A terminology proposal

What Is a Terminology Proposal, Exactly?

Before unpacking why any specific proposal matters, it helps to pin down the concept itself. A terminology proposal is a structured argument — sometimes a single sentence on a social platform, sometimes a multi-page terminology proposal PDF submitted to a standards body — that a particular word or phrase should be adopted, retired, or redefined within a given community of practice. It is distinct from a general writing proposal (a pitch to write a book or article) and from a proposal in the marriage sense — an offer made to a specific person seeking a personal commitment. The word "proposal" here carries its older, more formal sense of putting an idea before a group for collective consideration.

The proposal definition in law clarifies this lineage. In contract law and legislative drafting, a proposal is a formal offer to enter into an agreement or to adopt a rule; acceptance by the relevant body transforms it into binding text. Terminology proposals in technical fields follow the same logic informally: they invite a community to "accept" a new label by using it consistently until it achieves critical mass. The UN proposal meaning in multilateral diplomacy — a draft resolution or amendment tabled before a UN body — is the most institutionalized version of this process, complete with co-sponsors, procedural votes, and official language requirements across the organization's six working languages.

A proposal synonym that captures the spirit of terminology work is nomination: you are nominating a word to fill a gap or replace a flawed incumbent. Other useful synonyms include recommendation, submission, and motion, each of which emphasizes a slightly different dimension of the act. If you are asking what is a better word for proposal in a technical context, "specification" works well for engineering-oriented audiences, while "recommendation" suits standards bodies such as the W3C or ISO, where the term already has a defined procedural weight.

The Bluesky Post That Sparked the Conversation

Terminology proposal writing — the formal or informal act of suggesting that a community, institution, or industry adopt a specific new word or phrase — sits at an underappreciated intersection of linguistics, law, and technology. When Karl Olson (@karlrolson.com on Bluesky) posted a concise "terminology proposal" on April 1, 2024, it attracted approximately 3,900 engagements and generated visible discussion threads across the platform. That viral moment is a window into something much larger: how contested vocabulary shapes developer culture, policy, research credibility, and public understanding.

On April 1, 2024 — a date that might tempt readers to dismiss the post as a joke, though the breadth of its serious reception suggests many did not — Karl Olson published a two-word setup, "A terminology proposal:", accompanied by a visual using the widely-recognized disgusted face / interest face reaction-image format. In that meme structure, the "disgusted face" panel labels something the author finds unappealing or misleading, while the "interest face" panel presents the preferred alternative. The post positioned artificial intelligence as the term on the "disgusted" side of the ledger, with a preferred replacement shown in the image rather than in the post's plain text.

Because the proposed replacement term was embedded in the image rather than in indexable text, it was not accessible to search engines or screen readers — an irony not lost on some respondents who engaged with it. The post was reshared by multiple Bluesky accounts spanning individual technologists and academics, generating visible discussion threads about the adequacy of "AI" as a label. The viral spread is itself data: it signals that a meaningful segment of the technically literate Bluesky community feels that the phrase "artificial intelligence" is inadequate, misleading, or actively harmful as a descriptor for contemporary large language models and related systems.

Why it matters: When a terminology critique gains rapid traction among developers, researchers, and policy-adjacent professionals, it reveals a fault line in shared vocabulary. Contested language is not merely aesthetic — it shapes funding decisions, regulatory scope, public trust, and the kinds of questions researchers think to ask.

The structure and distribution of the post are consistent with a long tradition of informal terminology proposals that gain traction precisely because they are pithy, shareable, and emotionally resonant — qualities that formal standards documents rarely achieve. Whether the April 1 date was intentional provocation, coincidence, or a deliberate use of humor to lower rhetorical defenses remains unconfirmed. What is clear is that many readers engaged with it as a sincere intervention, and the image-based format — by placing the proposed alternative outside indexable text — inadvertently made the critique of "artificial intelligence" the most spreadable part of the post.

Why "Artificial Intelligence" Is a Perennial Terminology Battleground

The critique implied by Olson's post is not new, but it is newly urgent. The term "artificial intelligence" dates to a 1955 proposal by John McCarthy for the 1956 Dartmouth Summer Research Project. McCarthy's primary documented motivation for the coinage was to distance the nascent field from Norbert Wiener's cybernetics — specifically, from what McCarthy saw as cybernetics' misguided emphasis on analog feedback systems — rather than to attract funding per se. In a 1988 interview, McCarthy explained that one of his reasons for inventing the term was "to escape association with 'cybernetics'" and to avoid having to accept Wiener as a de facto intellectual leader of the field. Decades later, that original act of naming has itself become a liability. Critics argue the phrase:

• Overclaims cognition. "Intelligence" implies understanding, intentionality, and general reasoning capacity that current systems demonstrably lack under rigorous testing conditions.
• Underclaims mechanism. It says nothing about what the technology actually does — statistical pattern matching over enormous corpora, matrix multiplication at scale, probabilistic next-token prediction.
• Distorts regulation. Legislators who interpret the term literally may either over-regulate benign automation tools or under-regulate genuinely risky systems by treating "AI" as a single undifferentiated category.
• Inflates public fear and hype simultaneously. The same two words that generate existential-risk discourse also appear in marketing copy for spell-checkers and playlist generators, making the term semantically unusable as a policy boundary.

Competing terminology has been proposed at various points by researchers and critics. Stochastic parrots — coined in Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell's 2021 paper "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?" (presented at FAccT 2021; note that "Shmargaret Shmitchell" is a pseudonym used by one of the paper's authors) — emphasizes probabilistic mimicry without comprehension. Machine learning is more mechanistically accurate for a subset of techniques but has itself become a marketing umbrella stretched beyond usefulness. Automated decision systems is preferred in some legal and policy contexts precisely because it resists anthropomorphization. Statistical inference engines, language models, and generative AI each capture different facets while sacrificing others. The sheer abundance of competing proposals is evidence that no single alternative has yet achieved the community consensus required to displace an entrenched term.

Terminology Proposal Examples Across Fields

To understand what successful terminology proposal writing looks like, it helps to examine cases where proposed language actually displaced its predecessor. The table below compares several landmark terminology shifts, the mechanism by which they occurred, and the approximate timeframe from proposal to adoption.

Field	Displaced Term	Proposed Replacement	Mechanism	Approx. Adoption Timeline
Medicine	Mongolism	Down syndrome	WHO recommendation + journal style guide mandates	~15 years (1960s–1970s)
Software engineering	master (default git branch)	main	GitHub made main the default for all new repositories on 1 October 2020, following community pressure beginning around 2018	~3 years (2018–2021)
Climate science	Global warming	Climate crisis / climate change	IPCC reports + newspaper style guide updates	~10 years (ongoing)
Food regulation (EU)	Veggie burger / vegan steak	Plant-based product (restricted labeling)	EU meat terminology proposal via European Parliament debate; Parliament rejected a broad ban (Amendment 165) in October 2020; a revised amendment passed in October 2025, with a subsequent compromise protecting specific anatomical meat names while exempting generic terms like "burger"	~6 years (2019–2025)
AI / ML	Artificial intelligence	Multiple candidates (LLM, generative AI, automated decision system, etc.)	Academic papers, social media proposals, style guides	Ongoing, unresolved

The EU meat terminology proposal is particularly instructive as a case study in how terminology battles play out in formal regulatory contexts. The process began around 2019, when the EU's Committee on Agriculture and Rural Development (AGRI) circulated Amendment 165 arguing that allowing plant-based products to use terms such as "burger," "steak," "sausage," and "bacon" misled consumers and undermined the economic identity of conventional meat products. The European Parliament, however, rejected that amendment outright in October 2020, declining to impose the ban. The debate did not end there: in October 2025, the Parliament voted in favour of a new amendment, and subsequent negotiations with the Council of the EU produced a compromise — specific anatomical meat names (ribeye, T-bone, bacon, drumstick, and others) are now protected, while generic terms such as "burger," "schnitzel," and "sausage" remain available for plant-based products. The multi-year arc illustrates a recurring pattern: informal vocabulary objections must be translated into precise legislative amendment language, survive successive institutional votes, and survive inter-institutional negotiation before they become enforceable — a process that often dilutes the original proposal considerably. The lesson is straightforward: informal terminology proposals and formal regulatory ones follow fundamentally different timelines and require very different coalitions, but both begin with someone naming a problem and suggesting a more accurate label.

How to Write a Terminology Proposal That Actually Works

Whether you are drafting a one-paragraph Bluesky post or a multi-section terminology proposal PDF for a standards body, the elements that make a proposal persuasive are consistent. Strong proposal examples in the terminology space share the following structure:

1. Name the problem precisely. Explain what the incumbent term gets wrong — whether it overclaims, underclaims, carries harmful connotations, or creates legal ambiguity. Vague objections ("this word is bad") rarely persuade any audience beyond those already convinced.
2. Propose a specific alternative. Avoid proposing a category of alternatives; commit to one candidate. If multiple options genuinely exist, rank them with explicit, stated criteria so readers can interrogate your reasoning.
3. Demonstrate the alternative's precision. Show that the proposed term accurately describes the referent without the incumbent's flaws. Provide a definition crisp enough to be operationalized in a style guide or inserted into legal text without rewording.
4. Address adoption friction. Acknowledge that displacing entrenched vocabulary has real costs — backward compatibility with existing literature, reader confusion during transition periods, SEO disruption for publishers relying on established search volume — and explain concretely why the benefits outweigh them.
5. Identify the adoption mechanism. A term does not change because someone proposes it; it changes because key gatekeepers — journal editors, platform default settings, regulatory bodies, influential practitioners — adopt it consistently. Name who needs to act, in what sequence, and what would constitute sufficient uptake to declare the shift successful.
6. Provide terminology proposal examples of usage in context. Draft two or three sentences that deploy the new term in realistic scenarios so readers can evaluate how it reads in practice, whether it introduces new ambiguity, and whether it survives translation into adjacent fields.

In a formal technical documentation context, a terminology proposal PDF submitted to bodies such as ISO, the IETF (via an RFC or Internet-Draft), or the W3C typically adds several additional sections: scope (does this term apply to all documents in the standard, or only normative provisions?), a clear distinction between normative and informative definitions, explicit relationship to existing defined terms in the same or related standards, and a prior-art record — i.e., has anyone else proposed this term, and if so, why was it not previously adopted? That prior-art section is often the most revealing, because it forces the proposer to engage honestly with why previous attempts failed.

The Role of Social Media in Modern Terminology Proposals

Karl Olson's Bluesky post exemplifies a distinctly contemporary form of terminology proposal writing: the viral micro-proposal. Before the social-media era, terminology proposals moved through journals, conference proceedings, and standards-committee mailing lists — channels with high signal and relatively low reach. Social platforms have inverted that ratio, enabling proposals to reach large audiences within hours while compressing the deliberative process that formal channels enforce.

The meme format is particularly effective for terminology critique because it leverages the emotional logic of the reaction-image genre: it associates the incumbent term with a negative affect (disgust, dismissal) and the proposed alternative with a positive one (interest, approval) without requiring a written argument. This is rhetorically powerful but epistemically thin — it persuades through affective association rather than explicit reasoning, which means viral terminology proposals tend to succeed at spreading dissatisfaction with existing terms faster than they succeed at building durable consensus around replacements. Olson's post illustrates a further irony specific to image-based proposals: by embedding the alternative term in a non-indexable image, the proposal ensured that the critique of "artificial intelligence" was the only part that spread freely through text-based sharing and search, while the proposed replacement remained visible only to those who actually viewed the original image.

Still, the viral spread matters for reasons beyond immediate persuasion. When a terminology proposal circulates widely among the practitioners who actually write documentation, train models, draft policy, and build products, it creates a form of distributed peer review at speed. Researchers and engineers who reshared Olson's April 2024 proposal were, in effect, adding their names to a petition — signaling to editors, style-guide maintainers, and platform operators that demand for new vocabulary exists and is not confined to a small critical minority. That signal is legible to institutional decision-makers in a way that a single academic paper, however rigorous, often is not.

Proposal Definition in Law and Standards: Why Precision Is Non-Negotiable

In legal and regulatory contexts, the stakes of terminology proposal writing are highest. The proposal definition in law — an offer capable of acceptance so as to form a binding agreement — maps directly onto how terminology changes propagate in formal rule-making. When the EU's Common Agricultural Policy incorporates a definition of "meat product," every subsequent use of that phrase in contract law, customs declarations, and product-liability litigation is governed by that definition. A poorly worded terminology proposal, once adopted, can generate decades of interpretive problems that cost far more to resolve than careful original drafting would have.

The UN proposal meaning — a draft resolution or operative text tabled before a deliberative body — illustrates the most rigorous end of the spectrum. UN documentation style requires that every defined term appear consistently in all six official languages, with footnotes or annexes reconciling semantic gaps between languages that do not map cleanly onto each other. The terminology proposal PDF format used by UN specialized agencies typically runs to dozens of pages for a single term, tracing etymology, existing usage in prior international instruments, potential conflicts with other currently defined terms in the UN system, and recommended transitional language for documents that will need updating.

For developers and technical writers working at a much smaller scale, this institutional rigor still offers a useful model. A well-structured terminology proposal — even one submitted as a pull request to an open-source project's documentation repository — benefits from: an explicit scope statement (does this term apply to all documentation or only the API reference?), a clear and testable definition, examples of correct and incorrect usage, and a migration path specifying how existing content should be updated and over what timeframe.

Key Takeaways

• Terminology proposal writing is the formal or informal practice of nominating a word or phrase for adoption by a community, ranging from a viral social-media post to a multi-page standards document with prior-art analysis and normative definitions.
• Karl Olson's April 1, 2024 Bluesky post used the "disgusted face / interest face" meme format to critique the term "artificial intelligence" and propose an alternative embedded in the accompanying image, attracting approximately 3,900 engagements and reshares across the platform. Because the proposed replacement appeared only in the image, the critique of "AI" spread more widely than the specific alternative term.
• The term "artificial intelligence" was coined by John McCarthy in his 1955 Dartmouth proposal primarily to distance the emerging field from Norbert Wiener's cybernetics — not, as is sometimes claimed, to attract funding. The term has since been contested on grounds of overclaiming cognition, underclaiming mechanism, and distorting both regulation and public perception, making it a recurring target for terminology proposals from Bender et al.'s "On the Dangers of Stochastic Parrots" onward.
• Successful terminology shifts — such as GitHub making main the default branch for new repositories in October 2020, and the EU's multi-year meat-labeling debate — require not just a compelling proposal but a clear adoption mechanism: platform defaults, journal mandates, regulatory text, or some combination of these.
• The EU meat terminology proposal is a model for how informal vocabulary debates become binding law over a multi-year arc. The European Parliament rejected an initial broad ban (Amendment 165) in October 2020; a revised amendment passed in October 2025 and a subsequent compromise protected specific anatomical meat names while leaving generic terms such as "burger" available to plant-based producers.
• Effective terminology proposal examples name the problem with precision, commit to a specific alternative, demonstrate that alternative's definitional adequacy, address adoption friction honestly, and identify the institutional gatekeepers who must act for the change to take hold.
• Social media enables terminology proposals to reach large audiences rapidly but compresses deliberation — viral dissatisfaction spreads faster than the consensus-building required to make replacements durable and institutionally legible.
• In legal and standards contexts, the proposal definition in law and the UN proposal meaning are reminders that once a term is formally adopted, precision at the drafting stage prevents costly interpretive ambiguity that can persist for decades.
• A proposal synonym such as nomination or recommendation can help writers frame their intervention for different audiences — "nomination" emphasizes agency and advocacy, while "recommendation" signals deference to a body with formal adoption authority.

---

What comes next in the AI terminology debate will likely be determined less by any single proposal — viral or academic — and more by which institutions move first. If major style guides (AP, Chicago, major tech-publication house guides), influential peer-reviewed journals, or regulatory bodies such as the authorities tasked with implementing the EU AI Act settle on preferred alternatives to "artificial intelligence" for specific technical contexts, the broader community will follow their lead. The Bluesky post format may compress deliberation, but it accelerates the first and often hardest stage of any terminology shift: generating enough discomfort with the status quo that practitioners start actively looking for something more precise. In that sense, Karl Olson's terse "A terminology proposal:" does exactly what the best proposals do — not yet winning the argument, but successfully naming the problem in a form that others can share, debate, and eventually act on.