Do you have smart friends? I’d like to think that I do. They work in a range of professions, some traditional (teacher, doctor), some creative (music agent, graphic designer) and some that I will never truly understand, so I just never ask about (”sustainability consultant”, whatever that is).
They went to good universities and got good degrees. They’re funny, which I’ve long believed requires at least a little intelligence (with the exception of some rare outliers, like Joey Essex or Kermit the Frog).
But my lovely, otherwise erudite pals have been uncannily keen to rock a dunce cap in recent months. Nearly all of them are ChatGPT users, and many of them speak openly about their ChatGPT usage, completely unaware of my widened eyes and furrowed brow. Seriously — it’s pushing forward my Botox timeline in a way that I don’t appreciate.
And they aren’t reserving their AI chatbot usage for work anymore. They’re using it for genuinely ridiculous, mind-numbing everyday tasks and enquiries, half of which could be solved with a simple Google, the other half of which could be tackled by just... thinking about it.
The speed at which my mates turn to ChatGPT has become frankly alarming. The neurotransmitters once used for critical thinking are becoming increasingly depleted with every visit to the site. The tell-tale white screen is drawn up almost instantly, whether it be for the calculation of a restaurant bill (which ChatGPT often gets wrong, by the way) to writing their sexts (yes, really). I’m not sure what’s going to snap first: their synapses or me.
Last week, I received my first paragraph text message — the emotional kind — that was clearly written ChatGPT. It was sent by a friend, but ChatGPT’s fingerprints were all over the place: bullet points to summarise key arguments, em dashes (not used in the kind of fun, personality-laden way that I just used them, obviously), wide spacing between paragraphs, you name it.
I’m already aware that some particularly desperate people have been turning to ChatGPT for emotional purposes (bump them up the NHS waiting list for talking therapy, I beg), I just didn’t think my own friends would be outsourcing their emotions to a chatbot any time soon.
It’s bad enough when they ask it to assist them with simple, everyday tasks that they used to handle perfectly well with their own grey matter, but using an AI chatbot for intimate correspondence has pushed me to breaking point. At least getting ChatGPT to write your sexts is objectively funny. This is just concerning.
And that’s to say nothing of ChatGPT’s financials: OpenAI’s cofounder Greg Brockman and his wife reportedly donated a significant amount ($25 million) to US President Donald Trump’s super PAC this January — the largest donation in the six-month fundraising cycle. (Oh, and let’s not forget that Trump’s own cyber chief recently uploaded sensitive documents to a public version of ChatGPT. An internal review was launched, but it’s currently unclear what was concluded.)
Brockman’s donation has led people to boycott the chatbot, with a online “QuitGPT” campaign gaining traction, especially in the US, where there is ongoing violence from the government’s Immigration and Customs Enforcement (ICE) agents. It hasn’t had quite the same impact across the pond, which means my friends are still gleefully punching requests into the search bar, a dumb smile plastered across their faces.
It’s also bad for the environment: the training process for OpenAI’s GPT-3 was estimated to have consumed 1,287,000 KWh of electricity before it was even publicly released, enough electricity to make more than 2,500 tonnes of steel. It’s also thirsty as hell. Another study found that Microsoft used around 700,000 litres of fresh water during GPT-3’s training in its data centres — the same amount of water needed to produce 370 BMW cars or 320 Tesla vehicles. It has been estimated that a conversation of 20-50 questions consumes the equivalent of a 500ml bottle of water. Given that ChatGPT receives 29,000 prompts a second, that’s no small amount of water usage.
Not that Sam Altman, OpenAI’s CEO, is too bothered. In a recent interview, he said that a “significant fraction” of the world’s power “should” be used to power AI. And OpenAI is only one part of the large language model problem — there are many other models guzzling water and draining energy, too.
I can’t moderate my friends using chatbots for emotional or critical-thinking purposes. But I can control how I respond to it. Or, more accurately, whether I respond to it. So I’ve formulated one simple solution. If you ever text me using ChatGPT, don’t expect a response. Ever again. Instead, I’ll see you in person, where I can ver the provenance of your words. As for correspondence? It’s carrier pigeon or bust from that moment forward.
