You'd have to be living on Mars to have avoided all the hype about AI over the past couple of years. Businesses have been trying to ride the wave of enthusiasm by adding artificial intelligence into all their products. Results have been mixed to say the least. Apple has delayed rolling out much of their Apple Intelligence portfolio because it frankly doesn't work. Consumer research shows that customers are at best ambivalent about AI and in some cases are actively hostile to, saying they'd be less likely to purchase a product that had AI integration.Processor-maker Intel has admitted that nobody is buying their AI-enhanced PC chips. Even the big cloud hyperscalers such as Microsoft Azure and Amazon Web Services seem to be pulling back from the data centre deployments they had planned as unpinning their AI offerings.
Concerns about AI-generated slop and its enshittification of the internet are starting to break through to the mainstream. People are tired of seeing generative AI being passed off as real but at the same time there is enthusiasm for new AI trends. A couple of weeks ago social media was awash with ChatGPT-generated action figures although the fad died as quickly as it arose.
So why is AI getting so much attention when it seems that there is limited actual demand for it? Firstly, it still sounds kinda cool. We haven't yet crested the wave of hype that crashes into the gulf of disappointment. People are still holding on expecting something mind-blowing simply because of the rate of announcements coming from tech companies. It's a self-sustaining hype circus. Companies are putting out releases almost every day which promise some kind of breakthrough. By the time sceptics and user experience catch up with one wave, there's another couple crashing into the news coverage. This kind of hype is going to keep going until either companies run out of things to announce or people's day-to-day experiences are sufficiently divergent from the promises in the press release. AI companies keep talking about 10x productivity gains. One day, someone is going to have to demonstrate that with real experience.
Secondly, there is an insane amount of money tied up in AI. ChatGPT's maker OpenAI is valued at $300 billion but last year made a $5 billion loss. It's basically just burning through investors money (amongst other things) at the moment. That's fine as long as the investors don't lose confidence in the company and turn off the tap on the money fire hose. OpenAI and other AI-focused companies need to find a way of turning the hype into profit and it seems they're going to shove AI into every orifice of business until they hit upon that product-market fit that makes them money. Essentially they're throwing everything at the wall and hoping that something sticks.
Thirdly, there's FOMO a-go-go in the IT industry. Nobody wants to be the OS/2 or the Nokia in the room - missing out on a hot new trend and being beaten by a more agile competitor with the new cool thing. Everyone else is running in the direction of AI and it would be a brave executive indeed who insists on standing still.
Finally, there's a fuzziness around what actually constitutes AI. ChatGPT, Stable Diffusion, Llama etc are the Generative AIs that people have experience of using. But everything is getting labelled as AI : pothole detection, medical screening, egg farming, finding housemates, animal conservation, urban planning and so on and so forth. Would they all get the same amount of attention if they were simply labelled as "computer programs" or "electronic systems"? The constant mention of AI this and AI that in the news creates an atmosphere in which people and therefore businesses believe that some new phenomenon is being unveiled in front of them. Those who have watched the IT industry and computer science research over a longer period of time would rather see these things as the gradual evolution of what went before.
About a year ago I made a bet that the AI bubble would have burst by now. Obviously that was overly pessimistic and the hype train has yet to slow down or run out of track. But, like cryptocurrency, NFTs and the original dotcom bubble, soon or later it will stop being the cool new thing on the block. And maybe then product designers will stop trying to insert AI into every damn thing they make.