Lexical Compendium

Wait... what is this? Sometimes I come across a word, phrase, idiom, quote, reference, bit of slang, person of interest, etc that either I don't know or I find amusing, interesting, etc. This is a collection of those items so that I can refer back to them in emails, texts, etc.

Recent Entries:

Tag Cloud eponymous grammar tipofmytongue technology television vocabulary jargon entomology internet music spanish games podcast slang history quote movie acronym ai computer-science netspeak space idiom politics legal comics science software
Surprise Me Searching...

AGI

AGI stands for "Artificial General Intelligence." Unlike AI, AGI is meant to connote a system that exhibits human-like intelligence and is not trained for specific tasks.

Related: ASI (Artificial Super Intelligence), ANI (Artificial Narrow Intelligence), p(doom)


(link to this entry)

References:

  1. Wikipedia: Artificial General Intelligence
  2. YouTube: Wozniak: Could a Computer Make a Cup of Coffee? - Steve Wozniak talks pontificates about a computer being smart enough to make a cup of coffee in a random person's home; something that would take a lot of "general" intelligence.

Tags: ai , computer-science

Date Added:
Date Modified:

Engels' Pause

Engels' Pause is a period during the early Industrial Revolution (1790–1840) in England in which economic productivity and GDP per capita rose significantly, but real wages for the working class stagnated or even declined. It was named after Friederich Engles and coined by Robert C. Allen.

Many have theorized that AI will bring about similar changes for today's workers.


(link to this entry)

References:

  1. Wikipedia: Engels' Pause
  2. YouTube: Open AI Founder Sam Altman on Artificial Intelligence's Future | Exponentially - Azeem Azhar interviews OpenAI CEO Sam Altman in 2023 and brings up Engels' Pause.
  3. Aberdeen Investments: How is AI going to change the economy?

Tags: ai , jargon , economics , eponymous

Date Added:

Goodhart's Law

When a measure becomes a target, it ceases to be a good measure.

Coined by British economist and statistician Charles Goodhart, Goodhart's Law describes the tendency for metrics to become distorted once people are incentivized to optimize them directly. A measurement that accurately reflects success under normal circumstances often loses its usefulness when rewards, punishments, or status become tied to the number itself.

Goodhart coined the phrase in the 1970s in an article discussing monetary policy. [1] The law and the perverse incentives [2] they create exist elsewhere. Take a couple examples I've seen recently in my day-to-day:

  • A team judged solely on ticket closure rates may rush tickets closed without properly addressing the customer's problem.
  • Engineers evaluated on AI usage metrics may engage in "token-maxxing" -- artificially increasing AI token consumption to improve their standing on internal dashboards. The metric rises, but the underlying objective of greater productivity or better outcomes may not. [3]

As a joke (hopefully), someone wrote a small utility designed expressly to use up tokens: Burn, baby, burn (those tokens) [4]


(link to this entry)

References:

  1. Wikipedia: Goodhart's Law
  2. Wikipedia: Perverse Incentive
  3. Engineer's Codex: Tokenmaxxing, Promomaxxing, and Misaligned Incentives in Tech
  4. GitHub: "Burn, baby, burn (those tokens)" -- dtnewman

Tags: economist , statistics , eponymous

Date Added:

Jevons Paradox

Jevons Paradox is the economic principle stating that as technological improvements increase the efficiency of resource use, overall consumption of that resource may increase rather than decrease.

Originally observed by 19th-century economist William Stanley Jevons in relation to coal consumption during the Industrial Revolution, the paradox highlights how greater efficiency lowers costs, which can drive higher demand.

Jevons Paradox was in the news a lot early in 2025 after the unveiling of AI models developed by a Chinese start-up called DeepSeek. In a report titled "DeepSeek-V3 Technical Report" published in December, DeepSeek showed that they were able to create this AI model much more efficiently (and cheaper) than their counterparts (i.e. OpenAI's ChatGPT, Meta's Llama, etc).

AI research benefits from more efficient neural networks, yet Jevons Paradox suggests that these efficiency gains will lead to even greater overall computational demand, as more complex models are developed and deployed at scale.

While all of this was in the news, Microsoft's Satya Nadella posted on X: Jevons paradox strikes again! As AI gets more efficient and accessible, we will see its use skyrocket, turning it into a commodity we just can't get enough of.


(link to this entry)

References:

  1. Wikipedia: Jevons Paradox
  2. Planet Money: Why the AI world is suddenly obsessed with a 160-year-old economics paradox
  3. Reuters: Europe's AI bulls pin hopes on 'Jevons Paradox' after DeepSeek rout
  4. Tweet: @satyanadella 26 Jan 2025
  5. DeepSeek
  6. DeepSeek-V3 Technical Report

Tags: eponymous , ai

Date Added:

Lysine Contingency

The lysine contingency is intended to prevent the spread of the animals in case they ever got off the island. Dr. Wu inserted a gene that creates a single faulty enzyme in protein metabolism. The animals can't manufacture the amino acid lysine. Unless they're continually supplied with lysine by us, they'll slip into a coma and die. --Ray Arnold, Jurassic Park

When I was a kid I read Jurassic Park multiple times. I loved that book.

The Lysine Contingency is a made up concept in the novel Jurassic Park by Michael Crichton and is defined in the quote above. I think it could perhaps be used more generically to describe a deliberately engineered dependency on a single critical input, intended to constrain or control a system’s behavior. I've been thinking about this as it relates to AI and guardrails.


(link to this entry)

References:

  1. Jurassic Park (novel)

Tags: literature , science fiction

Date Added:

humanic

For a while now, I've been searching for a term that captures human-made creativity -- as opposed to something created by an LLM. Occasionally I see the word organic used in this context. [5] It works, but it feels a little off. Organic implies something that happens naturally, often without human involvement.

Here I'd like to proffer humanic: a word meaning pertaining to or originating from human agency or creativity, particularly in contrast to work produced by AI. An appropriate antonym in this context might be synthetic. So: humanic and synthetic. I like the symmetry.

The word humanic isn't actually new. Historically, it referred to the study of human nature.[6] But let's reclaim it to also denote cultural artifacts made by humans in a world where if feels like we're just moments away from being saturated with purely synthetic content. Maybe others will organically come upon this usage. I'll use this space to record any sightings.

The word humanic is not necessarily anti-AI, it is descriptive of origin, not ideology. We may need another word to convey sentiment and preference.

Example:

"Oh, I love the theater. It's so refreshing to see content that is entirely humanic."

Note: Anthropic might have also been a good choice, but this, ironically, is already the name of a popular LLM.[7]

Other contenders: anthrogenic, anthropogenic

Updates:

  • In Nov 2025 YouTuber/filmmaker Thomas Flight published a thought-provoking piece on art versus artifice [8]
  • In Nov 2025, a song made by AI topped Billboard's "Country Digital Song Sales" chart [1]
  • In Feb 2026, an AI bot tried to submit a PR, got rejected, and wrote a blog in retaliation. [3]
  • A May 2026 article in The Atlantic [2] contained a link to an interesting pre-print paper [4] suggesting that using LLMs has impacted human speach in terms of the words we use.

(link to this entry)

References:

  1. The Independent: The most downloaded country song in the US right now was made by AI
  2. The Atlantic: The Biggest Tell That Something Was Written by AI
  3. Mastodon Toot @anderseknert - highlighting an interaction between an AI bot and a human with a rejected pull request
  4. arXiv [preprint]: Empirical evidence of Large Language Model’s influence on human spoken communication
  5. Stat Significant: How Much AI Will Audiences Accept in Music and Movies?
  6. Definition: Humanics
  7. Wikipedia: Anthropic
  8. YouTube: Thomas Flight - Why AI "Art" Feels So Wrong

Tags: technology , ai , vocabulary , neologism

Date Added:
Date Modified:

p(doom)

p(doom) stands for "probability of doom" and is a term used when talking about AGI.


(link to this entry)

References:

  1. Wikipedia: p(doom)
  2. NYT: Silicon Valley Confronts a Grim New A.I. Metric
  3. Fast Company: P(doom) is AI’s latest apocalypse metric. Here’s how to calculate your score

Tags: ai , computer-science

Date Added:

stochastic parrot

Stochastic parrot is a term coined by Emily M. Bender, Timnit Gebru, et al. in a 2021 paper on the ethical risks of large language models called "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜."

It refers to how large language models generate text by probabilistically predicting the next word based on patterns learned from massive datasets, rather than understanding or reasoning like a human. The metaphor highlights how such systems mimic language without genuine comprehension.

In December 2022, shortly after ChatGPT was released, Sam Altman of OpenAI tweeted, "i am a stochastic parrot, and so r u."

Stochastic parrot was a 2023 American Dialect Society "Word of the Year."


(link to this entry)

References:

  1. Wikipedia: Stochastic parrot
  2. WSJ: ‘Stochastic Parrot’: A Name for AI That Sounds a Bit Less Intelligent
  3. American Dialect Society: 2023 Words of the Year
  4. Tweet: @sama 04 Dec 2022
  5. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜 - 2021, Emily M. Bender, Timnit Gebru, et al.

Tags: ai , computer_science , vocabulary

Date Added:

vibecoding

Vibecoding (also vibe coding) is the act of developing software by purely defining a project and letting AI do the work.

The term was popularized earlier this year (in 2025) by OpenAI founding member Andrej Karpathy. It has since taken off and I hear it everywhere.

The NYT's Kevin Roose wrote an article on his experience with vibecoding and also covered it in an episode of Hard Fork.

Update May 2025: added Freethink's piece about the technical and cultural shift happening around vibecoding.


(link to this entry)

References:

  1. Wikipedia: vibe coding
  2. NYT: Not a Coder? With A.I., Just Having an Idea Can Be Enough
  3. Tweet: @karpathy 02 Feb 2025 - @karpathy's tweet about "vibe coding"
  4. Hard Fork: Is Google Search Cooked? + We're Getting a U.S. Crypto Reserve? + What You're Vibecoding
  5. Freethink: Vibe coding is rewriting the rules of technology

Tags: ai , software , podcast , slang , netspeak

Date Added:
Date Modified: