Language Log

The August 6 Economist discussed AI threats to government bureaucracy:

• "How AI is breaking the British state: Equipped with models and agents, citizens could bring government to a halt"
• "From parking tickets to planning, the state is under siege by AI: The people-versus-state arms race"
• "The tragedy of the commons, AI edition: Britain’s employment courts are clogged with AI cases"

Travis LaCroix, Fintan Mallory, and Sasha Luccioni. "Strategic polysemy in AI discourse: A philosophical analysis of language, hype, and power." In The 2026 ACM Conference on Fairness, Accountability, and Transparency, 2026:

Abstract: This paper examines the strategic use of language in contemporary artificial intelligence (AI) discourse, focusing on the widespread adoption of metaphorical or colloquial terms like “hallucination”, “chain-of-thought”, “introspection”, “language model”, “alignment”, and “agent”. We argue that many such terms exhibit strategic polysemy: they sustain multiple interpretations simultaneously, combining narrow technical definitions with broader anthropomorphic or common-sense associations. In contemporary AI research and deployment contexts, this semantic flexibility produces significant institutional and discursive effects, shaping how AI systems are understood by researchers, policymakers, funders, and the public. To analyse this phenomenon, we introduce the concept of glosslighting: the practice of using technically redefined terms to evoke intuitive—often anthropomorphic or misleading—associations while preserving plausible deniability through restricted technical definitions. Glosslighting enables actors to benefit from the persuasive force of familiar language while maintaining the ability to retreat to narrower definitions when challenged. We argue that this practice contributes to AI hype cycles, facilitates the mobilisation of investment and institutional support, and influences public and policy perceptions of AI systems, while often deflecting epistemic and ethical scrutiny. By examining the linguistic dynamics of glosslighting and strategic polysemy, the paper highlights how language itself functions as a sociotechnical mechanism shaping the development and governance of AI.

I was curious about the Unicode characters involved in the "shrug" text art:

 ̄\_(ツ)_/ ̄

so I ran it through charnames.pl:

Too Faced is a cosmetics company acquired by Estée Lauder in 2016. The name is obviously a pun, substituting too "to an excessive degree" for  the two in two-faced  "deceitful" — though too here apparently connects to the literal idea of a second face, more than to the idea of deceit or hypocrisy.

A couple of days ago, Rachel Strugatz wrote an article with the headline "Et Too Faced?", which relies on a second punning layer, connecting too to Latin tu in the quotation "Et tu, Brute?" from Shakespeare's Julius Caesar.

A decade after selling Too Faced to Estée Lauder, co-founder Jerrod Blandino has lately been chirping about the brand’s post-acquisition fate. Does the guy have a point?

I was struck yesterday by the term Melt-Up in a story about the stock market: "Tech Melt-Up Drives $3.5T Nasdaq 100 Gain in Four Days", Bloomberg 8/5/2026:

A dramatic turnaround in technology stocks has powered a $3.5 trillion increase in the Nasdaq 100’s market capitalization in just four days, driven by strong earnings that have emboldened investors about the AI outlook. The tech benchmark has jumped 9.3% over this period, its sharpest rally since April 2025.

This morning's question:

I'm an ESL textbook editor in Korea, and a grammar dispute has come down to one question: is "What do you say that we get together tonight?" grammatical, with "that" included (vs. the more common zero-complementizer "What do you say we…")?

My understanding is "that" is simply optional here, not ungrammatical.

Email from JK:

My daughter remarked that on several occasions recently she has heard presumably college-educated people use the word “I’s”, in a sentence like "I just picked up Margaret and I’s cat from the vet."

I'm guessing the reason for this jarring (to me) usage is that the sentence could be more logically parsed as

I just picked up Margaret and (my cat) from the vet  than as
I just picked up (Margaret and my) cat from the vet.

The brain turns to I just picked up Margaret and I’s cat from the vet as a solution because because Margaret and I is a strong connection like my cat, and it's encountered first in the sentence.

So it's parsed as I just picked up (Margaret and I)’s cat from the vet.

That sounds awful, but is it acceptable usage (obligatory SMBC reference)?

According to a recent BoingBoing segment,

Ballots for a special election in England's Clacton-by-Sea are the longest ever produced in U.K. electoral history despite¹ only one major party contesting the race.

The footnote:

1. This is the journalist's despite, wherein the word implies disregard when there is in fact a blatant causal relationship.

The singularity may have arrived, but AI systems still can't label diagrams:

As you probably know, Emily Wilson wrote a critical review of Christopher Nolan's Odyssey movie, and Joyce Carol Oates posted a negative evaluation of Wilson's review, among lots of other (mass and social) media coverage.

My favorite so far is today's SMBC:

Last December, Rob Pike got big mad at AI Village. Simon Willison covered the associated events and discussions.

I've recently wrestled several times with similarly frustrating Gmail searches, for recent tickets and the like:

Impossible to put into words how useless Gmail search is. No wait, these guys did it!

[image or embed]

— Alex Selby-Boothroyd (@alexselbyb.bsky.social) July 28, 2026 at 2:22 PM

I get the impression that these problems have multiplied since Gemini started helping out in Gmail, but I might be wrong.