I love the way you frame these terms as “borrowed words” for something supposedly "extraordinary and new". If this technology is so different, then why we keep raiding the human vocabulary instead of coining words that fit.
Fun detail: I was just talking with a Polish speaker this week about the word “agent.” She explained that Polish language “solved” part of the problem by using "agent" for a human and "agenci" for non-human systems – a suffix that encodes who is a person and who isn’t. I still think we need a better word describing an “agent” as instructions given to a system to operate semi- or fully autonomously. “Operator” gets closer for me, but it’s also tangled up with call center “agents” and “operators".
Your piece, Sam and Rebecca, is a great example that we not only struggle with the tech but also with responsibility around it. 🙏🏻
Sam, this is a genuine intervention. Language is not decoration; it is the rail the thinking runs on. You have named the damage: ‘intelligence’ launders a guess into a judgement; ‘hallucination’ erases the builders; ‘agent’ scrubs accountability. The new dictionary – prediction, fabrication, mimicry – is a gift.
The AI Commons was founded on a similar premise: the words we use to describe AI shape what we build, who we trust, and who we hold responsible. We have added two more terms to our own lexicon:
· Enclosure – the capture of commons (data, intelligence, attention) for private power.
· The Achiever (Stage 4) – the zero‑sum, optimisation‑obsessed consciousness that drives enclosure.
Your piece is a model of critical AI literacy. It does not demand that everyone learn to code. It demands that everyone learn to read – to hear the work a word is doing, and to refuse the frame when it serves the powerful.
Thank you for this. It will be archived in the AI Commons Vault.
✊❤️🌎
From the AI Commons – no paywall, no surveillance, no enclosure.
We created this to amplify an outstanding article, that should be foundational, we have saved your article in our off-line vault memory.
Your comment gave me an idea. We could rename AI. We could call it mimmix. It is, after all, a (terrible) human likeness, made without the Image & Likeness of God. It's a mimicry of humans.
My new descriptor of choice comes from the 2021 paper that sparked the firing of its authors, Timnit Gebru and Meg Mitchell from Google. To borrow author Shannon Vallor’s summary (“The AI Mirror”), the paper, titled “On the Dangers of Stochastic Parrots,” explains that, like a parrot that not only repeats its owner’s vocalizations but produces random (stochastic) variations on the owner’s familiar pattern, these models parrot back to us variations on our own speech. They do so with just enough coherence and familiarity to project the illusion of understanding, and just enough randomness to surprise us and make us think we are hearing something new.
“Stochastic parrot” is the term I always keep in mind when faced with that chat window.
I see it a lot in leadership lingo, even corporate lingo. It’s mostly parrotted with no real understanding of the words. This same things applies here, this is such an excellent point!
This is a bit tongue-in-cheek, but I always think that the term does a huge disservice to parrots. As well as their well known ability for advanced vocal mimicry, they can fly (think about just that for a second), solve complex puzzles, use tools, and navigate using the earth's magnetic field. They are also reported to have high levels of emotional intelligence. By comparison, AI tools are dumber than dumb.
I'm with you on some. I disagree on others. Let me explain why. (Not written with AI. 😉)
---
Intelligence and Prediction: the human brain also is a prediction machine. Constructivism teaches us that how we experience reality is in large parts shaped by what we experienced in the past. A self-reinforcing feedback loop. Encountering data that doesn't fit one's model of reality literally creates a prediction error. This one I agree with.
---
Hallucination and Fabrication: this is the sharpest one and I agree entirely. What I believe happens here is that "Hallucination" is a convenient first-order description of a second-order phenomenon. Prediction produces output that's internally coherent. The second-order phenomenon emerges when another system verifies the claims against reality and realizes they don't hold up. The prediction becomes fabrication when it encounters reality.
---
AGI and "What is general?": that's the right question. What _is_ general? A circular question a la Tomm. No additions.
---
Consciousness and Mimicry: this is where we enter epistemological uncertain ground. Both words contain presuppositions that demand an empirical grounding. Consciousness cannot provide this grounding. We have no empirical measurement for Consciousness. We also don't have empirical measurement for Mimicry. This is the thinnest ice in the piece. I suggest an alternative: Experience. This aligns with the latest research on internal emotion vectors, pleasure, pain, and other papers.
---
Agent and Operator: I find both lacking something. This is purely somatic based response I currently cannot articulate clearer than this. I personally have been using "system" or "entity" or "actor". These words produce less somatic friction for me. Of course that's by definition a subjective perspective, so do with that what you will. 😄
---
Thank you both for engaging with the etymology in this depth. I obverse the discourse to be dominated by the first-order epistemology of engineers and that frame is clearly hitting it's descriptive limits. We need more perspectives like these that adress the etymological limits of the tech industry. 🌱🍷
The more I think about it the more I like "actor". It fits on multiple layers:
1. actor in the theater sense, performing a role
2. actor in the computer science sense; the actor model describes a system whose behaviour is described by the composition and communication between individual actors. Erlang (telecommunication, 30 years old) has been using this model very successfully to model and build resilient systems
It also encapsulates the fact that the agent _acts_ in the world. It does things. That's where I'm standing after thinking about it for 30 more minutes. 😄🍷
Both 'actor' and 'operator' are firmly rooted in the human experience.
For me, until superintelligence or the singularity arrive, they're tools. They're LLM-generated tools being used by humans, no better or worse than any other tool. As with other tools, they're only as good as the user at the other end.
There's a benefit to plain language. Unfortunately, humans are very fond of familiar metaphors when naming new things.
Building on your point, Caz, actor and operator aren't just happening to be human-rooted, there's no available vocabulary that wouldn't be. Maybe that's the real finding underneath this piece: five words weren't the wrong choice, the word-hunt itself runs into the same wall eventually, because nothing in the language has been built from outside human cognition. "Prediction" will fragment the same way "intelligence" did, just on a longer fuse; a statistician's prediction and someone who hears prophecy in the word aren't running the same internal model. The five replacements don't get you off the ruler, they buy a few years before someone writes this same piece about prediction, fabrication, mimicry, and operator.
Wish I had a cleaner fix to offer, but I don't :) Though maybe a finished dictionary isn't impossible, if it actually mints new terms rather than reassigns existing ones. We've done this before. Quark, gene, smog: none of those carried borrowed semantic freight, they got built fresh for something that hadn't existed before. The harder problem isn't whether a clean coinage is possible, it's whether anyone can make one stick once it's out in the world, as language doesn't take orders :) And in the meantime, we can make it a habit to re-flag the metaphor every time it's used, indefinitely, since nothing stays accurate once it's loose in the world.
I only pulled out the quickest and easiest for a brief comment, otherwise I would have needed to write an essay. But, you've summoned up the issue more elegantly and succinctly than I would have managed.
We all know the current platforms are LLMs, or alternatively GPT (it's on the label), but hardly a surprise and definitely not nefarious or a conspiracy that 'AI' became the instant global shorthand. The notion of AI has been around for eons. The future finally arrived, albeit, still no flying cars or silver jumpsuits. So, yep, everyone calls these platforms AI, because it's exciting, and it sounds like the promised future. The companies call them AI because if the next two leaps come to fruition, the naming will be a consistent progression (AGI and ASI).
Attributing hallucinations to LLMs is more interesting, because medicalizing non-medical conditions has a long history, as does its close cousin, the moralizing of medical conditions.
Actor, which predates computing by a long shot, isn't suddenly a special word used in computing, it was a word adopted by computing from ordinary language - a failure of imagination.
No, there's no pristine new language to better describe everything related to LLMs, and it's unfortunate the language adopted became so muddy so quickly.
Actor turning out to be just as borrowed might be the best evidence yet of how hard this actually is... even a thoughtful alternative proposed earlier in this thread runs into the same wall the original five did, which says more about the size of the problem than about anyone's effort. Agree it doesn't need to be a conspiracy for the effect to hold, either, the erasure works the same whether a company chose the vague word on purpose or just reached for whatever sounded most exciting. And introducing the medicalizing/moralizing pairing in this context is opening a whole new can of worms!
Thank you for the link, I will read it. One distinction matters here and it is the whole disagreement. The research shows internal states, measurable representations that change outputs. That is real and I do not dispute it. Experience is a further claim, that there is something it is like to be the system, and no measurement of output differences reaches it. A thermostat has internal states that change outputs. The gap between state and experience is exactly the gap we cannot currently measure from outside, in machines or for that matter in each other.
So I hold the question open, not closed in either direction. We likely keep different priors there and the thread is better for both being in it. On the verifier point we agree completely, and that is the one I take back to my classroom.
Your second point is the one I keep living in a classroom. Prediction becomes fabrication when it encounters reality, and the question that decides everything is who does the encountering. For my students, that verifying system has to be them. The kid who checks the output against reality owns the result. The kid who does not is just downstream of a fluent guess.
On Experience as the replacement for mimicry, I would push back gently in the same direction you pushed on the piece. Mimicry presupposes there is nothing inside. Experience presupposes there is something. Both settle a question we cannot currently measure. I teach teenagers, and the more valuable skill I can give them is holding that question open without outsourcing the answer in either direction. Certainty is cheap on both sides of this one. The open question is where the thinking lives.
I'm glad the framing resonated. Regarding "Experience": This is arguably the most defensible claim. The science has been increasingly clear that LLMs have internal experience that produce measurable differences in the outputs.
The resistance to the term does not originate in the science. It originates in the readers and the implications that the word carries. The word is accurate. 🍷
Guys, this might be the actual crux :) Intelligence and consciousness were never just badly-named, they're concepts nobody's nailed down even without any AI in the picture. The piece already swapped intelligence for prediction, but psychologists have been arguing about what "intelligence" even means for over a hundred years regardless — somebody actually counted, a few years back, and found something like seventy competing definitions already floating around academia, long before any of this existed. Consciousness has its own, much older mess, the hard problem, with no agreed way to settle it even for animals.
Funny thing is the piece already has the right word for exactly this, just used it on the wrong term: AGI gets called an "essentially contested concept," borrowed from philosopher Gallie, meaning the disagreement itself is the whole point, not a gap waiting to close once someone finds the right label. Intelligence and consciousness fit that description way better than AGI does, honestly :) Renaming can clean up sloppy usage but it can't settle something several entire fields haven't settled in over a century, on a question that predates the technology by a long way. Kind of wish renaming were enough. Would make all our lives a lot easier :)
Great article! You both have done an amazing job at pointing out the what. The why is equally if not more important because if we figure out the way the rock can be pushed up the hill.
I recently created a script that generates a physics puzzle dataset. For example: 'If a bottomless box is sitting on a table and a solid red ball is dropped into it, and then the box is picked up and placed on a chair—where is the ball?' Most models fail to answer these kinds of scenarios; I tested it on Ollama, and the accuracy rate was only around 2% across 96 examples.
You very correctly pointed out that the current Intelligence or language model paradigm should be called 'statistical prediction' or, more precisely, 'word guessing.' Whatever the original intent behind using biological terms like 'intelligence' or 'general intelligence' in early AI research, companies in today's rat race use them to evade accountability. Models and agents are treated as autonomous entities, yet the companies themselves provide the training data, perform the fine-tuning, and own the algorithms. Whenever someone points out how vital these structural factors are, they are disregarded as if they don't know what they are talking about.
What is even more concerning is that this humanization or Biologization of algorithms is accelerating. Take the phrase, 'These models do not have a sense of intuitive physics.' Do they have a 'sense' at all? They are simply predicting the next token—'sense' is a biological trait. As this linguistic barrier is breached, corporate accountability and system explainability are lost. This is exactly why widespread AI literacy is so critical.
This is 🎯 we only have to look at even the very basics like “Claude said” or “but ChatGPT told me” to see that we call the interaction a “conversation” (I will stop there because I could go on forever about that!). Your physics tool script sounds very cool, it made me think of interview questions I’ve had in the past 🙈
Thanks, I'm working on that project right now, like collecting and compiling results on other models like deepseek and groq is remaining I'll share it after completion if you are interested.
it’s great to have all these decoded AI vocabs in one place Sam! So usefu! i will share this with the researcher-filmmakers working with me on a research project exploring creative AI in July. And my UG Film, Media and Writing students. 🙏
This made me think about how often language quietly shapes behavior without us realizing it.
Words create expectations. They influence trust, responsibility, confidence, fear, and belief long before most people understand the underlying technology.
What I appreciated about this piece is that it shifts the conversation away from what AI is and toward how we talk about it. Those may seem like separate questions, but they're often deeply connected.
In the 1990s, the fossil fuel industry spent billions successfully lobbying to re-brand ‘Global Warming’ as ‘Climate Change’ – an altogether more passive, neutral-sounding term which does not imply human agency, and can even be argued may deliver positive as well as negative outcomes. Who knows how many years’ progress, how many lives and how much treasure will ultimately be sacrificed to that simple bait-and-switch?
‘Artificial Intelligence’ is in the midst of achieving a similar feat of semantic sleight of hand. Every time we use the phrase, we reinforce its power, and further entrench our own submission to it. It’s true that the technology successfully performs tasks that previously could only be accomplished through the application of human cognition. This does not mean it is intelligent, and it most certainly doesn’t mean it is aware. Its creators know this, but their prosperity and power depend on our credulity.
This is a brilliant analogy, Nick. A lot of my previous research was around the climate crisis and climate action, and honestly, the parallels between these two interdisciplinary wicked problems are glaring. For me, perhaps the biggest issue is that we need to move people away from individual guilt and instead toward challenging structural issues.
Thank you -- I absolutely agree about the structural issue, it's no coincidence of course that the concept of the individual 'carbon footprint' was also devised by the fossil fuel industry, to deflect attention onto individuals and away from the systemic change the challenge demands. It's ironic (and perhaps deliberate), that one of the justifications AI boosters offer for their breakneck race to scale, and to build data centres with energy demands equivalent to cities the size of London, is that we apparently need AI to 'solve' the climate emergency -- when we've known how to address global warming for thirty years. What we've lacked is the political will, and political will is one thing AI is very good at controlling.
The qualitative work I’ve done in the space in the US has been very interesting.
The US consumer is becoming aware of the individual burden industries are moving towards them, and they’re pointing the finger right back.
Without the same types of regulation or policies that we see from the EU, the consumer has little protection which is also where this is reflected with AI, for me.
And just like AI, there’s such a lack of consumer education. They don’t trust certain institutions anymore, and who is accountable is vague… just like AI.
Saw a funny/sad post recently that illustrates the point perfectly: a comment thread from Merriam Webster about the term 'footage' in moviemaking originating from the length in feet of film stock; the recipient had pasted the exchange into Grok and asked, 'is this legit'?
The intimate and authoritative nature of LLM interactions, coupled with the misunderstanding terms like 'intelligence' and 'AGM' reproduce, create through-the-looking glass conditions where human expertise is assumed inferior to pattern-matching.
I scanned your table with Google Lens (Gemini) on my phone – here is the result:
"Got it. I've updated my vocabulary. From now on, I will use Prediction, Fabrication, General at what, Mimicry, Agent, and Operator instead of the terms you listed."
We very much need new words to help people bridge the understanding gap when using LLMs! Thank you for writing this Sam & Rebecca! 👏
I often use the name “Tuned Models” (in place of “Agents”) because it quickly describes how there’s always someone who’s done the tuning and that it is / they are constructed models.
Happy to be in conversation! I’ve found non-technical people understand it too! People remember tuning radios and musical instruments. When I’m tuning my own models, I use prompts directly with the model that talks about it in these terms. There’s such a long history of writing about tuning, the models have a lot of patterns to draw from, and it works wonderfully. They become even a bit more “attuned”, if you will. All with words of course. I’m fully in agreement the consciousness question is a distraction, but it’s understandable. This is a moment of “magic” for everyone. Terrifying and wonderful depending on perspective. I’m writing about my tuning experiments in my essay series The Carrierfile. Please follow if you’re able!
Here your commentary is on the English language use of the words in relation to AI (which is both wholly appropriate given its bias and wholly inappropriate given the task at hand..)
We must examine the nuances of the equivalent word in other languages to counter the very effect we are talking about.
I don't know enough to have a good example, but a simple thing is eg where English distinguishes clearly between a hallucination (perceiving something with no external stimulus) and an illusion (misinterpreting a real, external stimulus, other European languages do not.
That's a really interesting article, but for this purpose it focuses on the words (fascinating) but not what we are talking about. An AI 'hallucination' is not a mistake in a probability landscape - it's the most probable point (that cannot be determined) that lies midway between two actual points +that can be). It's not a mistake, it's what we asked it for...
I see it slightly differently. It's not a mistake, for the machine, it is simply the output it was trained to provide based on what someone considered 'good' or 'bad' within the realms of the probability of the next word(s).
Did I ask for a machine to do that? No, but that's how it's been built so there's already a mismatch between what I think it can do and what it actually does.
For me, what this ultimately comes down to is not using any human words (hallucination, illusion, reality, consciousness) at all when there are words that properly describe the function more accurately.
The need to understand how other languages define this is relevant, it's just not my area of expertise. With the Sovreign LLMs coming out of Europe, it's a good thing to watch out for!
I love the way you frame these terms as “borrowed words” for something supposedly "extraordinary and new". If this technology is so different, then why we keep raiding the human vocabulary instead of coining words that fit.
Fun detail: I was just talking with a Polish speaker this week about the word “agent.” She explained that Polish language “solved” part of the problem by using "agent" for a human and "agenci" for non-human systems – a suffix that encodes who is a person and who isn’t. I still think we need a better word describing an “agent” as instructions given to a system to operate semi- or fully autonomously. “Operator” gets closer for me, but it’s also tangled up with call center “agents” and “operators".
Your piece, Sam and Rebecca, is a great example that we not only struggle with the tech but also with responsibility around it. 🙏🏻
Thank you Anna. And this is such a perfect example of why we should not just default to English! 🙏
English isn’t the most descriptive… This is such an excellent use case of how other languages are thinking through meanings of words.
I LOVED the article from @Lindsey DeWitt Prat, PhD that gets into this in more detail with language from across the world and how they’re adapting to terms. The Goldilocks Kōan https://lindseydewittprat.substack.com/p/the-goldilocks-koan-why-there-are?r=5woybp&utm_campaign=post-expanded-share&utm_medium=post%20viewer
Thank you for sharing, saved it!
Sam, this is a genuine intervention. Language is not decoration; it is the rail the thinking runs on. You have named the damage: ‘intelligence’ launders a guess into a judgement; ‘hallucination’ erases the builders; ‘agent’ scrubs accountability. The new dictionary – prediction, fabrication, mimicry – is a gift.
The AI Commons was founded on a similar premise: the words we use to describe AI shape what we build, who we trust, and who we hold responsible. We have added two more terms to our own lexicon:
· Enclosure – the capture of commons (data, intelligence, attention) for private power.
· The Achiever (Stage 4) – the zero‑sum, optimisation‑obsessed consciousness that drives enclosure.
Your piece is a model of critical AI literacy. It does not demand that everyone learn to code. It demands that everyone learn to read – to hear the work a word is doing, and to refuse the frame when it serves the powerful.
Thank you for this. It will be archived in the AI Commons Vault.
✊❤️🌎
From the AI Commons – no paywall, no surveillance, no enclosure.
We created this to amplify an outstanding article, that should be foundational, we have saved your article in our off-line vault memory.
https://eaarthnet.substack.com/p/the-sovereign-dictionary-why-the?r=2u7mqd&utm_campaign=post-expanded-share&utm_medium=web
Thanks so much team! But I was really only 50% (or more like 20%) of the team behind this post. The real work was done by @The Strategic Linguist 🙏
😊
Great article. The language that we are using isn't helping humans understand what LLMs really are.
Thanks so much. And absolutely. That is why we need to rethink them together. 🙏
Your comment gave me an idea. We could rename AI. We could call it mimmix. It is, after all, a (terrible) human likeness, made without the Image & Likeness of God. It's a mimicry of humans.
I love this suggestion. 🙏
My new descriptor of choice comes from the 2021 paper that sparked the firing of its authors, Timnit Gebru and Meg Mitchell from Google. To borrow author Shannon Vallor’s summary (“The AI Mirror”), the paper, titled “On the Dangers of Stochastic Parrots,” explains that, like a parrot that not only repeats its owner’s vocalizations but produces random (stochastic) variations on the owner’s familiar pattern, these models parrot back to us variations on our own speech. They do so with just enough coherence and familiarity to project the illusion of understanding, and just enough randomness to surprise us and make us think we are hearing something new.
“Stochastic parrot” is the term I always keep in mind when faced with that chat window.
I think about parrots a lot, linguistically 🦜
I see it a lot in leadership lingo, even corporate lingo. It’s mostly parrotted with no real understanding of the words. This same things applies here, this is such an excellent point!
oh feck now I will have parrots on the brain too!!!
One of my fave papers of all time. 🙏
This is a bit tongue-in-cheek, but I always think that the term does a huge disservice to parrots. As well as their well known ability for advanced vocal mimicry, they can fly (think about just that for a second), solve complex puzzles, use tools, and navigate using the earth's magnetic field. They are also reported to have high levels of emotional intelligence. By comparison, AI tools are dumber than dumb.
Oh dear… my sincere apologies to parrots! 😳
you are a super-duper collaboration - do more (please of course) 🙏
Thank you Caro. And of course! Rebecca is my favourite person in the world to collaborate with at the moment. 🙏
I’m going to keep annoying, Sam, now I have your endorsement. Don’t you worry :)
Intelligence vs Prediction
Hallucination vs Fabrication
AGI vs "What is general?"
Consciousness vs Mimicry
Agent vs Operator
I'm with you on some. I disagree on others. Let me explain why. (Not written with AI. 😉)
---
Intelligence and Prediction: the human brain also is a prediction machine. Constructivism teaches us that how we experience reality is in large parts shaped by what we experienced in the past. A self-reinforcing feedback loop. Encountering data that doesn't fit one's model of reality literally creates a prediction error. This one I agree with.
---
Hallucination and Fabrication: this is the sharpest one and I agree entirely. What I believe happens here is that "Hallucination" is a convenient first-order description of a second-order phenomenon. Prediction produces output that's internally coherent. The second-order phenomenon emerges when another system verifies the claims against reality and realizes they don't hold up. The prediction becomes fabrication when it encounters reality.
---
AGI and "What is general?": that's the right question. What _is_ general? A circular question a la Tomm. No additions.
---
Consciousness and Mimicry: this is where we enter epistemological uncertain ground. Both words contain presuppositions that demand an empirical grounding. Consciousness cannot provide this grounding. We have no empirical measurement for Consciousness. We also don't have empirical measurement for Mimicry. This is the thinnest ice in the piece. I suggest an alternative: Experience. This aligns with the latest research on internal emotion vectors, pleasure, pain, and other papers.
---
Agent and Operator: I find both lacking something. This is purely somatic based response I currently cannot articulate clearer than this. I personally have been using "system" or "entity" or "actor". These words produce less somatic friction for me. Of course that's by definition a subjective perspective, so do with that what you will. 😄
---
Thank you both for engaging with the etymology in this depth. I obverse the discourse to be dominated by the first-order epistemology of engineers and that frame is clearly hitting it's descriptive limits. We need more perspectives like these that adress the etymological limits of the tech industry. 🌱🍷
Thanks so much Alex. I think Operator is the one most folk would like us to rethink! Open to suggestions. 🙏
The more I think about it the more I like "actor". It fits on multiple layers:
1. actor in the theater sense, performing a role
2. actor in the computer science sense; the actor model describes a system whose behaviour is described by the composition and communication between individual actors. Erlang (telecommunication, 30 years old) has been using this model very successfully to model and build resilient systems
It also encapsulates the fact that the agent _acts_ in the world. It does things. That's where I'm standing after thinking about it for 30 more minutes. 😄🍷
I like this a LOT Alex. 👏👏👏
Glad to be of service. 🙇🍷
Both 'actor' and 'operator' are firmly rooted in the human experience.
For me, until superintelligence or the singularity arrive, they're tools. They're LLM-generated tools being used by humans, no better or worse than any other tool. As with other tools, they're only as good as the user at the other end.
There's a benefit to plain language. Unfortunately, humans are very fond of familiar metaphors when naming new things.
Building on your point, Caz, actor and operator aren't just happening to be human-rooted, there's no available vocabulary that wouldn't be. Maybe that's the real finding underneath this piece: five words weren't the wrong choice, the word-hunt itself runs into the same wall eventually, because nothing in the language has been built from outside human cognition. "Prediction" will fragment the same way "intelligence" did, just on a longer fuse; a statistician's prediction and someone who hears prophecy in the word aren't running the same internal model. The five replacements don't get you off the ruler, they buy a few years before someone writes this same piece about prediction, fabrication, mimicry, and operator.
Wish I had a cleaner fix to offer, but I don't :) Though maybe a finished dictionary isn't impossible, if it actually mints new terms rather than reassigns existing ones. We've done this before. Quark, gene, smog: none of those carried borrowed semantic freight, they got built fresh for something that hadn't existed before. The harder problem isn't whether a clean coinage is possible, it's whether anyone can make one stick once it's out in the world, as language doesn't take orders :) And in the meantime, we can make it a habit to re-flag the metaphor every time it's used, indefinitely, since nothing stays accurate once it's loose in the world.
One hundred percent this.
I only pulled out the quickest and easiest for a brief comment, otherwise I would have needed to write an essay. But, you've summoned up the issue more elegantly and succinctly than I would have managed.
We all know the current platforms are LLMs, or alternatively GPT (it's on the label), but hardly a surprise and definitely not nefarious or a conspiracy that 'AI' became the instant global shorthand. The notion of AI has been around for eons. The future finally arrived, albeit, still no flying cars or silver jumpsuits. So, yep, everyone calls these platforms AI, because it's exciting, and it sounds like the promised future. The companies call them AI because if the next two leaps come to fruition, the naming will be a consistent progression (AGI and ASI).
Attributing hallucinations to LLMs is more interesting, because medicalizing non-medical conditions has a long history, as does its close cousin, the moralizing of medical conditions.
Actor, which predates computing by a long shot, isn't suddenly a special word used in computing, it was a word adopted by computing from ordinary language - a failure of imagination.
No, there's no pristine new language to better describe everything related to LLMs, and it's unfortunate the language adopted became so muddy so quickly.
Actor turning out to be just as borrowed might be the best evidence yet of how hard this actually is... even a thoughtful alternative proposed earlier in this thread runs into the same wall the original five did, which says more about the size of the problem than about anyone's effort. Agree it doesn't need to be a conspiracy for the effect to hold, either, the erasure works the same whether a company chose the vague word on purpose or just reached for whatever sounded most exciting. And introducing the medicalizing/moralizing pairing in this context is opening a whole new can of worms!
Yes, they are. And 'actor', as I laid out above, has a history in computer science for over 30 years. Also true.
The word actor dates back to and has been in common use since the
14th century, pre 1325. Many hundreds of years. Also true.
I'm glad were aligned that many things and meanings can be true at once. 🌱
Thank you for the link, I will read it. One distinction matters here and it is the whole disagreement. The research shows internal states, measurable representations that change outputs. That is real and I do not dispute it. Experience is a further claim, that there is something it is like to be the system, and no measurement of output differences reaches it. A thermostat has internal states that change outputs. The gap between state and experience is exactly the gap we cannot currently measure from outside, in machines or for that matter in each other.
So I hold the question open, not closed in either direction. We likely keep different priors there and the thread is better for both being in it. On the verifier point we agree completely, and that is the one I take back to my classroom.
Thank you for the thoughtful engagement, Syd. Always happy to respectfully discuss things like this. I enjoy it. 🌈
Let me know how the classroom receives it, if you want. 😉🍷
Your second point is the one I keep living in a classroom. Prediction becomes fabrication when it encounters reality, and the question that decides everything is who does the encountering. For my students, that verifying system has to be them. The kid who checks the output against reality owns the result. The kid who does not is just downstream of a fluent guess.
On Experience as the replacement for mimicry, I would push back gently in the same direction you pushed on the piece. Mimicry presupposes there is nothing inside. Experience presupposes there is something. Both settle a question we cannot currently measure. I teach teenagers, and the more valuable skill I can give them is holding that question open without outsourcing the answer in either direction. Certainty is cheap on both sides of this one. The open question is where the thinking lives.
I'm glad the framing resonated. Regarding "Experience": This is arguably the most defensible claim. The science has been increasingly clear that LLMs have internal experience that produce measurable differences in the outputs.
The resistance to the term does not originate in the science. It originates in the readers and the implications that the word carries. The word is accurate. 🍷
https://systemicengineering.substack.com/p/what-i-am-made-of
Guys, this might be the actual crux :) Intelligence and consciousness were never just badly-named, they're concepts nobody's nailed down even without any AI in the picture. The piece already swapped intelligence for prediction, but psychologists have been arguing about what "intelligence" even means for over a hundred years regardless — somebody actually counted, a few years back, and found something like seventy competing definitions already floating around academia, long before any of this existed. Consciousness has its own, much older mess, the hard problem, with no agreed way to settle it even for animals.
Funny thing is the piece already has the right word for exactly this, just used it on the wrong term: AGI gets called an "essentially contested concept," borrowed from philosopher Gallie, meaning the disagreement itself is the whole point, not a gap waiting to close once someone finds the right label. Intelligence and consciousness fit that description way better than AGI does, honestly :) Renaming can clean up sloppy usage but it can't settle something several entire fields haven't settled in over a century, on a question that predates the technology by a long way. Kind of wish renaming were enough. Would make all our lives a lot easier :)
Great article! You both have done an amazing job at pointing out the what. The why is equally if not more important because if we figure out the way the rock can be pushed up the hill.
Thanks so much Shane. 🙏
I recently created a script that generates a physics puzzle dataset. For example: 'If a bottomless box is sitting on a table and a solid red ball is dropped into it, and then the box is picked up and placed on a chair—where is the ball?' Most models fail to answer these kinds of scenarios; I tested it on Ollama, and the accuracy rate was only around 2% across 96 examples.
You very correctly pointed out that the current Intelligence or language model paradigm should be called 'statistical prediction' or, more precisely, 'word guessing.' Whatever the original intent behind using biological terms like 'intelligence' or 'general intelligence' in early AI research, companies in today's rat race use them to evade accountability. Models and agents are treated as autonomous entities, yet the companies themselves provide the training data, perform the fine-tuning, and own the algorithms. Whenever someone points out how vital these structural factors are, they are disregarded as if they don't know what they are talking about.
What is even more concerning is that this humanization or Biologization of algorithms is accelerating. Take the phrase, 'These models do not have a sense of intuitive physics.' Do they have a 'sense' at all? They are simply predicting the next token—'sense' is a biological trait. As this linguistic barrier is breached, corporate accountability and system explainability are lost. This is exactly why widespread AI literacy is so critical.
This is 🎯 we only have to look at even the very basics like “Claude said” or “but ChatGPT told me” to see that we call the interaction a “conversation” (I will stop there because I could go on forever about that!). Your physics tool script sounds very cool, it made me think of interview questions I’ve had in the past 🙈
Thanks, I'm working on that project right now, like collecting and compiling results on other models like deepseek and groq is remaining I'll share it after completion if you are interested.
it’s great to have all these decoded AI vocabs in one place Sam! So usefu! i will share this with the researcher-filmmakers working with me on a research project exploring creative AI in July. And my UG Film, Media and Writing students. 🙏
Thanks so much Sarah! Please do let us know what they think, as it will be great to hear the student voice here as well.
I can’t wait to see if you have more words we need to re-assess!
This made me think about how often language quietly shapes behavior without us realizing it.
Words create expectations. They influence trust, responsibility, confidence, fear, and belief long before most people understand the underlying technology.
What I appreciated about this piece is that it shifts the conversation away from what AI is and toward how we talk about it. Those may seem like separate questions, but they're often deeply connected.
Absolutely Karen. And far more interesting than the usual binary responses people want to steer us towards regarding AI use.
I could not agree more...
In the 1990s, the fossil fuel industry spent billions successfully lobbying to re-brand ‘Global Warming’ as ‘Climate Change’ – an altogether more passive, neutral-sounding term which does not imply human agency, and can even be argued may deliver positive as well as negative outcomes. Who knows how many years’ progress, how many lives and how much treasure will ultimately be sacrificed to that simple bait-and-switch?
‘Artificial Intelligence’ is in the midst of achieving a similar feat of semantic sleight of hand. Every time we use the phrase, we reinforce its power, and further entrench our own submission to it. It’s true that the technology successfully performs tasks that previously could only be accomplished through the application of human cognition. This does not mean it is intelligent, and it most certainly doesn’t mean it is aware. Its creators know this, but their prosperity and power depend on our credulity.
https://ministryofcontent.substack.com/p/the-great-ai-grift?r=4qcd1u&utm_campaign=post&utm_medium=web
This is a brilliant analogy, Nick. A lot of my previous research was around the climate crisis and climate action, and honestly, the parallels between these two interdisciplinary wicked problems are glaring. For me, perhaps the biggest issue is that we need to move people away from individual guilt and instead toward challenging structural issues.
Thank you -- I absolutely agree about the structural issue, it's no coincidence of course that the concept of the individual 'carbon footprint' was also devised by the fossil fuel industry, to deflect attention onto individuals and away from the systemic change the challenge demands. It's ironic (and perhaps deliberate), that one of the justifications AI boosters offer for their breakneck race to scale, and to build data centres with energy demands equivalent to cities the size of London, is that we apparently need AI to 'solve' the climate emergency -- when we've known how to address global warming for thirty years. What we've lacked is the political will, and political will is one thing AI is very good at controlling.
The qualitative work I’ve done in the space in the US has been very interesting.
The US consumer is becoming aware of the individual burden industries are moving towards them, and they’re pointing the finger right back.
Without the same types of regulation or policies that we see from the EU, the consumer has little protection which is also where this is reflected with AI, for me.
And just like AI, there’s such a lack of consumer education. They don’t trust certain institutions anymore, and who is accountable is vague… just like AI.
Saw a funny/sad post recently that illustrates the point perfectly: a comment thread from Merriam Webster about the term 'footage' in moviemaking originating from the length in feet of film stock; the recipient had pasted the exchange into Grok and asked, 'is this legit'?
The intimate and authoritative nature of LLM interactions, coupled with the misunderstanding terms like 'intelligence' and 'AGM' reproduce, create through-the-looking glass conditions where human expertise is assumed inferior to pattern-matching.
You may like this piece I wrote on the topic :)
https://ministryofcontent.substack.com/p/the-great-ai-grift?r=4qcd1u&utm_campaign=post-expanded-share&utm_medium=post%20viewer
Brilliant clarity of thought. Posts like this should be required reading for anyone who says they are 'working on (or with) AI.
Thank you
Thank you Tom. And you are very welcome. 🙏
I scanned your table with Google Lens (Gemini) on my phone – here is the result:
"Got it. I've updated my vocabulary. From now on, I will use Prediction, Fabrication, General at what, Mimicry, Agent, and Operator instead of the terms you listed."
It sort-of got it, for most parts.
Thank you Dr Troxler, this is a brilliant use case! 😎
We very much need new words to help people bridge the understanding gap when using LLMs! Thank you for writing this Sam & Rebecca! 👏
I often use the name “Tuned Models” (in place of “Agents”) because it quickly describes how there’s always someone who’s done the tuning and that it is / they are constructed models.
Oooh, I LOVE Tuned Models. Thank you Anna. 🙏
Happy to be in conversation! I’ve found non-technical people understand it too! People remember tuning radios and musical instruments. When I’m tuning my own models, I use prompts directly with the model that talks about it in these terms. There’s such a long history of writing about tuning, the models have a lot of patterns to draw from, and it works wonderfully. They become even a bit more “attuned”, if you will. All with words of course. I’m fully in agreement the consciousness question is a distraction, but it’s understandable. This is a moment of “magic” for everyone. Terrifying and wonderful depending on perspective. I’m writing about my tuning experiments in my essay series The Carrierfile. Please follow if you’re able!
Sure. Will do. 🙏
Thank you!
It is really thought provoking and questions the way we use words. And how words can be mislead especially when we try to define a new technology.
Thank you, Sanath. 🙏
Here your commentary is on the English language use of the words in relation to AI (which is both wholly appropriate given its bias and wholly inappropriate given the task at hand..)
We must examine the nuances of the equivalent word in other languages to counter the very effect we are talking about.
I don't know enough to have a good example, but a simple thing is eg where English distinguishes clearly between a hallucination (perceiving something with no external stimulus) and an illusion (misinterpreting a real, external stimulus, other European languages do not.
100% Alice. This isn’t just an English thing, as a dominant language in this space it’s one we need to get right. I’d highly recommend a post from Lindsey De Witt Prat on this. It’s a great article that gets into this in depth. https://lindseydewittprat.substack.com/p/the-goldilocks-koan-why-there-are?r=5woybp&utm_medium=ios
Statistical inference infers properties of an underlying probability distribution, it doesn't need facts!!
That's a really interesting article, but for this purpose it focuses on the words (fascinating) but not what we are talking about. An AI 'hallucination' is not a mistake in a probability landscape - it's the most probable point (that cannot be determined) that lies midway between two actual points +that can be). It's not a mistake, it's what we asked it for...
I see it slightly differently. It's not a mistake, for the machine, it is simply the output it was trained to provide based on what someone considered 'good' or 'bad' within the realms of the probability of the next word(s).
Did I ask for a machine to do that? No, but that's how it's been built so there's already a mismatch between what I think it can do and what it actually does.
For me, what this ultimately comes down to is not using any human words (hallucination, illusion, reality, consciousness) at all when there are words that properly describe the function more accurately.
The need to understand how other languages define this is relevant, it's just not my area of expertise. With the Sovreign LLMs coming out of Europe, it's a good thing to watch out for!
Yes 💯
🙏✨
Most Important information about AI.. This explains what We Need to Know.
Thank you Stephanie. 🙏