You know how sometimes you say a word too many times and it starts to sound like it's not a real word and loses all meaning? That's where we are with "AI".
"Artificial intelligence" has always been something of a nebulous term in common usage. It summons up visions of science fiction in which a robot or computer behaves in a human-like fashion, responding to verbal instructions and conversing as if sentient.
Since the release of ChatGPT in 2022 the term "AI" has come to be thought of as just that, an apparently intelligent entity with which you can chat and which will respond intelligently. As I've previously discussed, this is largely an illusion caused by the fact that we tend to assign humanity to things that look or act a bit like humans (a form of pareidolia).
But AI has nevertheless taken the world by storm and a lot of people are investing a lot of money in telling us that it's going to change the world. Even if they're not really clear on what "it" is.
These days, if you want to get investment or press coverage, you have to be shoving "AI" into your product. It doesn't matter whether it really exhibits any intelligence or even whether your customers actually want it, as long as you can call it "AI" you're halfway to the home straight.
This week the UK's National Health Service announced that it would rolling out an AI triage tool within its app. Details on what this actually meant were conspicuously absent. Was it a cutting-edge large language model? Was it a simple yes/no questionnaire? No idea. But it was called an AI so it got plenty of press coverage.
The irony is that the NHS already has a digital triage system. It's part of NHS 111, the out-of-hours care service for non-emergency health issues. Available on the web and over the phone, it is a series of questions in a branching tree structure where each answer leads either to another refinement question or to an assessment of medical urgency (eg wait til Monday and see your GP; go to the pharmacy; come in and see an out of hours doctor).
Thirty years ago we were calling these things "expert systems". They were built, not by hoovering up every single word ever published, but by having actual subject matter experts working alongside engineers to codify the expertise. Unlike large language models, these expert systems are deterministic (the same answers will always follow the same path and get the same result) and can easily be verified by a human expert.
At my day job we've started using a system that will apparently give our board a strategic overview of the organisation. That's what an uninitiated person might call a "dashboard" or a business intelligence SaaS. But not the smart marketing bods at this company. No, what we've bought is a "an AI-powered ally in technology decision-making" complete with illustrations of a Iron Man-esque humanoid robot standing shoulder-to-shoulder with a stock photo businessman.
If it really is so artificially intelligent why are we the ones doing all the work inputting our data into it? A truly intelligent system would just be plugged in to our existing systems, fed a bunch of internal documents and come up with the answers itself. To call this an "artificial intelligence" is to render the term meaningless.
Nowadays it seems that any computerised process can be labelled as "AI".
That little chat window that's been sitting unloved for years in the corner of your website providing next to no help to visitors? Now it's an AI customer service bot!
Those APIs you've been consuming since "Web 2.o" was a thing? Now they're agentic AI APIs!
The tentative calendar entries that your email client automatically adds whenever you get a mail with a future date mentioned in it? AI personal assistant!
It's such a frustrating additional bit of cognitive load every time I read the word "AI" to have to try figuring out "Is this an LLM? Is this a standard bit of computing functionality? Is this actually an optimised way of doing something that has to be branded as AI just to get anyone to notice?"
We've reached (and probably passed) AI saturation. When everything is "AI", nothing is. It's a meaningless term and it's turning people off. "AI" is becoming synonymous with "slop". Stop calling your product "AI" if you want people to take it seriously!