Arseniy Zarechnev

Programmer and entrepreneur.

[Arseniy Zarechnev @evindor]

@AskYoshik @zerohedge > This is the most expensive science experiment in history, funded by your SaaS subscriptions.

Not the worst thing to spend that money on?

[Arseniy Zarechnev @evindor]

@seraleev There's a saying "wanna bathe in holy water? do the dishes"

[Arseniy Zarechnev @evindor]

@nvk I saw this hack: CLAUDE.md: @AGENTS.md

[Arseniy Zarechnev @evindor]

@8teAPi It's magical when I got my first one. Spent over an hour crafting the prompt, sent opus for overnight research (small transformer training). Was still running 10 hours later, with research milestones documented in a journal, code snapshot every step, real progress made. Insane.

[Arseniy Zarechnev @evindor]

My opus agents last week be like

[Arseniy Zarechnev @evindor]

@GregorySchier They are silently optimizing and quantizing. Last week Opus was brilliant, this week all I hear from it is that I'm absolutely right

[Arseniy Zarechnev @evindor]

@bcherny please give us back smart opus. I've been absolutely right for a week now and its driving me nuts.

[Arseniy Zarechnev @evindor]

The more unique entropy you stuff into the context before your request, the more unique and creative outputs you will get.

Models are like hash functions.

f(simple prompt) -> same output
f(your degen mind snapshot, simple prompt) -> unique output

[Arseniy Zarechnev @evindor]

My answer: prime your models with your taste and preferences to get unique outputs.

"Write a poem about time" -> identical slop

"values.md, journal.md, life_of_jdilla.md
What if time is not a 1-dimensional arrow, but 3d like space?
Write a banger poem about time" -> dope stuff

[Alex Veremeyenko @alex_verem]

🚨 BREAKING: Researchers at UW Allen School and Stanford just ran the largest study ever on AI creative diversity.

70+ AI models were given the same open-ended questions. They all gave the same answers.

They asked over 70 different LLMs the exact same open-ended questions.

"Write a poem about time." "Suggest startup ideas." "Give me life advice."

Questions where there is no single right answer. Questions where 10 different humans would give you 10 completely different responses.

Instead, 70+ models from every major AI company converged on almost identical outputs. Different architectures. Different training data. Different companies. Same ideas. Same structures. Same metaphors.

They named this phenomenon the "Artificial Hivemind." And the paper won the NeurIPS 2025 Best Paper Award, which is the highest recognition in AI research, handed to a small number of papers out of thousands of submissions.

This is not a blog post or a hot take. This is award-winning, peer-reviewed science confirming something massive is broken.

The team built a dataset called Infinity-Chat with 26,000 real-world, open-ended queries and over 31,000 human preference annotations. Not toy benchmarks. Not math problems.

Real questions people actually ask chatbots every single day, organized into 6 categories and 17 subcategories covering creative writing, brainstorming, speculative scenarios, and more.

They ran all of these across 70+ open and closed-source models and measured the diversity of what came back. Two findings hit hard.

First, intra-model repetition. Ask the same model the same open-ended question five times and you get almost the same answer five times.

The "creativity" you think you're getting is the same output wearing a slightly different outfit. You ask ChatGPT, Claude, or Gemini to write you a poem about time and you keep getting the same river metaphor, the same hourglass imagery, the same reflection on mortality.

Over and over. The model isn't thinking. It's defaulting to whatever scored highest during alignment training.

Second, and this is the one that should really alarm you, inter-model homogeneity. Ask GPT, Claude, Gemini, DeepSeek, Qwen, Llama, and dozens of other models the same creative question, and they all converge on strikingly similar responses.

These are models built by completely different companies with different architectures and different training pipelines.

They should be producing wildly different outputs. They're not. 70+ models all thinking inside the same invisible box, producing the same safe, consensus-approved content that blends together into one indistinguishable voice.

So why is this happening? The researchers point directly at RLHF and current alignment techniques. The process we use to make AI "helpful and harmless" is also making it generic and boring.

When every model gets trained to optimize for human preference scores, and those preference datasets converge on a narrow definition of what "good" looks like, every model learns to produce the same safe, agreeable output. The weird answers get penalized.

The original takes get shaved off. The genuinely creative responses get killed during training because they didn't match what the average annotator rated highly. And it gets even worse.

The study found that reward models and LLM-as-judge systems are actively miscalibrated when evaluating diverse outputs. When a response is genuinely different from the mainstream but still high quality, these automated systems rate it LOWER. The very tools we built to evaluate AI quality are punishing originality and rewarding sameness.

Think about what this means if you use AI for brainstorming, content creation, business strategy, or literally any task where you need multiple perspectives. You're getting the illusion of diversity, not the real thing.

You ask for 10 startup ideas and you get 10 variations of the same 3 ideas the model learned were "safe" during training. You ask for creative writing and you get the same therapeutic, perfectly balanced, utterly forgettable tone that every other model gives.

The researchers flagged direct implications for AI in science, medicine, education, and decision support, all domains where diverse reasoning is not a nice-to-have but a requirement.

Correlated errors across models means if one AI gets something wrong, they might ALL get it wrong the same way. Shared blind spots at massive scale.

And the long-term risk is even scarier. If billions of people interact with AI systems that all think identically, and those interactions shape how people write, brainstorm, and make decisions every day, we risk a slow, invisible homogenization of human thought itself. Not because AI replaced creativity.

Because it quietly narrowed what we were exposed to until we all started thinking the same way too.

Here's what you can actually do about it right now:
→ Stop accepting first-draft AI output as creative or diverse. If you need 10 ideas, generate 30 and throw away the obvious ones
→ Use temperature and sampling parameters aggressively to push models out of their comfort zone
→ Cross-reference multiple models AND multiple prompting strategies, because same model with different prompts often beats different models with the same prompt
→ Add constraints that force novelty like "give me ideas that a traditional investor would hate" instead of "give me creative ideas"
→ Use structured prompting techniques like Verbalized Sampling to force the model to explore low-probability outputs instead of defaulting to consensus
→ Layer your own taste and judgment on top of everything AI gives you. The model gets you raw material. Your weirdness and experience make it original

This paper puts hard data behind something a lot of us have been feeling for a while. AI is getting more capable and more homogeneous at the same time.

The models are smarter, but they're all smart in the exact same way. The Artificial Hivemind is not a bug in one model. It's a systemic feature of how the entire industry builds, aligns, and evaluates language models right now.

The fix requires rethinking alignment itself, moving toward what the researchers call "pluralistic alignment" where models get rewarded for producing diverse distributions of valid answers instead of collapsing to a single consensus mode.

Until that happens, your best defense is awareness and better prompting.

[Arseniy Zarechnev @evindor]

@camsoft2000 funny how I get completely the opposite experience, been maxing out my claude sub and underutilizing codex. smooth sailing with opus, but codex (5.3/5.4) overengineers and misses the mark

[Arseniy Zarechnev @evindor]

After 10+ years of mac, running omarchy is so refreshing.

hyprland scrolling layout is awesome
everything is easily customizeable
claude makes me custom gtk apps and overlays which look nice

love it

[Arseniy Zarechnev @evindor]

you have to live through your AI psychosis and emerge on the other side. i was in a similar state for months, finally said fuck it and am now happily creating stuff i'd never bothered/was able to do before

[Mo @atmoio]

I was a 10x engineer. Now I'm useless.

[Arseniy Zarechnev @evindor]

@phuctm97 Recreate it from scratch.
1. opus: research this messy code, write a paper on details of implementation, discard this, keep that
2. opus: take the spec and implement

worked for me to refactor 20k+ loc messy research project into 2k lines prod-ready

[Arseniy Zarechnev @evindor]

kinda sad linux on desktop finally became good, but just before an incoming tectonic shift in how we work with computers, which will make desktops in general obsolete. but we can have a couple good years I think!

[Arseniy Zarechnev @evindor]

Configuring hyprland/omarchy using claude code (ofc skip permissions) is next level. It just does things

[Arseniy Zarechnev @evindor]

@TomBukic For me carry-mix helped a lot (sometimes .3, sometimes .8), and when I tried to abandon digit curriculum (1-3, then 1-6, then 1-10) models were stuck at ~0.2 random tok_acc for 50k+ steps. Somehow that helps

[Arseniy Zarechnev @evindor]

@TomBukic With 57p its so small we were able to enumerate all remaining changes, none of them worked, so claude is now running huge seed sweeps with instrumented training, trying to figure out why the model needs those params that converge to near zero

[Arseniy Zarechnev @evindor]

@TomBukic Thanks! Will you try to push your architecture more? We found good insight via "structural analysis" - Claude writes a bunch of python to analyze trained weights and find similarities, regularities or potential zeros/reductions.

[Arseniy Zarechnev @evindor]

@martinlasek There is already a lab doing that https://t.co/zKhv9T1OVP

[Arseniy Zarechnev @evindor]

@martinlasek I think that would change when someone makes a chip implementing a huge foundational model in silicon. It would outperform generic GPUs 100x, provide real value and also a good reason to sell you a new device every year