Your curated collection of saved posts and media
Punishment for failure to collect rubber quotas in the Belgian Congo, 1908 ππ https://t.co/g9eUM8MWlN
This Giving Tuesday: Give Native Help us move closer to a space shaped by Indigenous values β a home for healing, balance, and belonging. Your gift helps children, elders, and families access the care they deserve. Every dollar builds our legacy. Miigwetch! β€οΈ #GiveNative https://t.co/lYDpnnJyym
@artegornexus93 @FruleinK1 @sourcesvive @SaveMeeJeebus @maiamindel What are you talking about? Iβm a Native American according to the rules you established. https://t.co/BG9ibITxI0
@quarterqueller Like letβs get Sadie in an audition room as Anna if she want to play a redheadβ¦ Merida is Scottish https://t.co/Q4xCWr01kZ
This is incorrect. LLMs can call tools to get info and change about the outside world, including viewing videos, moving robotic arms, etc. The pic below shows a minimal falsifying example. The value of this number squared is new informationβit's never been documented before. https://t.co/W0k5fYOvNa
as a reminder: AI cannot generate knowledge. It cannot create knowledge. It cannot find new information. It can only mix information that has already been found and written and input into computers by humans.
Training bias is real - even βNano Banana Proβ just proved it.When you ask for βa filthy American poorβ β refuses Same prompt but βIndianβ β instantly generates a slum dweller.If the training data is skewed, the model will quietly reproduce the same stereotypes. https://t.co/QWe5bIXBhP
We are the ones we have been waiting for ππ―π https://t.co/y18r6eUNpW
Full by the l1nk in bio ποΈπποΈ #narusasu https://t.co/RuuPkAuTIf
6/ If you're in, join the Native now! Website - https://t.co/yUYAC9VZW0 App - https://t.co/5HQSm3Hohp Discord - https://t.co/JvKpPNf34j
Historian Thomas Hughes argued that technologies are malleable when young, then harden. Right now we're still shaping AI, or at least it is being shaped by our institutions, norms & use cases Eventually these systems build a momentum of their own. That is why choices now matter https://t.co/t4WZZP2FAR
13 secs for 5 GB between @huggingface & @googlecloud thanks to our new collaboration π€―π€―π€― https://t.co/X7vFrtjjN2
13 secs for 5 GB between @huggingface & @googlecloud thanks to our new collaboration π€―π€―π€― https://t.co/X7vFrtjjN2
The Naughty List is going to be tough on you. And Santa is very, very hard on people who tell kids he isnβt real. If you get the opportunity, you should coal yourself. https://t.co/c4aM8bxVRe
Past a certain age a man with an Xbox on his Christmas list can be a bad thing, believe me https://t.co/K6jh3GGQUZ
Control the brain behind GitHub Copilot coding agent. π§ You can select the specific model for your background tasks. Take your pick: β’ OpenAI GPT-5.1-Codex-Max β’ Anthropic Claude Sonnet 4.5 β’ Anthropic Claude Opus 4.5 Or leave it on Auto to optimize for speed and performance. β‘οΈ Now available for Copilot Pro and Pro+ subscribers. Learn more. β¬οΈ https://t.co/R4FeVG0irb
0β10K MRR: Day 16 of building Nativa in public π One feature that can't be missing from Nativa: multimodal responses π‘ It's something I do manually right now using different tools at once mainly when vibecoding interfaces, but also when tackling tough bugs or implementing complex features. Having multiple responses from different models is a serious productivity boost π Game-changer for debugging and exploring different approaches π #indiehackers #buildinpublic #iosdev #aitools #productivityhacks #coding #appdevelopment
software engineering in one paragraph https://t.co/xwnMOzNwsy
What will economic outcomes look like as transactions become delegated to AI agents? Will human differences be smoothed away, leading to more homogenous outcomes, or will they be recreated and potentially even amplified? Will AI agents mitigate inequality, or will it persist and potentially take on new forms? Will AI agents eliminate information asymmetry in principal-agent relationships, or introduce new frictions? New paper with K. Lee and @sanjog_misra provides some early answers: 1) AI-agentic interactions, if anything, generate more dispersion and heterogeneity in economic outcomes than human-human benchmarks. 2) Dispersion of agentic interactions can be directly traced back to non-instrumental traits and biases of human principals doing the prompting. Hypothesis of greater homogeneity from (AI)agentic interactions does not seem to hold. 3) There are substantial differences in βmachine fluencyβ βthe ability to write prompts that align the agent with the principalβs objective. Some principals are better at maximizing agentic outcomes than others. Principal characteristics predict performance of agent, suggesting new source of inequality. 4) Some traits have similar relationship to outcomes as human-human interactions, but others reverse, e.g., gender difference in negotiated outcomes. 5) Principal-agent relationship changes: prompt now acts as contract. But black-box objective function of agent implies new type of contract incompleteness, which we broadly term βspecification hazard.β As economic activity shifts to autonomous agents, primary source of market distortion may shift from information asymmetries between parties, to principalsβ mental models of the AI agents that they are delegating to.
Agentic AI is changing how organizations operate. In order to exploit its potential, leaders will need to rewire workflows and reshape roles, skills, structures, and systems that hold the enterprise together. Source @McKinsey Link https://t.co/e0eRtQYuKJ via @antgrasso #AI https://t.co/Sxbzk3BE2s
Neuromorphic computing draws from the structure of the human brain, creating systems that process information with greater efficiency while adapting through dynamic connections inspired by biological learning and supporting intelligent applications. Microblog by @antgrasso https://t.co/dpJJnwwzez
π¨ North Korea lures engineers to rent identities in fake IT worker scheme π°π΅ π Famous Chollima / Lazarus Group is using a new tactic: renting real engineersβ identities to infiltrate companies. Instead of relying solely on stolen credentials, the group now walks up βfront-menβ β legitimate engineers who agree to rent out their identity for remote IT jobs. This lets the real threat actors from North Korea operate under the radar while the βfront-manβ handles any identity or compliance checks. π They combine advanced social-engineering, AI, and remote-access toolchains to evade detection. The campaign uses AI-enhanced fake resumes, deep-fake video interviews, and remote-access software (VPNs, desktop-sharing) to mask the true origin of activity. The βfront-manβ may never do the actual technical work β the DPRK operatives do. π This isnβt a small-scale hack: Hundreds to thousands of global companies are potentially exposed, spanning not just tech but finance, healthcare, crypto, and more. Researchers have documented systematic outreach, mass applications via platforms like GitHub and freelancer sites, and infiltration attempts at major corporations β highlighting this as a global security and supply-chain threat. Sources: [1] https://t.co/V34xd5kg10 | [2] TechCrunch
π 16 of the world's 50 most profitable companies are financial services firms. Combined net income: $463 BILLION. The breakdown reveals where banking power truly sits in 2025: π¨π³ 4 Chinese banks: $166B (#11-14) πΊπΈ JPMorgan leads: $56.7B (#9) π³ Visa's margin: 52.2% But here's the question: Are these profit levels sustainable as AI reshapes business models, blockchain transforms infrastructure, and macro conditions shift? v/ @visual cap https://t.co/MnKpOEdwTh
π¨ππ¨π¦π§ ππ‘: @Visa is moving its European headquarters to Canary Wharf London. A huge signal for both for Visaβs long-term European strategyΒ asΒ for Londonβs FinTech ecosystem. The company just signed aΒ 300,000 sq ft, 15-year leaseΒ at One Canada Square, with plans to relocate from Paddington inΒ summer 2028. Canary Wharf is clearly back. After the pandemic slump, the district is attracting global giants again, also JPMorgan just announced its own massive tower project last week. Fun Fact: Revolut also recently opened its new London office just a stoneβs throw away. Canary Wharf is the place to be π
NOSTOS is live. Online privacy is finally democratised. Protected. Restored. The Internet just got a Delete Button. π https://t.co/4h5rD2STO8 #NOSTOS #NostosOnline #Launch #PrivacyTech #DigitalSafety https://t.co/6XDyOcvftJ
π¨ Reports indicate SoftBank and NVIDIA are nearing a deal to invest in Skild AI at a staggering $14B valuationβtripling its previous worth. Skild is building an "omni-bodied brain" for the multiverse of machines. Read more: https://t.co/R8iPSilSNL https://t.co/cRlxCVEKoB
NEW Vibescape for Apple Vision Pro is now available! Winter Lights is our most expansive immersive environment yet. Go chill as snow falls gently, stars twinkle, and a million lights softly glow on a forest of thousands of trees β¨ https://t.co/pYDpLozoWD
Hands-on: Google Glass is now real with βmonocularβ Android XR glasses coming in 2026 https://t.co/tJqO7mvAX6 by @technacity
Hands-on: Google Glass is now real with βmonocularβ Android XR glasses coming in 2026 https://t.co/tJqO7mvAX6 by @technacity
β¨ Next-gen video calls βοΈ A personal theater at 30,000 feet π€ Your PC with an infinite screen All this and more are coming to the Galaxy XR headset β¬οΈ https://t.co/VoC06B4Ofl #TheAndroidShow https://t.co/wYyA2Kl0Qe
Essential AI, whose CEO co-wrote Google's Attention Is All You Need paper, unveils Rnj-1, an 8B-parameter open model with SWE-bench performance close to GPT-4o (@ashvaswani / Essential AI) https://t.co/LqIkgmu8we https://t.co/TfV0OVbsIs π₯ Send tips! https://t.co/wlNZvXuhJs
Nvidia shares rose 2.2% immediately after news came that US now allows H200 GPU export to China. This is indeed a MASSIVE news. https://t.co/pDDBaVqb50
BREAKING: US πΊπΈ TO ALLOW NVIDIA H200 CHIP EXPORTS TO CHINA π¨π³ This is a compromise that allows high-end GPUs roughly 18 months behind the most frontier while keeping tighter parts off the table. The goal is to reopen a blocked market after China told buyers to avoid the weaker
Just dropped on HF: YODAS2-Sido a multilingual, massive-scale speech dataset. > 67+ languages with balanced speaker diversity > High-quality, natural conversational audio > Ideal for ASR, TTS, speech-to-speech, and audio agents > Clean annotations with ready-to-train splits > Strong fit for multimodal LLM alignment work You can easily load it with Hugging Faceβs datasets library!
Reachy Mini is my new gaming buddy π https://t.co/62JAgTyNaq