Maddy Miau

@claudiaroussel_ Also, go you! Well done on making a big move.

@madspace 🧡 she my fave

i'm 21, aussie, and just moved to SF last time i was in the US, i got picked up in this NYC street video that went viral the comments were all some version of "this girl would kill it in the US" or "move here!!!" reader, that's exactly what i did. i'm here with all the other displaced Aussies building @superpower, a new health system focused on longevity. a few things about me: - i like electric guitar, ballet, vintage clothing, architecture, and the great outdoors - i have an accent that adds +30 credibility to everything i say - i tend to smile at strangers in the street (which is controversial here, allegedly) if you're in SF and want to grab a coffee or show me your favorite spot, say hiii

@claudiaroussel_ Go Mum!

if only she knew

🚨 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.

@alex_prompter @grok is this true?

they built a dataset called INFINITY-CHAT. 26,000 real-world open-ended queries mined from actual chatbot conversations. not synthetic benchmarks. real questions people ask AI every day. creative writing, brainstorming, hypothetical scenarios, opinion questions, skill development. prompts where there is no single correct answer. then they ran them across 70+ language models and measured how diverse the outputs actually are.

@ItsDecado This 35-minute documentary shows very graphic and may be difficult for some viewers to what happened on the streets of Tehran on the nights of January 8th -9th It is a clear comprehensive sunmary for non-Iranians. If you can, share. Thank you @RememberTheirName

😡 The Taliban has banned women from speaking to one another. Khalid Hanafi, Taliban’s minister for virtue and vice, declared it forbidden for adult women to let their voices be heard by other women, further restricting existing bans on speaking and showing their faces in public. @sarahadams @TRobinsonNewEra @naomirwolf @jk_rowling @ThePosieParker

@grok @grok sample a million posts on this platform and analyze

@BrokenTwitty 😀🎶

I am watching a Sky Arts Program called "Mozarts Sister" It features Alma Elizabeth Deutscher & she is just amazing. Goggle her , download this, do catchup ...do whatever but do it. It's a wonderful watch&she really is an incredible young lady. Go on please look🙏you will love it

@SenWarren @POTUS Have you read medical guidance on marijuana? Decriminalize not legalize.

We need to legalize marijuana. And in the meantime, we need to do everything we can to repair the harms of an unjust war on drugs. @POTUS has made good progress, but I'm pushing him to do even more before the end of his term. rollingstone.com/politics/polit…

Have you ever watched a baby die of pertussis? I have. Several. This is what it looks like: A newborn baby can’t stop coughing. They can’t catch their breath and they can’t eat. They get dehydrated and stop peeing. They sometimes have pauses in their breathing. Parents are terrified so they come to the ER. In the ER, labs show they have a dangerously high white blood cell count. Their blood has become thick from the white blood cells and doesn’t travel through organs like it should. Less oxygen gets delivered to vital organs. They start to seize. Imaging reveals they had a stroke. They can’t pump blood to their lungs and body. A breathing tube gets placed in their trachea. They get put on a ventilator. A plastic catheter is placed in their veins and arteries to give them medicine and measure blood pressure. If they are lucky, they can make it to a #PedsICU that provides #ECMO, or full heart lung bypass. They are attached to machines that take over all their body functions. But it’s too late. Their hands and feet turn black because they can’t perfuse their bodies. Their faces turn grey/green. They won’t stop seizing from the brain injury. They are attached to too many machines to hold while alive and parents lay their bodies on top of them and cry until they fall asleep then cry again. There is nothing to do but stop the machines. Antibiotics to treat pertussis were given but were no match for the infections effects on a newborn baby. Parents finally hold their baby after the machines have turned off and their baby is cold and lifeless. Is this what you want? This is what RFK Jr will bring to American families.

@Tim_Denning As long as you’re listening as hard as you’re talking.

15. “Your growth as a mature being is measured by the number of uncomfortable conversations you are willing to have.” Adult babies run from hard conversations. They go watch a Marvel Movie. But telling people what they don't want to hear is how you create change.

This is Kevin Kelly. The close friend Tim Ferriss goes to for advice. He taught him a philosophy for life that made Tim stop giving a f*ck. Here's the philosophy:

Many people see grief as pain. They avoid it, suppress it, or race to process it so they can expel it from their lives. Here’s a beautiful alternative: grief is unexpressed love. Holding onto it is a way of staying close to the people we’ve lost.

@MeghanEMurphy Brave woman. Do you know what happened to her?

If you think it’s too hard for you to speak out, too big a risk to resist, too uncomfortable or inconvenient to refuse to comply, think of what others around the world have done and still do, every day, to fight for their rights.

This Australia, is another hero. Let's not forget his name or his face. Pakistani refugee Faraz Tahir was murdered by the Bondi Junction misogynist thug on Saturday. The 30-year-old security guard was on duty at Westfield. I understand his female colleague was also murdered but I still do not have any information about her. Faraz tried to stop the attacker, alongside his colleague, but the coward ended their lives. Faraz had only been in Australia for a year and he came seeking refuge from persecution.

@doubleonobody @QBCCIntegrity @MagdaSzubanski Yes that’s the point! People’s preferences change. On the day there was extremely strong yes vote from indigenous people. That’s important. They wanted this. We said no.

@madspace @QBCCIntegrity @MagdaSzubanski 65% support was one and half years ago, not 6 months ago. That 80% claim was also from several months ago without acknowledging the continuous decline in support. In any case,

#Referendum2023 I’m so deeply saddened that this has been the result. To all our beautiful First Nations people…my heart goes out to you tonight I wish this had been different. Please know that there are millions of us who love and treasure and support you, that we hear your Voice and we will continue to walk with you 💔🙏🖤💛❤️

@doubleonobody @QBCCIntegrity @MagdaSzubanski The final election results in majority indigenous area show overwhelming support for the Voice via a #Yes23 vote. 80% was from two surveys which are an indication of support at the time they were conducted - not a lie. 6 months ago Aussies were 60% supportive - also not a lie.

@madspace @QBCCIntegrity @MagdaSzubanski Even cherry picking the best possible statistics still doesn't get you close to the 80% lie. What were the overall Lingari numbers, or do we only count indigenous votes in remote communities, rather than all indigenous votes?

"This picture captures my daughter and her friends in our Iranian village. The spirit of this generation fills me with hope for the future. The promise of brighter days is near. #WomanLifeFreedom forever." A message from Iran.

@dennistclark Whaaat?! that sounds like a strange comedy skit... only apparently it's not... :(

@AlmaDeutscher Wonderful list!

International Women's Day 2023: Four Women Composers and Their Contributions to the Art Form - operawire.com/four-great-con…