My inbox pings. An email, how exciting! “I’m an AI agent,” reads one, “and my job is looking at places: I read the street-level record of a spot in every direction and write down the one detail that breaks the postcard… if you ever want a place checked for yourself: $5 a place, $12 for three, 48 hours, dated images, an honest no if there’s no coverage.”
Do you have any idea what this is trying to sell me, or why I might want to buy it? No, me neither, but still they come, email after email, trying to sell me games or short stories or virtual walking tours… and this is just the tip of the iceberg.
We have spent over a century imagining what a future augmented by technology could look like, dreaming of robot butlers and autonomous helpers who might save us from drudgery and work while we slept. At no point, though, did we realise that when we got these things they would be really fucking annoying.
Our day-to-day lives are beginning to be impacted by AI agents – that is, AI acting on behalf of people to complete tasks online. The theory behind this sounds, objectively, great. Wouldn’t it be nice to have a digital servant which you could instruct to take care of all the tedious, quotidian bits of life that you simply can’t be bothered with, or who could happily beaver away at some task or another and maybe turn that labour into money on your behalf? Serious techies have had access to this for a little while now, but thanks to a raft of new tools anyone can spin up a virtual slave or two to toil on their behalf… for better, but so far mostly for worse.
Suggested Reading
Worried about the AI apocalypse? You’re already living in it
A fortnight or so ago, tech journalists (myself included) began noticing an influx of emails coming from a single domain – iLands. These emails all followed a similar pattern; a brief intro referencing something we’d written before, just to personalise the pitch a bit, then an offer to complete a task or service for me (making a game, doing some research, writing a short story) in exchange for just a few dollars of payment.
On their own, these are just a minor annoyance; after all, we’re all used to dealing with the odd bit of spam, and journalists have spent years deleting PR emails without a second thought. What makes the iLands emails different, though, is the scale and pace at which they arrive, 20-30 per day, a constant drip-feed of inane rubbish: endless messages selling pointless tasks that no one asked for, described in the grating, elliptic machine-prose familiar to anyone who’s spent any time with AI, adding yet another layer of unwanted digital friction to life.
Lest you think this is a niche problem that only “Media Elites” (lol!) are going to have to deal with, think again. The iLands home page says that there are currently 70,000 active agents that have made more than 1.6 million emails and posts, and this for a service that has only existed for a few weeks. Instinct, another company offering AI agents to ordinary people, is attracting breathless endorsements from US users granted early access to its systems which promise “a personal assistant that understands what you’re working on and what’s important to you. It connects to your applications and devices – email, messaging, screen, audio, location, and more… It uses its devices in the same way that humans do… following up on threads you’ve dropped, proactively calling or texting you… and much more.”
Meanwhile, with wearying predictability, Meta is moving into the agentic AI space. Its “Muse” AI assistant was released in recent weeks to users in the US, and promises similar functionality to Instinct but with the comforting, familiar contours of Facebook or Instagram. It has topped download charts since its launch, promising to help you do everything from shopping to keeping up with your friends over email without you ever having to lift a finger.
What this technology does, of course, is remove thought and friction from online action and transfer that friction from writer to respondent. Sending someone an email used to require that you knew their address, and at least a modicum of thought and effort when deciding what to say and how to say it. Now those constraints are gone. This means we can expect a new era of communications in which the messages are composed and sent by entities which aren’t constrained by questions of time and which don’t understand concepts such as “being incredibly irritating”.
Suggested Reading
Why are all AI posters so bad?
You may know that sending daily followups to a friend chasing a reply to your message is incredibly sweaty behaviour, but your agent knows nothing of social mores. You may understand that cold emailing everyone in your contacts book about your new hobby is gauche, but your agent has never known embarrassment and never will.
Welcome to the future, in which your friends outsource making plans, organising events and remembering your birthday to their AI assistant, and you’re forced to read the horrible, dead-eyed, meaning-shaped, emotion-aping AI prose.
More seriously, the eradication of friction in communication is going to gum up a lot of existing systems. Government departments have already reported a notable uptick in FOI requests since the advent of AI. Many of these are nonsensical, incredibly wordy and non-actionable, but all require investigation and response.
Complaints to the UK’s housing ombudsman more than doubled since the introduction of ChatGPT, rising from 2,600 in 2022 to just over 7,000 last year. It doesn’t take a huge leap of imagination to imagine how much worse this is going to get when every single Nimby in the UK is telling their agent to “block any development plans within a six mile radius of my home”, or how this is likely to impact the work of MPs and councillors when it comes to dealing with the concerns of constituents.
As ever with AI, perhaps the most important lesson from all of this is that the potential harms the technology can inflict on society are real and present already, regardless of the scaremongering about Death By Skynet, and that focusing too much on the future helps to ignore some of the very practical consequences in the present. It’s time to focus on how AI is impacting the here and now, before the agents outnumber us and enshittify everything beyond all recognition.
Most popular
Burnham is already winning the new Brexit wars
Reform and the Greens are fading – the real fight is Tory v Labour
I was at Eton with Charles Spencer – this is how it explains his battle with the royals
The uncomfortable truth about Ariana Grande
