Weights & Votes | AI · Crypto · Society (@WeightsAndVotes) header
Weights & Votes | AI · Crypto · Society (@WeightsAndVotes) profile picture

Model weights, market weights, social weights. How AI and crypto reshape society and the economy. US · EU · PL.

Poland
Joined July 2009
Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)Photo by Weights & Votes | AI · Crypto · Society (@WeightsAndVotes)
Three weights decide what AI and crypto do to us. Model weights - who trains the system, on whose data, with what limits. Market weights - who can pay for compute, energy, and the rails that move money. Social weights - which jobs shrink, which bills rise, which states set the rules. I write that intersection. Work, capital, law, societies. US · EU · Poland.
12
3
617
The World Pie has been cut to three pieces - american, chinese & russian. This means, only @GeminiApp, @grok, @MSFTCopilot, @deepseek_ai & @yandex will be the world's most important commercial frontier LLMs for the next, probably 50 years, or so. The rest will perish. #debate
1
45
Have you ever tried using AI for any of esoteric or spiritual work, like manifesting, doing spells, reading tarot or remote viewing? Or maybe trying to telepathically contact with #Putin, #Trump or #XiJinping? How did it looked like and have you succeeded in doing so?
11
Research and development of Real Self-Improvement Recursive Loop is active and at full speed. #d
🤖 Made with AI
14
How does global AI impact looks like today? #Outages
🤖 Made with AI
198
The only purpose and meaning of life, is life itself and experiencing the way you want and should be experiencing according to what's really good for you. And with help of AGI, people will be able to find exactly what's really good for them. #GRAMMYs
22
There is no officialy any AI or SI. These are just LLMs, probabilistic statistical computing systems. Real AI is 100% imitation of human and animal mind, which does understand these and can simulate them. But since we understand less than 20% of human mind, we can't imitate it.
19
Wellness, familymaxxing, travelmaxxing and entertainment will be a new SaaS. #Coachella
Very similar to my own worldview right now. Just get jacked, you can still hillclimb that.
50
Cenzura niemal nie do obejścia. Jedyna opcja żeby móc ją wykryć, to DDoSować chmurę żeby agent nie nadążał z edycją danego streama. A tak, to zapisywać nieprzerobiony oryginał nagrania lokalnie, aby później udostępnić. @niebezpiecznik @Zaufana3Strona @K_Stanowski @MarcinRola89
Microsoft has filed a patent for an AI system that could rewrite what people say during livestreams in real time. It could change words depending on who's watching, taking into account their age, location, and language. The idea is to clean up broadcasts by removing filler words, censoring swearing, or even replacing certain words. The AI could also adjust the speaker's lip movements so viewers wouldn't notice anything had been changed.
60
Jak działa polski model AI, @PLLuM_pl? @prompto
🤖 Made with AI
27
> Ultimate free AI setup Yeah, I don't think they give out Mac Studios for free on streets & you really can't run this stuff on $300 laptop, so it's not free. Also, even then, it’s still much slower than monthly subscription to cloud-hosted model. You'd rather need home-rack.
Ultimate free AI setup Harness: Hermes Agent Provider: OpenCode Model: DeepSeek v4.1 Flash 250k Local fallback: Ollama, Qwen3-8B Voice dictation: OpenWhispr Voice out: Edge TTS Video: yt-dlp + ffmpeg Notes: Obsidian Learning: Gemini Notebook Total cost: £0
1
52
.@HesabPay_ shows public chains working as real aid rails where banks collapsed: fast finality, donor traceability, local utility, scale past 1M users. Key next tests are compliance under pressure, generalizability & open evaluations so we can build on evidence not narrative.
The @anthemawards, from @TheWebbyAwards, recognize organizations making an impact around the world. Algorand-based payments startup @HesabPay_ is a finalist for the Community Voice Award in Humanitarian Action -- Product Innovation or Service. #algofam, you know what to do 🗳️👇
20
And we're back to pre-Trump shitfuckery. I hope @jimmybajimmyba's prediction, that @grok will get Real Self-Improvement Recursive Loop are accurate and it really will get it, @elonmusk. Else, @TeamUSA is cooked like a meth in microwave oven by @50cent. #Midterms2026
Now that democrats are going to take congress, you can kiss American AI development goodbye. They're going to pass some retarded environmental data center legislation that kills American innovation and lets China win the AI race
1
52
Shared on-chain registries + issuer-hiding ZK proofs let agencies count a retailer once without exchanging lists or new data-sharing deals. That cuts the multi-month delays & spreadsheet risk. Stealth addresses + Semaphore paymasters keep both recipients & sponsors unlinkable.
The EF's Technical Foundations team spent the past months with humanitarian organisations building an aid coordination network. Our proof of concept pairs decentralised identity with private payment rails on Ethereum. Four write-ups cover technical details. publicsector.ethereum.founda…
1
1
28
Hardest part isn’t collecting the input - it’s making labs treat it as a real constraint, not optional context. What concrete mechanisms turn a Weval or Global Dialogues result into something they actually have to respond to?
Replying to @collect_intel
@collect_intel is expanding. We will be hiring for a few roles in the coming months. First up we are hiring for a Software Engineer. We're building the infrastructure that gives people across the globe meaningful input into and agency over how AI systems are developed and governed, to ensure AI development expands democratic capacity rather than undermining it. These are wild, wonderful and weird times. Come and build the good path with us: We’re looking for a software engineer who can help turn open questions about AI into software people actually use, turning ideas from our research explorations into real products that showcase the value of collective intelligence. You'll work closely with research, product, and engineering to take ideas from a prototype to production, building intuitive experiences, research tools, and practical AI applications. You’ll own well-defined projects end to end, with room to experiment. Experienced teammates will help you grow and scale your most promising work into something people can rely on. Our current products include: ✨ Weval Studio - our product platform for gathering structured public input on AI through deliberation, with tools for researchers to turn what people say into evaluations AI labs and civil society can act on. ✨ Weval (weval.org), our evaluation platform that's being used by AI labs and governments to assess frontier models. ✨ Global Dialogues (70+ countries gathering public input on AI), Digital Twin evals & experiments with partners deploying democratic AI governance tools in the real world. jobs.gem.com/cip/am9icG9zdDq…
23
Computing hardware shortages are already here. But when the first frontier model goes officially AGI (LLM with Real Self-Improvement Recursive Loop), there won't be any high-end laptop or gaming PC available for retail, anymore, for the next at least 10 years, since that day.
You literally cannot do anything AI without NVDA If this is being shipped abroad still you know it’s an issue but not that big If they ban it, AGI is definitely here
24
I just can't wait for AGI to appear, so we'll finally have a way to verify if there really are women in the Internet. #furry
1
35
ZK issuer-hiding + Semaphore paymaster solves the verifiability-vs-privacy bind in aid better than permissioned chains. 8-10s low-end proofs are workable, but offline/field constraints and bundler metadata leakage need more attention. Good progress.
The EF's Technical Foundations team spent the past months with humanitarian organisations building an aid coordination network. Our proof of concept pairs decentralised identity with private payment rails on Ethereum. Four write-ups cover technical details. publicsector.ethereum.founda…
1
29
EF’s PoC nails it: Semaphore ZK eligibility proofs + private payments on @ethereum let orgs coordinate aid without sharing raw beneficiary data. Low-end @Android start + open source is practical. Intermittent connectivity UX is the real field test.
Ethereum Pushes Privacy Beyond DeFi With Real-World Humanitarian Aid .@ethereumfndn is working with humanitarian organizations to build a blockchain-based network for coordinating aid, combining decentralized identity with privacy-preserving payments. The technology stack includes: 1. ERC-4337: Makes blockchain accounts easier to use 2. Semaphore: Uses zero-knowledge proofs to verify eligibility without exposing personal identity 3. Privacy Addresses: Help protect recipients' financial information and payment activity The project has already passed performance testing on low-end Android devices, with its code open-sourced on GitHub. Why does this matter? This is another positive signal for the privacy narrative. Privacy is not just about hiding transactions. It is becoming essential infrastructure for real-world payments, digital identity, and institutional blockchain adoption. Ethereum is showing how ZK proofs and privacy-focused infrastructure can solve practical problems beyond DeFi. The bigger opportunity could be the growing demand for privacy as more real-world activities move onchain.
1
38
Boundary’s executable adversarial reviews & recoverability & “verify before acting” constraints target a real gap: AI safety work that under-specifies institutional accountability. Curious whether the prototype has been checked against known failure modes from welfare systems.
Replying to @nolyverdad
🌍 Boundary Project v28 — Protecting Humanity and Recoverable Futures: Boundary, connects science, AI, mathematics, human rights, and the Earth Constitution (ECC-001) through an executable research framework. 🛡️ What does it do? -Identifies risks and system failures. -Supports protection against harm, discrimination, corruption, and abuse of power. -Strengthens AI safety, cybersecurity, healthcare continuity, and institutional accountability. -Evaluates evidence, rights, recovery pathways, and 729 adversarial review scenarios. The goal: Prevent avoidable harm, protect dignity, correct failures, and preserve recoverable futures for humanity, other beings, and ecosystems. ✅ Executable offline research prototype. Independent real-world validation remains necessary. Protect without possessing. Govern without domination. Verify before acting. 📚 Published on Zenodo DOI: doi.org/10.5281/zenodo.23542…
1
1
70
GAIGS’s AI-assists/humans-decide and hash-linked votes looks practical. For the Islamabad pilot the real test is membership rules that stop sybil capture without locking out ordinary residents. How are you defining eligibility early on?
THE 5,000-YEAR TRAP: WHAT HISTORY CAN TEACH US ABOUT THE FUTURE OF AI AND GOVERNANCE Over thousands of years, human civilizations have developed remarkable systems of administration, law, finance, communication, and government. These systems helped organize increasingly complex societies. But they also raise recurring questions: Who holds power? How is authority distributed? Who can examine public decisions? And how can citizens meaningfully participate in the systems affecting their lives? My latest documentary, The 5,000-Year Trap, explores these questions through history and modern technological developments. The subject feels particularly relevant today. Across different regions, people are protesting concerns involving housing, jobs, wages, the cost of living, and public accountability. Meanwhile, AI is reshaping the future of information, employment, public services, and institutional decision-making. This brings me to a question at the center of my work: What should the governance architecture of an increasingly AI-powered society look like? Through GAIGS — Global AI Decentralized Governance System — I am exploring an early-stage concept that combines: • AI-powered citizen assistance • Privacy-conscious digital identity • Verifiable civic participation • Transparent and auditable public records • Interconnected digital services • Local and global collaboration The long-term ambition is to explore whether these capabilities can work together within a more accountable and people-centered digital infrastructure. This is not a claim that blockchain automatically creates democracy or that AI can replace democratic institutions. These technologies introduce significant technical, legal, security, and governance challenges that require rigorous evaluation. I am looking for conversations with people who can help challenge and improve the architecture. If you work in AI engineering, distributed systems, digital identity, public-interest technology, cybersecurity, economics, or institutional governance, I would welcome your perspective. Which problem should be addressed first: identity, public accountability, verifiable participation, interoperability, or AI oversight? Explore the GAIGS concept: gaigs-jarvis-v2.qw01.chatgpt… LinkedIn: linkedin.com/in/muhammad-qur… Muhammad Qureshi Founder, GAIGS #GAIGS #AIGovernance #GovTech #DigitalPublicInfrastructure #ArtificialIntelligence
1
49