Recently, I have observed people musing, applauding, warning about an ‘AI bubble’ on Bluesky, especially after Sam Altman, CEO of Open AI, talked about it and compared it to the dot-com bubble of the 1990s. Shrewd observers then asked whether we were talking about a bubble in AI technology or in AI markets and whether it’s more comparable to the tulip bulb mania of the 1630s or the railway mania of the 1840s. Others pointed out that the dot-com bubble didn’t make the internet obsolete…..

In this post I will not contribute to these clever discussions but just do what I sometimes do, that is, engage in some conceptual archaeology, focusing on the two metaphorical phrases ‘AI winter’ and ‘AI bubble’, which themselves are surrounded by a penumbra of metaphors such as bursting, crashing, stalling, popping, puncturing, wobbling, deflating, freezing and so on and also by related phrases like AI hype, AI boom and even AI con.

Almost all the phrases I mentioned (AI hype, boom, bubble, winter) are older than you might think, with AI winter being the oldest, as there was already talk of such a winter in the 1970s, and the others making their way into our language during the 1980s. They all have risen in prominence after the 2022 AI explosion, as this Google Trends graph shows.

In the following I explore the meaning and history of the two metaphors ‘AI winter’ and ‘AI bubble’, which are of course both linked to AI hype, as hype leads to winter and to bubble bursting – the typical hype disillusionment cycle (you can learn more about hype here and about the long history of AI hype here).