1982: Scott Fahlman proposes the emoticon

On 19 September 1982, artificial-intelligence researcher Scott E. Fahlman typed a short message onto a Carnegie Mellon University bulletin board that still shapes how humans and machines communicate. Amid a lively thread about the difficulty of telling jokes from serious remarks in plain text, Fahlman suggested two simple character sequences: :-) to flag humour and :-( to mark seriousness.

Fahlman was working at one of the world’s foremost AI laboratories. His own research concerned knowledge representation, semantic networks and the design of Common Lisp—languages and formalisms that underpinned early expert systems and cognitive models. The board he posted to was frequented by colleagues wrestling with the same questions that defined the field: how meaning, context and intention could be captured in computational form.

The twin “emoticons” spread rapidly across the academic networks that became the internet. What began as an ad-hoc solution inside an AI community evolved into a global visual vocabulary. Emoticons later gave rise to emoji, filling digital text with emotional colour that purely logical symbol systems had never possessed.

That September post now threads directly into contemporary AI. Sentiment-analysis models and large language models must parse not only words but the affective cues carried by :-) and its descendants. Multimodal systems are trained on billions of emoji-laden messages; affective computing—the effort to make machines recognise and respond to human emotion—rests in part on the insight that text needs an explicit signal for feeling.

Thus a light-hearted suggestion by an AI researcher on a departmental message board became part of the data and the design problems that today’s intelligent systems confront. It reminds us that the intellectual history of artificial intelligence includes not only theorems and architectures, but also the ordinary human wish to smile across the wire.