strogonoff

The argument that those who refuse to use an LLM will fall behind because they “won’t be able to produce as much” hinges on more LoC being an objective good and typing speed being a bottleneck, which always struck me as a suspicious line of reasoning. In a well-thought-through architecture, you don’t need to type that much; add to that the observation that more LoC usually means more opportunity for error and the net effect suddenly becomes less obviously positive (and this ignoring other negative effects LLM-driven development might have on OSS ecosystem, such as GPL-washing).

Furthermore, when it does come to typing speed, good old non-LLM-powered autocomplete does still exist, and offers less opportunity for subtle errors within otherwise plausibly looking blocks of code.

Measuring information entropy seems to have parallels with the incompleteness theorem and with the hard problem. As we approach the seams (so to speak) in our ability to model the system in its own terms, our otherwise precise maps become distorted, things get self-referential and paradoxical.

I do remember that we are promised to work less thanks to this tech, and yet somehow the opposite happens. We have to work more, and we are paid less (if we are able to get a job).

People who are chronically unemployed always sort of become broken

The causal link may be unclear but I recall there was research showing that at different ages people decline faster once they retire. Sense of purpose is a real thing.

What’s this new trend with allegations how people have jobs for the sake of having jobs?

People have jobs because this is an organised way for us to provide value to other humans and give our lives meaning; a way for us to express ourselves; to earn money that puts food on the table and helps raise our children.

Personally I thought opening doors was about the only thing they did—never seen them engage in any other activity.

I am not sure you provide a counter-argument, IMO offering more people a way to earn money and do something useful (as opposed to delegating these duties to [generally foreign-made] equipment) as a way of maintaining a healthy society still holds.

To assume that the universe is disorderly presumes knowing some esoteric objective truth about how the universe works. Since the field of natural sciences does not provide that truth (by design), and practices that do aim to provide that truth (metaphysics, religion, etc.) seem to have different versions of it, I would say that there’s a fair chance that the universe is not that disorderly and we just lack better models for describing it.

If comparing past industrialisation and automation events with hypothetical LLM dominance in the workforce, some of my questions would be: (0) how limited is the change to a specific industry (e.g., weaving vs. most of intellectual and creative work by everyone starting from children as soon as they can read and type to the elderly)? (1) how many people are affected—in absolute terms not just percentages that ignore population growth? (2) how quickly are they affected? (3) what can people do—where can people move to where they can keep working (e.g., operating/maintaining/manufacturing tractors and looms, intellectual/white collar work, etc.)? (4) what are the organised reskilling processes in place that facilitate said migration? (5) how competitive and diverse, vs. monopolistic and regulated, are the new industries that power the change? how concentrated is created wealth and how many new jobs do they create? (6) what laws (in letter or in spirit) are violated as the change is happening? (7) if it is shown that the current change is about similar to the past in above aspects, does that imply it’s OK that it happens all over again and we have not learned a lesson to go ahead with these things more carefully so that the majority doesn’t suffer as much?

When I was in Bangkok a good while back, I was surprised how there’re doormen at shopping mall—humans to handle simple things which in many parts of the world would certainly not involve a person.

A bit later I thought that it makes sense: it improves customer experience and slightly reduces unemployment in local population—a doorman job is better than no job.

Looking just at ambient air temperature is an easy mistake to make. I used to be like that, always surprised why people whine so much near the coast in winter with their comfortable numbers and thinking I’m superhuman compared to them. I stopped when I learned first-hand that their negative 17 degrees feel as bad as our negative 30.

If you live in a town in New South Wales where the average humidity is less than 50% in the wettest (or, should I say, least dry) season, you might not understand what it feels like in London where the average doesn’t dip below 65% any time of year.

Today London will feel at least 4 degrees Celsius hotter than Hong Kong. The latter is already an extremely unpleasant place to be in these conditions (and had in fact its own very hot weather warning issued), and unlike London it has a very strong culture of air conditioning.

4 degrees might seem like not a lot, but heat extremes are a tricky beast. Once your body cannot evaporate heat fast enough, you’re literally toast.

“A memory begins with tiny changes inside the brain” as truth statement is a basic fallacy of naive physicalism. There is no falsifiable way to ascertain in which direction causality points, nor is natural science even intended provide a definitive answer—it is designed to make predictions and any models that arise in the process are necessarily faulty and do not describe the true nature of underlying reality, which this ultimately comes down to.

I’m not sure it’s possible to truthfully describe what we are missing in reality with a photo.

You can publish a photo with default automatic JPEG processing, say by a phone, and it will certainly look flat. You could also present a masterful interpretation of raw sensor data that uses the most out of the available display space, and the impression might be different.

There is no objectively correct way to represent reality in a photo; even the concept of neutral grey is not a real thing as soon as perception is concerned. A default camera interpretation of light is baseline and safe to maximally avoid awkward edge cases. We all know that time we photograph a bright pink sunset but our phone renders it as pale yellow or orange. However, give the same shot human attention, and even though it may never be as pink as what you have perceived in reality it will pop enough that the viewer will have a similar response.

It is photographer’s job to work raw data in specific ways and make what impressed you stand out to your audience, arranging colours both relative to each other and in absolute display space, however limited it is. Human eyes are incredibly adaptive: we lower our relevant thresholds, adjust our idea of neutral grey—in short, we adapt to given display medium, to given photographic style, etc., and in the end perceive a true lush lagoon in a photo even if our eyes only receive a truly minuscule amount of colour range present in the scene.

How Big AI plans to profit from this inter­me­di­a­tion is an open ques­tion. One AI company has suggested taking a cut of AI-assisted discov­eries. The logis­tics and legal­i­ties would be boggling. Details—what­ever.

Interesting if they pull it off, because clearly they did not have the logistics to pay the people whose IP they used to power the LLMs.

For now, AI compa­nies largely agree on the first step: make workers depen­dent on AI to do their jobs, just as tech fore­bears made workers depen­dent on a certain soft­ware program to share a file, or on a certain website to have friends. This time, however, the soft­ware ulti­mately consumes the worker.

Imagine sudoku with hundreds of subtle, sometimes mutually exclusive rules, and no single valid solution.

This is not about LLMs, by the way. It’s about reviewing any code, including by a fellow human. It’s just that many people mistakenly feel like with LLMs they can lower their guard and accept even if they have not gone through the steps of themselves coming up with their solution and comparing it to the one suggested by the LLM.

The reason is that many correctly see proper review as duplicate work, and while it is justified with another human (because it is (A) instructive and (B) reducing bus factor) with LLMs most people simply can’t be bothered. If you personally can, you are a minority.

Proper review should take longer than writing it yourself, because you need to know the correct solution, understand the proposed solution, and evaluate the difference between the two. When designing it yourself, you just need to know the correct solution and write it, and with modern high-level languages and IDEs with autocomplete writing it is hardly a bottleneck.

Is it really JS with type hints? I had thought it was really a different language that compiles to JS, like a less hardcore answer to Purescript.

No experience with Purescript, but otherwise yes, a language with a different feel. Beats Python in my books. Typings do also serve as documentation but errors won’t compile so it’s not just linting. Make sure to use strict mode so it forces you to pay even more attention to types.

I still occasionally use Python, but TypeScript is my go-to currently.

ML is real. Chatbots are real. “AI” is a marketing term that John McCarthy invented because he wanted more money for a summer study at Dartmouth—direct quote from him.

I tend to get triggered when TypeScript is painted as “JS with type hints”. Coming from Python background, TS and Python with type hints are just so different.

With Python I can’t see myself type-annotating everything (or bringing in pydantic anymore for that matter, it is indeed becoming a blight), but with TypeScript my process is turned on its head: I find it natural and easy to start writing with types and have everything fully typed, and I find the fact that it simply won’t compile if anything is off (compared to Python where it’s more like “one of my N type checkers/linters failed, oh well it still runs though) a useful constraint that gives peace of mind.

What’s your current process for discovering new music?

I quickly got used to mine. It’s not exactly at a 90 degree angle to the desk, the finger pushes about as much down as to the side, and since it’s a natural pinching gesture the thumb on the other side stops any remaining horizontal force. Just find a mouse that has a good design, I’ve seen some I could never use.

A few months ago I switched to a vertical mouse and can’t recommend it enough.

I stopped using VS Code and switched to Neovim some years ago, once I noticed that the former would automatically install random Python packages with typings for libraries without stock typings. The “feature” (part of Microsoft’s official Python extension, which was the only one that worked acceptably well for me in other regards) ended up installing type definitions for a different version of a library than the one my project would use, seemed wildly insecure as it casually ran third-party unvetted code, and was evidently not configurable.

I wish I could add “and I never looked back”, but honestly in the past year or two Neovim started regularly breaking my setup (approximately every upgrade). Had some inklings it might happen eventually… Strictly speaking, 10 years in, nvim is yet to have its first stable version released—which means technically one can’t blame it for instability, but which is useful to keep in mind.

Considering going back to plain vim. I’m sure I will lose many niceties, but hopefully it would not require me to troubleshoot broken functionality in the middle of work.

In all of your examples, a human author would usually prefix with the word "the".

Not really. At least one of my examples is literally based on a real headline from pre-LLM days[0]. If we are talking about a generalised idea of bias, there is no need for “the”. In fact, it would be wrong.

Nobody routinely says the things in your example without some supporting words.

This is a post. One is expected to put more creativity into public speech. Good writing can be expected to be full of metaphors and denser than casual speech. Some writing purposefully uses headline style for impact. It doesn’t mean the author routinely talks like that.

it's density

If it’s a wall of text with filler, it’s LLMs. If it’s dense, it’s LLMs.

You can find plenty of examples of denser, less legible posts from days before LLMs.

are you making the claim that this post was not LLM-generated or at least LLM-assisted

I am making the claim I am making: don’t say someone used an LLM based on such a weak foundation and nothing else.

People will sound like LLMs. Blurring the line with actual writing is the entire point of LLMs and is fully by design; if you encounter a text with zero “tells”, it might as well be made with an LLM if the product works as intended.

[0] Here’s another one: https://ribbonfarm.com/2017/05/25/blockchains-never-forget/

If you have only seen it in LLM output, then it is clearly novel to you when it comes to actual writing. “Speech” in “this turn of speech is novel to you” is intended to mean human written speech.

I will rephrase: if you see a phrase in human speech that you have not seen used in human speech before, don’t penalise human creativity by saying it came from an LLM.

More specifically, the pattern you continue to latch onto dominates writing and has done so for decades. Right here on HN you can find instances of not merely “inanimate subject + verb” but specifically the phrase “bias compounds” from 5 years ago and beyond.

Other examples in use by humans all the time:

— River overflows

— Camera clicks

— LLM hallucinates

— Engine roars

— Secrets rest

— CD player stutters

— Ecosystem explodes

— Door invites

— Stock dips

— Airplane crashes

Antropomorphisation more generally and metaphor even more generally have existed since forever. Authors have played with form and tried to convey the point in different interesting ways since forever. Do you think Homer vibe-wrote The Odyssey?

Yes, LLM chatbots make it exceedingly easy—and it is one of their societal harms—but please do not discredit creativity by insinuating the author didn’t do the work themselves.

I recommend to think twice before penalising potential human creativity by saying that a novel to you turn of speech is a sign of LLM use. If you base your judgement on “unusual phrase”, it should be a sign that you are probably unable to tell.

Considering that LLMs output continuously becomes more human-sounding (by design), you’d either have to continuously run what you write through various detectors and keep changing it or you must resign to inevitably be called an LLM at some point.

Simultaneously, because humans subconsciously mimic what we see, we also converge to sound more LLM-ish.

The harsh reality is that no matter what you write and how much research you put into it—especially if you try to be legible to others and not make grammatical mistakes—someone could discount all that and claim you just prompted an LLM. If they want, they can always find some magic “AI checker” that will return a high enough probability. We all know that with a good enough prompt and with round-trip validation against a checker (there are definitely products with this all built in) it will avoid the common tells, it’s just the matter of a few extra tokens.

It’s somewhat demotivating.

The term “playing field” as far as I am aware usually implies the context of competitive sport, not play in the playful sense.

Trying to “level the playing field” is antithetical to what art is. Art is about self-expression and communication.

If we viewed art as some sort of competition or race, then someone using neural–network-based generative tools could avoid losing the race; however, everybody would be participating in some sport A and the person using ML participates in a completely different sport B. Everyone is running, but one person is riding a scooter.

However, art is not a finite zero-sum game[0]. Despite what formal music education for kids sometimes tries to make it look like, it’s not a competition, there is no global ranking and scoring system for your skill. Many people have an intuitive understanding of that; try going to a live jam to see people participating regardless of their hypothetical skill level.

[0] As further reading on this topic more generally, I recommend Finite and Infinite Games by James Carse.

Humans mirror. We subconsciously use the style of authors we read just like we subconsciously use our parents’ speech patterns. If the majority of text is now created with LLMs, how do we expect humans not to sound like LLMs?

Considering that LLMs output simultaneously becomes more human-sounding, you’d either have to continuously run what you write through various detectors and keep changing it or you must resign to inevitably be called an LLM at some point.

One of the most problematic aspects of this technology in human society is precisely that—not that it absolves us of having to spend effort and be creative, but that if you do apply work someone will easily and legitimately claim that it’s all machine output and you deserve no appreciation for it.

Our joy and suffering depends on our connection with other people, and being useful to other people has since the dawn of humanity been one of the ways we connect.

Our joy and suffering depends on our ability to fulfil our basic needs like shelter and food, and in current socioeconomic environment that ability intimately depends on the ability to create wealth, which in turn depends on our ability to provide value to others.

I can see both being undermined. What public policies reflecting this do you envision?

There’s a world of difference between people simply “copying websites” and providing tools that, along with other kinds of plagiarism[0], do so at scale while benefitting from that commercially.

Sure, you can do the same thing with people, but it’s 1) time-consuming, 2) expensive, 3) prone to whitleblowers refusing to do the shady thing, 4) prone to any competent and productive person involved quitting to do something worthwhile and more profitable instead.

[0] Mind you, “copying websites” is but a drop in the ocean in the grand scale of things.