but that’s not random behaviour: that’s what they’re designed to do. I have referred to Chat GPT in the past as a “glorified autocorrect algorithm” and I stand by that. AI evaluates what you expect to read in a response and serves it up with care and forethought, even though the response itself may be incorrect. It analyses your query and responds with a carefully constructed heap of bollocks that looks, on the surface, to be not bollocks at all – even though bollocks is definitely what it is.
One of my favourite demonstrations of this was when RetroNews, a subscription-supported media project that digitises old newspapers, asked Chat GPT last year to pick its most striking front page in any newspaper published before 1950. Chat GPT duly chose one – describing it in precise and emotive detail – that simply did not exist. Repeatedly challenged on this point, the AI doubled down and kept insisting that it did.
When asked to choose some key articles, photos or caricatures that best represented their eras from all the digitised newspapers in France’s National Library, Chat GPT produced a list of detailed but entirely bogus items, each helpfully documented with weblinks that led directly to 404s.
My point is that AIs do not busk answers or evade questions it cannot answer. It gives clear, well thought-out responses that just happen to be wrong, then calmly and respectfully tries to back up the falsehoods with fake proof.
But not even that is my real point. AIs will of course get better at finding the right information from its datasets and will eventually provide more accurate and fair responses. It will increasingly take over the dull chores of research and the evaluation of factual data to provide better answers.
But where does that leave human decision-making? Not in a good place, according to the late (and already much-missed) economist Daniel Cohen. Paraphrasing the theories of neuroscientist Michel Desmurget, Cohen agreed that the human mind can be said to operate on two levels: one is impulsive, approximate and simplifying, which helps us to move quickly and make decisions on the spur of the moment; the other is reflective, analytical and careful, which is sensible but slow and boring.
Yeah, so? Well, AI is really good at the slow and boring tasks that require reflection, analysis and care. Those are the kind of tasks that AI is already revolutionising and may take over from humans altogether. That leaves us humans with nothing else to do but practise the other – i.e. thoughtless random behavioural impulsiveness – all the time, just like the kind of behaviour we see on social media, for example.
It’s like how people stopped doing simple arithmetic in their heads or on a scrap of paper once they knew a calculator could do it for us. Apply this to every other challenge to the human mind as well and we will, as Cohen and Desmurget predict, left to become “algorithmically managed cretins”.
Yeah whatever. All I can say is that it took me ages to get back to town after my rooftop-motorbike-train escapade.
How was I to know that the train I landed the motorbike on was nonstop to nowhere? I rode most of the way back on the motorbike, until the police pulled me over as apparently someone had reported it stolen. I had to jog the rest of the way.
The good news is that the shop is still open and my shopping is still there. In fact, the electronic voice is still bleating that I have left it in the baggage area and that a customer service assistant is coming over to help.
I wait a few more minutes and eventually an assistant does come over to touch the machine with her magic keyfob. She informs me the card machine has successfully reconnected – in fact, it had done so a few seconds after I had dashed out of the store. It is now happily awaiting my PIN to complete the purchase.
Tap tap tap tap. Oh dear, it is not accepting my PIN. Perhaps I typed in the wrong one?
Fuck it, who cares?
With a yelp, I crash through a closed window next to the open front door, somersault over a low bollard, dance lightly over the roof of a parked car and, landing elaborately with a roll and springing upright onto the pavement, I commandeer a passing Segway.
“Follow that car!” I yell, pointing the wrong way up a one-way street, “And step on it.”
Here we go again!
Alistair Dabbs is a freelance technology tart, juggling IT journalism, editorial training and digital publishing. [*] He was commissioned to research a major publishing project revolving around Universal Films’ audacious plan to remake all of its classic-era horror films, one by one, featuring today’s big-name Hollywood stars. Unfortunately, they launched the series with The Mummy starring Tom Cruise. It was such an embarrassing turkey that Universal abandoned the entire plan, along with all associated side-projects. Oh well.
We had a rep who planned his visits to customers around his detailed map of public toilets. In later years, this map had a lot of crossings-out resulting in a lot of crossing-legs, I suppose.
A 'Uritrottoir' would never catch on in blighty; It would get vandalised leaving the alleyway free for the usual activities that walls provide to the ex-pubgoer. Perhaps Chat GPT might be able to provide a logical argument and explanation for this outcome.
As far as I recall, you never see a 'Uritrottoir' in the action films where the hunter/hunted chase/race/drive down ginnels.
We have no need of AI to turn people into cretins, mobile phones have got that task well in hand