[HN Gopher] LLM Year in Review
       ___________________________________________________________________
        
       LLM Year in Review
        
       Author : swyx
       Score  : 346 points
       Date   : 2025-12-19 20:49 UTC (1 days ago)
        
 (HTM) web link (karpathy.bearblog.dev)
 (TXT) w3m dump (karpathy.bearblog.dev)
        
       | swyx wrote:
       | xposted to https://x.com/karpathy/status/2002118205729562949
        
         | CamperBob2 wrote:
         | And also accessible sans login via
         | https://xcancel.com/karpathy/status/2002118205729562949 .
        
       | TheAceOfHearts wrote:
       | I think one of the things that is missing from this post is
       | engaging a bit in trying to answer: what are the highest priority
       | AI-related problems that the industry should seek to tackle?
       | 
       | Karpathy hints at one major capability unlock being UI
       | generation, so instead of interacting with text the AI can
       | present different interfaces depending on the kind of problem.
       | That seems like a severely underexplored problem domain so far.
       | Who are the key figures innovating in this space so far?
       | 
       | In the most recent Demis interview, he suggests that one of the
       | key problems that must be solved is online / continuous learning.
       | 
       | Aside from that, another major issues is probably reducing
       | hallucinations and increasing reliability. Ideally you should be
       | able to deploy an LLM to work on a problem domain, and if it
       | encounters an unexpected scenario it reaches out to you in order
       | to figure out what to do. But for standard problems it should
       | function reliably 100% of the time.
        
         | lukax wrote:
         | Google is doing that with A2UI. LLM will be able to decide how
         | to present info to the user.
        
       | victorbuilds wrote:
       | Notable omission: 2025 is also when the ghosts started haunting
       | the training data. Half of X replies are now LLMs responding to
       | LLMs. The call is coming from inside the dataset.
        
         | vlod wrote:
         | Any tips to spot this? I want to avoid arguing with a X bot.
        
           | shtack wrote:
           | Really easy: don't argue on the internet. The approach has
           | many benefits.
        
             | jckahn wrote:
             | Also, don't use X.
        
               | bdangubic wrote:
               | also, please just do not use X
        
               | dr_dshiv wrote:
               | Ok, fine, but do you have a better way to build a bot
               | following and expose oneself to trending MAGA memes?
        
               | bdangubic wrote:
               | "truth" social :)
        
       | thoughtpeddler wrote:
       | I appreciate Andrej's optimistic spirit, and I am grateful that
       | he dedicates so much of his time to educating the wider public
       | about AI/LLMs. That said, it would be great to hear his
       | perspective on how 2025 changed the concentration of power in the
       | industry, what's happening with open-source, local inference,
       | hardware constraints, etc. For example, he characterizes Claude
       | Code as "running on your computer", but no, it's just the TUI
       | that runs locally, with inference in the cloud. The reader is
       | left to wonder how that might evolve in 2026 and beyond.
        
         | D-Machine wrote:
         | The section on Claude Code is very ambiguously and confusingly
         | written, I think he meant that the _agent_ runs on your
         | computer (not inference) and that this is in contrast to agents
         | running  "on a website" or in the cloud:
         | 
         | > I think OpenAI got this wrong because I think they focused
         | their codex / agent efforts on cloud deployments in containers
         | orchestrated from ChatGPT instead of localhost. [...] CC got
         | this order of precedence correct and packaged it into a
         | beautiful, minimal, compelling CLI form factor that changed
         | what AI looks like - it's not just a website you go to like
         | Google, it's a little spirit/ghost that "lives" on your
         | computer. This is a new, distinct paradigm of interaction with
         | an AI.
         | 
         | However, if so, this is definitely a distinction that needs to
         | be made far more clearly.
        
           | realcul wrote:
           | Well Microsoft had thier "localhost" AI before CC but that
           | was a ghost without a clear purpose or skill.
        
         | magicalhippo wrote:
         | From what I can gather, llama.cpp supports Anthropic's message
         | format now[1], so you can use it with Claude Code[2].
         | 
         | [1]: https://github.com/ggml-org/llama.cpp/pull/17570
         | 
         | [2]: https://news.ycombinator.com/item?id=44654145
        
         | karpathy wrote:
         | The CC point is more about the data and environmental and
         | general configuration context, not compute and where it happens
         | to run today. The cloud setups are clunky because of context
         | and UIUX user in the loop considerations, not because of
         | compute considerations.
        
           | CamperBob2 wrote:
           | Agree with the GP, though -- you ought to make that clearer.
           | It really reads like you're saying that CC runs locally,
           | which is confusing since you obviously know better.
        
             | karpathy wrote:
             | Yeah, I made some edits to clarify.
        
             | ramoz wrote:
             | I think we need to shift our mindset on what an agent is.
             | The LLM is a brain in a vat connected far away. The agent
             | sits on your device, as a mech suit for that brain, and can
             | pretty much do damn near anything on that machine. It's
             | there, with you. The same way any desktop software is.
        
         | simonw wrote:
         | One of the most interesting coding agents to run locally is
         | actually OpenAI Codex, since it has the ability to run against
         | their gpt-oss models hosted by Ollama.                 codex
         | --oss -m gpt-oss:20b
         | 
         | Or 120b if you can fit the larger model.
        
           | AlexCoventry wrote:
           | What do you find interesting about it, and how does it
           | compare to commercial offerings?
        
             | simonw wrote:
             | It's rare to find a local model that's capable of running
             | tools in a loop well enough to power a coding agent.
             | 
             | I don't think gpt-oss:20b is strong enough to be honest,
             | but 120b can do an OK job.
             | 
             | Nowhere NEAR as good as the big hosted models though.
        
               | ontouchstart wrote:
               | Think of it as the early years of UNIX & PC. Running
               | inferences and tools locally and offline opens doors to
               | new industries. We might not even need client/server
               | paradigm locally. LLM is just a probabilistic library we
               | can call.
        
               | AlexCoventry wrote:
               | Thanks.
        
         | ramoz wrote:
         | What he meant was, agents will probably not be these web
         | abstractions that run in deployed services (langchain, crew);
         | agents meaning the Harnesses (software wrapper) specifically
         | that call the LLM API.
         | 
         | It runs on your computer because of its tooling. It can call
         | Bash. It can literally do anything on the operating system and
         | file system. That's what makes it different. You should think
         | of it like a mech suit. The model is just the brain in a vat
         | connected far away.
        
       | bgwalter wrote:
       | Vibe coding is sufficient for job hoppers who never finish
       | anything and leave when the last 20% have to be figured out. Much
       | easier to promote oneself as an expert and leave the hard parts
       | to other people.
        
         | zingar wrote:
         | I've found incredible productivity gains writing (vibe coding)
         | tools for myself that will never need to be "productionised" or
         | even used by another person. Heck even I will probably never
         | use the latest log retrieval tool, which exists purely for
         | Claude code to invoke it. There is a ton of useful software yet
         | to be written for which there _is_ no "last 20%".
        
           | diamond559 wrote:
           | These tools are so useful and make you so much more
           | "productive" that you don't think anyone else would want to
           | pay anything for them huh? Did your boss at least give you a
           | big raise for your "productivity" increase, or maybe lay off
           | some of your underperforming coworkers bc you are just so
           | much better now?
        
         | simonw wrote:
         | Do you mean vibe coding as-in producing unreviewed code with
         | LLMs and prompting at it until it appears to work, or vibe
         | coding as a catch-all for any time someone uses AI-assistance
         | to help them write code?
        
           | bgwalter wrote:
           | Karpathy uses the term for all of this in the exuberant
           | paragraph 5. of his blog post.
        
         | augment_me wrote:
         | All software is not meant to be open-source, in production and
         | working on 100 platforms.
         | 
         | Sometimes the point of the software is to make an app with 2
         | buttons for your mom to help her do her grocery shopping easier
        
       | delichon wrote:
       | > I like this version of the meme for pointing out that human
       | intelligence is also jagged in its own different way.
       | 
       | The idea of jaggedicity seems useful to advancing epistemology.
       | If we could identify the domains that have useful data that we
       | fail to extract, we could fill those holes and eventually become
       | a general intelligence ourselves. The task may be as hard as
       | making a list of your blind spots. But now we have an alien
       | intelligence with an outside perspective. While making AI less
       | jagged it might return the favor.
       | 
       | If we keep inventing different kinds of intelligence the sum of
       | the splats may eventually become well rounded.
        
         | visarga wrote:
         | I don't think it will become well rounded because that is not
         | cost sensitive. Intelligence is sensitive to cost, it is the
         | core constraint shaping it. Any action has a cost - energy,
         | materials, time, opportunity or social. Intelligence is solving
         | the cost equation, if we can't solve it we die. Cost is also
         | why we specialize, in a group we can offload some intelligence
         | to others. LLMs also have their own costs, and are shaped by it
         | into some kind of jagged intelligence, they are no spherical
         | cows either.
        
       | mvkel wrote:
       | > In this world view, nano banana is a first early hint of what
       | that might look like.
       | 
       | What is he referring to here? Is nano banana not just an image
       | gen model? Is it because it's an LLM-based one, and not
       | diffusion?
        
         | dragonwriter wrote:
         | I think he is referring to capability, not architecture, and
         | say that NB is at the point that it is suggestive of the near-
         | future capability of using GenAI models to create their own UI
         | as needed.
         | 
         | NB (Gemini 2.5 Flash Image) isn't the first major-vendor LLM-
         | based image gen model, after all; GPT Image 1 was first.
        
         | simonw wrote:
         | What's interesting about Nano Banana (and even more so video
         | models like Veo 3) is that they act as a weird kind of world
         | model when you consider that they accept images as input and
         | return images as output.
         | 
         | Give it an image of a maze, it can output that same image with
         | the maze completed (maybe).
         | 
         | There's a fantastic article about that for image-to-video
         | models here: https://video-zero-shot.github.io/
         | 
         | > We demonstrate that Veo 3 can zero-shot solve a broad variety
         | of tasks it wasn't explicitly trained for: segmenting objects,
         | detecting edges, editing images, understanding physical
         | properties, recognizing object affordances, simulating tool
         | use, and much more.
        
       | starchild3001 wrote:
       | The distinction Karpathy draws between "growing animals" and
       | "summoning ghosts" via RLVR is the mental model I didn't know I
       | needed to explain the current state of jagged intelligence. It
       | perfectly articulates why trust in benchmarks is collapsing; we
       | aren't creating generally adaptive survivors, but rather over-
       | optimizing specific pockets of the embedding space against
       | verifiable rewards.
       | 
       | I'm also sold on his take on "vibe coding" leading to ephemeral
       | software; the idea of spinning up a custom, one-off tokenizer or
       | app just to debug a single issue, and then deleting it, feels
       | like a real shift.
        
         | graemefawcett wrote:
         | I've been doing it for months, it's lovely
         | 
         | https://tech.lgbt/@graeme/115749759729642908
         | 
         | It's a stack based on finishing the job Jupyter started. Fences
         | as functions, callable and composable.
         | 
         | Same shape as an MCP. No training required, just walk them
         | through the patterns.
         | 
         | Literally, it's spatially organized. Turns out a woman named
         | Mrs Curwen and I share some thoughts on pedagogy.
         | 
         | There does in fact exist a functor that maps 18th century piano
         | instruction to context engineering. We play with it
        
         | HarHarVeryFunny wrote:
         | > The distinction Karpathy draws between "growing animals" and
         | "summoning ghosts" via RLVR
         | 
         | I don't see these descriptions as very insightful.
         | 
         | The difference between general/animal intelligence and
         | jagged/LLM intelligence is simply that humans/animals really
         | ARE intelligent (the word was created to describe this human
         | capability), while LLMs are just echoing narrow portions of the
         | intelligent output of humans (those portions that are amenable
         | to RLVR capture).
         | 
         | For an artificial intelligence to be intelligent in it's own
         | right, and therefore be generally intelligent, it would need to
         | need - like an animal - to be embodied (even if only
         | virtually), autonomous, predicting the outcomes of it's own
         | actions (not auto-regressively trained), learning incrementally
         | and continually, built with innate traits like curiosity and
         | boredom to put and keep itself in learning situations, etc.
         | 
         | Of course not all animals are generally intelligent - many
         | (insects, fish, reptiles, many birds) just have narrow "hard
         | coded" instinctual behaviors, but others like humans are
         | generalists who evolution have therefore honed for adaptive
         | lifetime learning and general intelligence.
        
         | fourside wrote:
         | > I'm also sold on his take on "vibe coding" leading to
         | ephemeral software; the idea of spinning up a custom, one-off
         | tokenizer or app just to debug a single issue, and then
         | deleting it, feels like a real shift.
         | 
         | We should keep in mind that currently our LLM use is
         | subsidized. When the money dries up and we have to pay the real
         | prices I'll be interested to see if we can still consider
         | whipping up one time apps as basically free
        
       | mips_avatar wrote:
       | I would love Andrej's take on the fast models we got this year.
       | Gemini 3 flash and Grok 4 fast have no business being as good +
       | cheap + fast as they are. For Andrej's prediction about LLMs
       | communicating with us via a visual interface we're going to need
       | fast models, but I feel like AI twitter/HN has mostly ignored
       | these.
        
         | gnerd00 wrote:
         | check out Sasha Luccioni
        
           | mips_avatar wrote:
           | Do you have a link to anything they wrote about this?
        
         | HarHarVeryFunny wrote:
         | Just guessing here, but these small models may well be
         | essentially distillations of larger ones, with this being where
         | their power comes from. e.g. Use a large model to generate
         | synthetic reasoning traces, then train a small model on those.
        
       | jkubicek wrote:
       | > In the same way, LLMs should speak to us in our favored format
       | - in images, infographics, slides, whiteboards,
       | animations/videos, web apps, etc.
       | 
       | You think every Electron app out there re-inventing application
       | UX from scratch is bad, wait until LLMs are generating their own
       | custom UX for every single action for every user for every
       | device. What does command-W do in this app? It's literally
       | impossible to predict, try it and see!
        
         | johnfn wrote:
         | On the other side of the spectrum, I see some of the latest
         | agents, like Codex, take care to get accessibility right --
         | something not even many humans bother to do.
        
           | becquerel wrote:
           | It's an extension of how I've noticed that AIs will generally
           | write very buttoned-down, cross-the-ts-and-dot-the-is code.
           | Everything gets commented, every method has a try-catch with
           | a log statement, every return type is checked, etc. I think
           | it's a consequence of them not feeling fatigue. These things
           | (accessibility included) are all things humans generally know
           | they 'should' do, but there never seems to be enough time in
           | the day; we'll get to it later when we're less tired. But the
           | ghost in the machine doesn't care. It operates at the same
           | level all the time
        
         | Aiisnotabubble wrote:
         | But that's exactly what an LLM solved.
         | 
         | It's the best ui ever.
         | 
         | It understands a lot of languages and abstract concepts.
         | 
         | It will not be necessary at all to let LLM generate random uis.
         | 
         | I'm not a native English speaker. I sometimes just throw in a
         | German word and it just works.
        
         | tim333 wrote:
         | >our favored format - in images, infographics, slides,
         | whiteboards, animations/videos, web apps, etc
         | 
         | If you look at how humans actually communicate I'd guess #1 is
         | text/speech, #2 pictures
        
       | ausbah wrote:
       | tl;dr seems like llms are maturing on the product side and for
       | day-day usage
        
       | socketcluster wrote:
       | For me, Claude Code was the most impressive innovation this year.
       | Cursor was a good proof of concept but Claude Code is the tool
       | that actually got me to use LLMs for coding.
       | 
       | The kind of code that Claude produces looks almost exactly like
       | the code I would write myself. It's like it's reading my mind.
       | This is a game changer because I can maintain the code that
       | Claude produces.
       | 
       | With Claude Code, there are no surprises. I can pretty much guess
       | what its code will look like 90% to 95% of the time but it writes
       | it a lot faster than I could. This is an amazing innovation.
       | 
       | Gemini is quite impressive as well. Nano banana in particular is
       | very useful for graphic design.
       | 
       | I haven't tried Gemini with coding yet but TBH, Claude Code does
       | such a great job; if I could code any faster, I would get
       | decision fatigue. I don't like rushing into architecture or UX
       | decisions. I like to sit on certain decisions for a day or two
       | before starting implementation. Once you start in a particular
       | direction, it's hard to undo and you may try to double down on
       | the mistake due to sunk cost fallacy. I try hard to avoid that.
        
         | tarsinge wrote:
         | I don't have much time to evaluate tools every months and I
         | have settled on Cursor. I'm curious on what I'm missing when
         | using the same models?
        
           | wahnfrieden wrote:
           | If you switch to Codex you will get a lot of tokens for $200,
           | enough to more consistently use high reasoning as well.
           | Cursor is simply far more expensive so you end up using less
           | or using dumber models.
           | 
           | Claude Code is overrated as it uses many of its features and
           | modalities to compensate for model shortcomings that are not
           | as necessary for steering state of the art models like GPT
           | 5.2
        
             | ccmcarey wrote:
             | I disagree, the claude models seem the best at tool
             | calling, opus 4.5 seems the smartest, and claude code (+
             | claude model) seems to make good use of subagents and
             | planning in a way that codex doesn't
        
               | wahnfrieden wrote:
               | Opus 4.5 is so bad at instruction following (30% worse
               | per benchmark shared above) that it requires a manual
               | toggle for plan mode.
               | 
               | GPT 5.2 simply obeys instruction to assemble a plan and
               | avoids the need to compensate for poor steerability that
               | would require the user to manually manage modalities.
               | 
               | Opus has improved though so the plan mode is less
               | necessary than it was before, but it is still far behind
               | state of art steerability.
        
             | MrOrelliOReilly wrote:
             | I think this is a total misunderstanding of Anthropic's
             | place in the AI race. Opus 4.5 is absolutely a state of the
             | art model. I won't knock anyone for preferring Codex, but I
             | think you're ignoring official and unofficial benchmarks.
             | 
             | See: https://artificialanalysis.ai
        
               | woadwarrior01 wrote:
               | > Opus 4.5 is absolutely a state of the art model.
               | 
               | > See: https://artificialanalysis.ai
               | 
               | The field moves fast. Per artificialanalysis, Opus 4.5 is
               | currently behind GPT-5.2 (x-high) and Gemini 3 Pro. Even
               | Google's cheaper Gemini 3 Flash model seems to be
               | slightly ahead of Opus 4.5.
        
               | dr_dshiv wrote:
               | https://lmarena.ai/leaderboard/webdev
               | 
               | LM Arena shows Claude Opus 4.5 on top
        
               | HarHarVeryFunny wrote:
               | I wonder how model competence and/or user preference on
               | web development (that leaderboard) carries over to more
               | complex and larger projects, or more generally anything
               | other than web development ?
               | 
               | In addition to whatever they are exposed to as part of
               | pre-training, it'd be interesting to know what kind of
               | coding tasks these models are being RL-trained for? Are
               | things like web development and maybe Python/ML coding
               | overemphasized, or are they also being trained on things
               | like Linux/Windows/embedded development etc in different
               | languages?
        
               | ramoz wrote:
               | https://x.com/giansegato/status/2002203155262812529/photo
               | /1
               | 
               | https://x.com/METR_Evals/status/2002203627377574113
               | 
               | > Even Google's cheaper Gemini 3 Flash model seems to be
               | slightly ahead of Opus 4.5.
               | 
               | What an insane take for anybody uses these models daily.
        
               | MrOrelliOReilly wrote:
               | Yes, I personally feel that the "official" benchmarks are
               | increasingly diverging from the everyday reality of using
               | these models. My theory is that we are reaching a point
               | where all the models are intelligent enough for day-to-
               | day queries, so points like style/personality and proper
               | use of web queries and other capabilities are better
               | differentiators than intelligence alone.
        
               | MrOrelliOReilly wrote:
               | Totally, however OP's point was that Claude had to
               | compensate for deficiencies versus a state of the art
               | model like ChatGPT 5.2. I don't think that's correct.
               | Whether or not Opus 4.5 is actually #1 on these
               | benchmarks, it is clearly very competitive with the other
               | top-tier models. I didn't take "state of the art" to here
               | narrowly mean #1 on a given benchmark, but rather to mean
               | near or at the frontier of current capabilities.
        
               | gessha wrote:
               | One thing to remember when comparing ML models of any
               | kind is that single value metrics obscure a lot of nuance
               | and you really have to go through the model results one
               | by one to see how it performs. This is true for vision,
               | NLP, and other modalities.
        
               | fzzzy wrote:
               | is x-high fast enough to use as a coding agent?
        
               | wahnfrieden wrote:
               | Yes, if you parallelize your work, which you must learn
               | to do if you want the best quality
        
               | wahnfrieden wrote:
               | What am I missing? As suspicious as benchmarks are, your
               | link shows GPT 5.2 to be superior.
               | 
               | It is also out of date as it does not include 5.2 Codex.
               | 
               | Per my point about steerability compensated for by
               | modalities and other harness features: Opus 4.5 scores
               | 58% while GPT 5.2 scores 75% for the instruction
               | following benchmark in your link! Thanks for the hard
               | evidence - GPT 5.2 is 30% ahead of Opus 4.5 there. No
               | wonder Claude Code needs those harness features for the
               | user to manually reign in control over its instruction
               | following capability.
        
           | andai wrote:
           | I have only compared Claude Code with Crush and a tool of my
           | own design. In my experience, Claude code is optimized for
           | giant codebases and long tasks. It loves launching dozens of
           | agents in parallel. So it's a bit heavy for smaller, surgical
           | stuff, though it works decent for that too.
           | 
           | If you mostly have small codebases that fit in context, or
           | make many small changes interactively, it's not really great
           | for that (though it can handle it too). It'll just be
           | spending most of its time poking around the codebase, when
           | the whole thing should have just been loaded... (Too bad
           | there's no small repo mode. I made startup hook that just
           | dumps cat dir into context, but yeah, should be a toggle.)
        
           | afro88 wrote:
           | You're not missing much. You can generally use Cursor like
           | Claude Code for normal day to day use. I prefer Cursor
           | because I like reviewing changes in an IDE, and I like being
           | able to switch to the current SOTA model.
           | 
           | Though for more automated work, one thing you miss with
           | Cursor is sub agents. And then to a lesser extent skills
           | (these are pretty easy to emulate in other tools). I'm sure
           | it's only a matter of time though.
        
             | Ozzie_osman wrote:
             | Claude Code's VS Code integration is very easy to set up
             | and pretty helpful if you want to see/review changes in an
             | IDE.
        
               | ollysb wrote:
               | The big limitation is that you have to approve/disapprove
               | at every step. With Cursor you can iterate on changes and
               | it updates the diffs until you approve the whole batch.
        
               | fzzzy wrote:
               | There is an auto accept diffs mode
        
           | ramoz wrote:
           | You are missing an entire agentic experience. And I wouldn't
           | call it vibe coding for an engineer; you're more or less
           | empowered to truly orchestrate the development of your
           | system.
           | 
           | Cursor has agent, but that's like whoever else tried to copy
           | the Model T while Ford was developing it.
        
             | senordevnyc wrote:
             | This hasn't been my experience at all. I'm finding Cursor
             | with Opus 4.5 and plan mode to be just as capable as CC.
             | And I prefer the UI/UX.
        
         | andai wrote:
         | I first got into agentic properly with GLM coding plan (it's
         | like _$2 /month),_ but I found myself very consistently asking
         | Claude to make the code more elegant and readable. At which
         | point I realized I was being silly and just switched to Claude
         | code.
         | 
         | (GLM etc. get surprisingly close with good prompting but...
         | $0.60/day to not worry about that is a no brainer.)
        
         | Daniel_sk wrote:
         | I don't even see much reason to use Cursor. I am used to
         | IntelliJ IDEA, so I just downloaded the Claude Code plugin and
         | basically now I use the IDE only for navigating in the code,
         | finding references and reviewing the code. I can't even
         | remember the last time I wrote more than 2 lines of code.
         | Claude Code has catapulted my performance at least 5x if not
         | more. And now that the cost of writing test is so minimal I am
         | also able to achieve much better (and meaningful!) test
         | coverage too. The AI agents is where the most productivity is.
         | I just create a plan with Claude, iterate over, ask questions,
         | then let it implement the plan, review, ask to do some
         | adjustments. No manual writing of code at all. Zero.
        
           | esafak wrote:
           | IntelliJ has its own Claude integration too, but it does not
           | use your Claude subscription:
           | https://blog.jetbrains.com/ai/2025/09/introducing-claude-
           | age...
        
           | RealityVoid wrote:
           | Maybe I'm holding it wrong, but the finer aspects of a
           | codebase it still messes up. If I ask it to implement some
           | weird thing off the beaten path it gets lost. But I
           | completely agree at the test part. I actually test much more
           | now since it's so easy!
        
         | spaceman_2020 wrote:
         | Nano Banana Pro is legitimately an insane tool if you know how
         | to use it. I still can't believe they released it in the wild
        
           | IAmGraydon wrote:
           | It's decent for things that would take a long time in
           | Photoshop. Like most AI, sometimes it works great and
           | sometimes it goes off the rails completely. Most recently, I
           | used it to process some drone photos that were taken during
           | late fall for the purpose of marketing a commercial property.
           | All of the trees/grass/plants were brown, so I told it to
           | make it look like the photos were taken during the summer but
           | not to change anything else. It did a very good job, not just
           | changing the color, but actually adding leaves to the plants
           | and trees in a way that looked very realistic. It did in
           | seconds what would have taken one of my team members hours,
           | leaving them to work on other more pressing projects.
        
           | rolymath wrote:
           | What is there to using it more than asking it to generate an
           | image of something?
        
             | kakapo5672 wrote:
             | For one: modifying existing images in interesting ways ...
             | adding characters, removing elements, altering or enhancing
             | certain features, creating layers, and so on. Things that
             | would take a while on Photoshop, done almost instantly.
             | Really unlocks the imagination.
        
             | dgacmu wrote:
             | For me: I've only tried using it seriously a few times but
             | my experience is that you have to juggle carefully when to
             | start a fresh session. It can get really anchored on
             | earlier versions of images. It was interesting balancing
             | iteration and from-scratch prompt refinement.
        
             | disease wrote:
             | I gave it an image of my crappy art and asked what steps I
             | could take to make it look better. It gave me specific
             | advice like varying the line widths and how to use this on
             | specific parts of the character. It also pointed out that
             | the shading in my piece was inconsistent and did not
             | reflect the 3d form I was representing and again gave me
             | specific fixes I could implement. I asked for it to give me
             | an updated version of the piece with all of its advice
             | implemented and it did so. I was pretty shocked at all of
             | this.
        
         | yread wrote:
         | Do you guys all work 100% on open source? Or are you uploading
         | bits of your copyrighted code for future training to Anthropic?
         | I hate patents so copyright is the only IP protection I have.
        
           | dukeyukey wrote:
           | We use AWS Bedrock, so everything stays within our AWS
           | account. It's not like we aren't already uploading our code
           | to GitHub for version control, AWS for deployment, Jetbrains
           | for development, all of ours logs to Datadog, Sentry,
           | Snowflake, and more.
        
             | yread wrote:
             | Yeah, my source code is on my computers, in self-hosted
             | version control and self-hosted CI runners
        
         | thefourthchime wrote:
         | I've used all of these tools and for me Cursor works just as
         | well but has tabs, easy ways to abort or edit prompts, great
         | visual diff, etc...
         | 
         | Someone sell me on how Claude Code, I just don't get it.
        
           | senordevnyc wrote:
           | I'm with you, I've used CC but I strongly prefer Cursor.
           | 
           | Fundamentally, I don't like having my agent and my IDE be
           | split. Yes, I know there are CC plugins for IDEs, but you
           | don't get the same level of tight integration.
        
       | augment_me wrote:
       | I noticed that despite really liking Karpathy and the blog, I was
       | am kind of wincing/involuntarily reacting to the LLM-like "It's
       | not X, its Y"-phrases:
       | 
       | > it's not just a website you go to like Google, it's a little
       | spirit/ghost that "lives" on your computer
       | 
       | > it's not just about the image generation itself, it's about the
       | joint capability coming from text generation
       | 
       | There would be no reaction from me on this 3 years ago, but now
       | this sentence structure is ruined for me
        
         | huevosabio wrote:
         | Same, I cringe when I read this structure.
        
         | d-lisp wrote:
         | I hated these sentences way before LLMs, at least in the
         | context of an explanation.
         | 
         | > it's not just a website you go like Google, it's a little
         | spirit/ghost that "lives" on your computer
         | 
         | This type of sentence, I call rhetorical fat. Get rid of this
         | fat and you obtain a boring sentence that repeats what has been
         | said in the previous one.
         | 
         | Not all rhetorical fats are equal, and I must admit I find
         | myself eyerolling on the "little spirit" part more than about
         | the fatness.
         | 
         | I understand the author wants to decorate things and emphasize
         | key elements, and the hate I feel is only caused by the
         | incompatible projection of my ideals to a text that doesn't
         | belong to me.
         | 
         | > it's not just about the image generation itself, it's about
         | the joint capability coming from text generation.
         | 
         | That's unjustified conceptual stress.
         | 
         | That could be a legitimate answer to a question ("No, no, it's
         | not just about that, it's more about this"), but it's a text.
         | Maybe the text wants you to be focused, maybe the text wants to
         | hype you; this is the shape of the hype without the hype.
         | 
         | "I find image generation is cooler when paired with text
         | generation."
        
           | killerstorm wrote:
           | It is not a decoration. Karpathy juxtaposes ChatGPT (which
           | feels like a "better google" to most people) to Claude Code,
           | which, apparently, feels different to him. It's a comparison
           | between the two.
           | 
           | You might find this statement non-informative, but without
           | two parts there's no comparison. That's really the semantics
           | of the statement which Karpathy is trying to express.
           | 
           | ChatGPT-ish "it's not just" is annoying because the first
           | part is usually a strawman, something reader considers trite.
           | But it's not the case here.
        
             | d-lisp wrote:
             | Indeed, I was probably grumpy at the time I wrote the
             | comment. I do find some truth in it still.
             | 
             | You're right ! The strawman theory is based.
             | 
             | But I think there's more to it, I find dislikable the
             | structure of these sentences (which I find a bit
             | sensationnalist for nothing, I don't know, maybe I am still
             | grumpy).
        
               | killerstorm wrote:
               | Well, language is a subject to 'fashion' one-upmanship
               | game: people want to demonstrate their sophistication,
               | often by copying some "cool" patterns, but then over-used
               | patterns become "uncool" cliche.
               | 
               | So it might be just a natural reaction to over-use of a
               | particular pattern. This kind of stuff have been driving
               | language evolution for millennia. Besides that, pompous
               | style is often used in 'copy' (slogans and ads) which is
               | something most people don't like.
        
           | amelius wrote:
           | Karpathy should go back to what he does best: educating
           | people about AI on a deep level. Running experiments and
           | sharing how they work, that sort of stuff. It seems lately he
           | is closer to an influencer who reviews AI-based products.
           | Hopefully it is not too late to go back.
        
             | flakiness wrote:
             | I feel these review stuff is more like a side / pass time
             | to him. Look at nanochat for example. My impression is that
             | these are the thongs he spends most of his energy still.
             | 
             | After all,l he's been a "influencer" for a long time,
             | starting from the "software 2.0" essay.
        
         | another_twist wrote:
         | Same here, had to configure ChatGPT to stop making these
         | statements. Also had to configure bunch of other stuff to make
         | it bland when answering questions.
        
           | andai wrote:
           | The way to make AI not sound like ChatGPT is to use Claude.
           | 
           | I realized that's what bothered me. It's not "oh my god, they
           | used ChatGPT." But "oh my god, they couldn't even be bothered
           | to use Claude."
           | 
           | It'll still sound like AI, but 90% of the cringe is gone.
           | 
           | If you're going to use AI for writing, it's just basic
           | decency to use the one that isn't going to make your audience
           | fly into a fit of rage every ten seconds.
           | 
           | That being said, I feel very self conscious using emdashes in
           | current decade ;)
        
             | ionwake wrote:
             | I dont think Ive ever noticed someone use an emdash until
             | chatgpt appeared
        
               | andai wrote:
               | https://xkcd.com/3126/
               | 
               | I mostly use them in Telegram because it auto converts --
               | into emdash. They are a pain to type everywhere else
               | though!
        
             | dr_dshiv wrote:
             | I love em dashes--they basically indicate a more deliberate
             | pause than a ... without the tight vibes of a semicolon.
        
             | another_twist wrote:
             | I don't use LLM for writing just factual research stuff.
             | And this would happen even in those questions.
        
             | sungho_ wrote:
             | If a reader gets angry simply because the author used
             | ChatGPT instead of Claude, then the reader is an idiot.
        
         | nathias wrote:
         | It's not text - it's clickbait distillied to grammar.
        
         | yard2010 wrote:
         | I cannot unsee this anymore and it ruins the whole internet
         | experience for me
        
         | matsemann wrote:
         | Yeah, came to read Karpathy's thoughts, but might as well ask
         | an LLM myself..
        
         | spaceman_2020 wrote:
         | I used to use a lot of em dashes normally in my writing - they
         | were my go-to replacements for commas and semicolons
         | 
         | But I had to change how I write because people started calling
         | my writing "AI generated"
        
           | athrowaway3z wrote:
           | 2026 will be the year of the ;
        
             | vatsachak wrote:
             | Please no that's my go to
        
           | fzzzy wrote:
           | so you switched to using hyphens instead?
        
             | nicwolff wrote:
             | En dashes!
        
         | karpathy wrote:
         | You're absolutely right!
         | 
         | Jk jk, now that you pointed it out I can't unsee it.
        
         | kakapo5672 wrote:
         | Very broadly, AI sentence-structure and word choice is
         | recursing back into society, changing how humans use language.
         | The Economist recently had a piece on word usage of British
         | Parliament members. They are adopting words and phrases
         | commonly seen in AI.
         | 
         | We're embarking on a ginormous planetary experiment here.
        
       | nkko wrote:
       | Beyond graduating students, I see model labs as
       | "accelerators/incubators" bundling, launching, and productizing
       | observed ideas that gain traction. The sheer strength of their
       | platforms, the number of eyes watching them, near-zero marginal
       | costs, and seemingly unlimited budgets mean that only slow
       | decision-making can prevent them from becoming the next Amazons
       | of everything.
        
       | alexgotoi wrote:
       | LLMs still need to bring clear added value to enterprise and
       | corporate work; otherwise, they remain a geek's toy.
       | 
       | Big media agencies that claim to use AI rely on strong creative
       | teams who fine-tune prompts and spend weeks doing so. Even then,
       | they don't fully trust AI to slice long videos into shorter clips
       | for social media.
       | 
       | Heavy administrative functions like HR or Finance still don't get
       | approval to expose any of their data to LLMs.
       | 
       | What I'm trying to say is that we are still in the early stages
       | of LLM development, and as promising as this looks, it's still
       | far from delivering the real value that is often claimed.
        
         | gessha wrote:
         | I think their non-deterministic nature is what's making it
         | difficult to adopt. It's hard to train somebody in the old way
         | of "if you see this, do this" because when you call the LLM
         | twice you most likely get different results.
         | 
         | It took a long time to computerize businesses and it might take
         | some time to adopt/adapt to LLMs.
        
       | dandelionv1bes wrote:
       | Something I've been thinking about is how as end stage users (eg
       | building our own "thing" on top of an LLM) we can broadly verify
       | it's doing what we need without benchmarks. Does a set of custom
       | evals built out over time solve this? Is there more we can do?
        
       | andai wrote:
       | The bit about o3 being the turning point is very interesting. I
       | heard someone say that o3 (or perhaps the cheaper o4-mini) should
       | have been called gpt-5, and that people would have been mind
       | blown. Instead it kind of went under the radar as far as the
       | mainstream goes.
       | 
       | Whereas we just got the incremental progress with gpt-5 instead
       | and it was very underwhelming. (Plus like 5 other issues at
       | launch, but that's a separate story ;)
       | 
       | I'm not sure if o4-mini would have made a good default gpt
       | though. (Most use is conversational and its language is very
       | awkward.) So they could have just called it gpt-5 pro or
       | something, and put it on the $20 tier. I don't know.
        
         | karpathy wrote:
         | I agree with this fwiw, for many months I talked to people who
         | never used o3 and didn't know what it was because it sounded
         | weird. Maybe it wasn't obvious at the time but that was a good
         | major point release to make then.
        
       | andai wrote:
       | Here's the source for the jagged spiky intelligence diagram:
       | 
       | https://x.com/colin_fraser/status/1994235521812328695
       | 
       | https://karpathy.bearblog.dev/the-space-of-minds/
        
       | distalx wrote:
       | Friendly reminder: There is no ghost in the machine. It is a
       | system executing code, not a being having thoughts. Let's admire
       | the tool without projecting a personality onto it.
        
         | ngruhn wrote:
         | Consciousness is weird and nobody understands it. There is no
         | good reason to assume that these systems have it. But there is
         | also no good reason to rule it out.
        
         | squidbeak wrote:
         | You sound as if you have grounds for certainty about this. What
         | are they?
        
         | dr_dshiv wrote:
         | That's the old way of thinking about it. there is a new way.
        
         | skybrian wrote:
         | For me, that's kind of the point. It's similar to how the
         | characters in a novel don't really exist, and yet you can't
         | really discuss what happens in a novel without pretending that
         | they do. It doesn't really make sense to treat the author's
         | motivations and each character's motivations as the same.
         | 
         | Similarly, we're all talking to ghosts now, which aren't real,
         | and yet there is something there that we can talk about. There
         | are obvious behavioral differences depending on what persona
         | the LLM is generating text for.
         | 
         | I also like the hint of danger in "talking to ghosts." It's
         | difficult to see how a rational adult could be in any danger
         | from just talking, but I believe the news reports that some
         | people who get too deep into it get "possessed."
        
       | lysecret wrote:
       | It's funny how every podcaster/public ai figure is so certain
       | text as a Ui will go away and it's not going anywhere.
        
         | tim333 wrote:
         | It's probably increased during my lifetime. People used to
         | talk, now they sit and text into smartphones.
        
           | gessha wrote:
           | There might be some confusion about the transition to what
           | some call post-literate era: era where text is not the
           | primary medium. That's not necessarily bad because you get
           | the advantages of other mediums - oral and visual but it is
           | something to keep in mind.
        
             | tim333 wrote:
             | I'm bit skeptical that a post-literate era is happening. I
             | gather it appears in some sci-fi but I don't see much sign
             | in reality. I mean here we are on a text only site. If
             | anything we seem to be heading for a 100% literate society.
             | Literacy graphs here:
             | https://ourworldindata.org/grapher/cross-country-literacy-
             | ra...
        
               | gessha wrote:
               | I don't think the post-illiterate era means that text
               | will disappear. I think it's just not going to be
               | dominant anymore but I also have my reservations since I
               | do prefer the text medium.
        
         | devalexwells wrote:
         | A few days ago I was trying to unsubscribe to a service
         | (notably an AI 3D modeling tool that I was curious about).
         | 
         | I spent 5 minutes trying to find a way to unsubscribe and
         | couldn't. Finally, I found it buried in the plan page as one of
         | those low-contrast ellipses on the plan card.
         | 
         | Instead of unsubscribing me or taking me to a form, it opened a
         | convos with an AI chatbot with a preconfigured "unsubscribe"
         | prompt. I have never felt more angry with a UI that I had to
         | waste more time talking to a robot before it would render the
         | unsubscribe button in the chat.
         | 
         | Why would we bring the most hated feature of automated phone
         | calls to apps? As a frontend engineer I am horrified by these
         | trends.
        
       | metalman wrote:
       | find on page:slop=0
        
       | cheesecompiler wrote:
       | Excellent more grounded review. A few questions:
       | 
       | > LLMs are emerging as a new kind of intelligence, simultaneously
       | a lot smarter than I expected and a lot dumber than I expected
       | 
       | Isn't this concerning? How can we know which one we get? In the
       | realm of code it's easier to tell when mistakes are being made.
       | 
       | > regular people benefit a lot more from LLMs compared to
       | professionals, corporations and governments
       | 
       | We thought this would happen with things like AppleScript, VB,
       | visual programming. But instead, AI is currently used as a
       | smarter search engine. The issue is that's also the area where it
       | hallucinates the most. What do you think is the solution?
        
       | sireat wrote:
       | What is current state of the art workflow when working with
       | legacy code across multiple languages?
       | 
       | This would be a 100 kLOC legacy project written in C++, Python,
       | and jQuery era Javascript circa 2010. Original devs have long
       | left. I would rather avoid C++ as much as possible.
       | 
       | I've been Github Copilot (in VS Code) user since June of 2021 and
       | still use it heavily, but the "more powerful intellisence"
       | approach is limiting me on legacy projects.
       | 
       | Presumably I need to provide more context on larger projects.
       | 
       | I can get pretty far with just ChatGPT plus and feeding bits and
       | pieces of project. However that seems like using the wrong tool.
       | 
       | Codex seems better for building things but not sure about
       | grokking existing things.
       | 
       | Would Cursor be more suitable for just dumping the whole project
       | (all languages) basically 4 different sub projects and then
       | selectively activating what to include in queries?
        
         | sandos wrote:
         | I dont understand, the agent mode of copilot will search for
         | and be pretty good and filling its own context afaik. I never
         | really feed any of our 100k+ lines legacy codebase explicitly
         | to the LLM.
        
       ___________________________________________________________________
       (page generated 2025-12-20 23:01 UTC)