vladzely.zip

"Design isn't waterfall, or the double diamond, or "iterate and test", or coding or comping or any of the dozens of processes ... Design is at its core the *intention* to improve material conditions under conditions of uncertainty based on understanding as much as you can about the context. Sometimes this is a competitive advantage, and sometimes it's not. (This is where so much "UX" talk goes wrong.)"

Erika Hall
src

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.

…
Src

“Design leadership, at its highest level, is not about managing complexity. It’s about deciding where complexity belongs, and where it doesn’t.”

Src

FS


Charlie Munger once asked me: ‘How can someone give away fifty percent of profits and make billions more than if he’d kept it all?’ Before I could answer, he told me about Les Schwab, a tire shop owner who understood incentives better than almost anyone.

What Schwab discovered will change how you think about business and life.

Here are a few of his lessons:

1. Win Win, The Math of Generosity:Les discovered that splitting profits 50/50 with store managers didn’t cut his wealth in half, it multiplied it. His reasoning was pure math: “If I share half the profits, I still have half. And if Frank makes more money, he’ll work harder to make the store successful. If the store is more successful, my half is worth more than my whole used to be.” You get rich by making others rich.

2. All-In or All-Out: At 34, Les sold his house, borrowed against his life insurance, and scraped together $11,000 to buy a failing tire shop with no running water. He’d never changed a tire. His competitors had decades of experience. But Les had something they didn’t: no backup plan. That total commitment forced him to figure it out. One year later, he’d quintupled revenue. Half-measures guarantee half-results.

3. High Agency: Everything is your job. Les bought his first tire shop having never fixed a flat in his life. On day one, a customer needs tires mounted. Les fumbles with hand tools on the cold concrete, making a complete mess until his employee arrives. He insisted on being taught, so the situation never repeated. Sometimes, the only qualification you need is the willingness to figure it out.

4. Go Positive, Go First: Les instituted free flat repairs for anyone, customer or not. Competitors called him crazy. Why fix flats for people who bought tires elsewhere? But Les understood reciprocity: humans are biologically wired to return favors, even those that are unearned. Those free repairs created a loop, doing more marketing than marketing could ever do. Most businesses wait for the transaction before the service. Consistently going positive and going first is the most powerful force in the universe.

5. Dark Hours: Every morning before dawn, teenage Les ran his paper route. Not biked, ran. For two months, he sprinted through dark streets on foot, saving enough to buy a bicycle. While his classmates slept, he earned. By senior year, Les owned all nine routes in town. When your competition sleeps, you can build your lead.

The three elements of human value

In what ways do we contribute to the world?

Ivan posits an elegant framework to the above question that breaks down value into three components:

Our capabilities — these are the skills and knowledge that we possess. Can you build a fire? Calculate the trajectory of a rocket? Craft a website? Determine who stole the cookie among a group of kindergartners?

Our taste — these are our values and preferences. What draws you more: Art Nouveau's organic curves or Bauhaus's clean functionality? Are you moved more by a single person's suffering right in front of you, or by the possible harm that might affect thousands of future generations? Does your focus lean toward an impressive end, or an elegant means?

Our agency — this consists of our will and drive. Even if we know how to do something, will we be sufficiently motivated to do it? Do we care enough to be moved to action? These three elements work together to determine how we actually contribute: our capabilities define what we can do, our taste guides what we want to do, and our agency determines what we will do.


—
JULIE ZHUO