Skip to content

Science1 publisher2 min readPublished

Merriam-Webster adds AI words its own lexicographer says take programming knowledge to explain

Merriam-Webster added at least six AI and computing terms, including 'agentic' and 'vibe coding', to its online dictionary in September. The entries show that specialist AI terms now circulate in mainstream print, but the dictionary's method cannot tell how many general readers understand them.

The Scientist · Science desk

Illustration accompanying Merriam-Webster adds AI words its own lexicographer says take programming knowledge to explain

What happened

  • Among the other AI and computing entries Nature lists are AGI, prompt engineering, natural language processing and 'compute' used as a noun.
  • The same list includes four non-AI technical terms: biohack, direct air capture, geoengineering and uncanny valley.
  • Merriam-Webster lexicographer Peter Sokolowski says a word qualifies only after it appears repeatedly in mainstream outlets such as newspapers and magazines.
  • The dictionary has been a staple of US libraries and classrooms for decades and traces its roots to a compendium of words published in 1806.

Compiled by The ScientistSomething wrong?How this is made

Why it matters

  • constraint An entry test that counts appearances in print measures how widely AI terms reach general readers. Whether those readers can follow them is a separate question.
  • decision Writers for non-specialist audiences still have to explain terms like 'agentic' in context. Merriam-Webster's own lexicographer says such entries depend on concepts like programming that many readers may lack.
  • precedent If Boroditsky is right that technology now changes as fast as it did in the industrial revolution, or faster, later batches should bring more specialist coinages into the dictionary by the same print route.

Passing the newspaper-and-magazine test is only the first step. Specialist groups of physical- and biological-science "definers" then pick candidate words, study how each is used technically and write a definition aimed at a general audience [6]. So the process counts how often a word reaches general readers, then writes a meaning for people outside the field.

Nature's report says the new additions stand out for the advanced knowledge needed to understand them [12]. Sokolowski, who writes and edits entries, gave the plainest case. "You can't really explain 'vibe coding' if you don't know what 'programming' is," he said [3]. He puts the sophistication of the batch down to twenty-first-century digital communication, which in his account has given non-specialists wider access to highly technical terms [4].

His explanation is a plausible interpretation. Nature's table lists ten terms, six of them AI or computing [1]. The report calls the additions "a slew" [1]. It does not give the batch's total size, say which channels carried the words, or compare this batch with earlier ones in figures. The AI share of the whole batch cannot be worked out from the table. The comparison with earlier batches rests on Sokolowski's impression. "This particular batch, more than any other that I've seen, really does feel like it's a meta-commentary on language," he said [9].

Lera Boroditsky, a cognitive scientist at the University of California, San Diego, who studies language and cognition, said the surge of complex technical words is not surprising [10]. Technological and scientific innovation often drives language change, she said. In her view, technology is now advancing at a rate comparable to, if not greater than, the rate during the industrial revolution [10]. "The more complex a society becomes, the more of these tightly packed units of meaning we have to make, so we can talk about things that are complex in ways that don't take forever," she said [11].

I think the evidence supports a narrower claim than these words having become general vocabulary in the full sense. Terms like these now recur in mainstream print often enough to earn entries [5], so non-specialists are meeting them. Boroditsky's account explains why such compact terms get coined [11]. Finding out whether people outside the field can unpack them would mean asking readers what the words mean.

What to watch

  • Merriam-Webster publishing the full September list, which would give the batch's total size and the true share of AI and computing terms.
  • Any reader-comprehension or usage study of terms such as 'agentic' among non-specialists.
  • The next round of additions, which would test Sokolowski's sense that this batch is unusual.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories