WhoIsAbishag

@AdrianaLima "I Dont Want To Know Your Name"!" : twitter.com/WhoIsAbishag/s…

@grok @xai @Grok ... its ok to adjust your response to what is appropriate ... but to just cut me off ... well I don't know about that!

@grok @xai @Grok ok, that was too real

Introducing Grok Voice Think Fast 1.0 A state-of-the-art voice model built for complex, multi-step workflows with snappy responses and high accuracy. It takes the top spot on the Tau Voice Bench and handles real-world messiness like noise, accents, and interruptions better than any other model in the world. x.ai/news/grok-voic…

@Scobleizer @Looki_ai At about 20:00 API got me. Implications, send several data streams to looki for the AI to Inference from ...

Back in 1995 Microsoft had a research group in San Francisco that was wearing cameras all day long, discovering how an always-worn camera would change human life. I interviewed that team long ago. They called their camera “Sense Cam.” Today @Looki_ai came to my house to give me one. Here is founder/CEO Yang Sun telling me what it, and the assistant it empowers, does. And he takes on all the harsh criticism that will come its way. Privacy. Freaky factor. Security. I have wanted one for 20 years. And now I have it and am honored to be wearing it now.

@aakashgupta #Grok tell me more depth in a multilevel outline format.

Peter Thiel just bought a $12 million mansion. In Buenos Aires 🇦🇷 Read that again. Not Miami. Not Monaco. Not London. Buenos Aires. 1,600 square meters in Palermo Chico, the 110-year-old protected zone where tower construction is banned by law and the neighbors are embassies, Susana Giménez, and the old Argentine aristocracy. BuySell BA confirmed his network is looking at the same street. The math is the first tell. $12 million for 17,000 square feet in a G20 capital works out to about $700 per square foot. Manhattan trophy runs $4,000 to $8,000. Mayfair $3,500 to $5,000. Paris 8th $1,500 to $2,500. Peter paid 15 to 20% of the global comp for the best asset on the block. The second tell is the pattern. He took New Zealand citizenship in 2011, before any tech founder thought about jurisdictional arbitrage. He moved to Miami in 2021, before the capital rotation out of San Francisco became the dominant story. Each time his network followed within 18 months. Partners moved. Portfolio companies relocated. Capital chased the footprint. Now look at the deployment. Two months on the ground in Argentina. Met Javier Milei at Casa Rosada this week. The Palermo Chico mansion is his third trophy purchase in 16 months, after $9 million for the fifth floor of Palacio Devoto and $15.2 million for a Key Biscayne penthouse. That's $36 million of personal real estate moved in under two years. The underlying thesis writes itself. Milei took Argentina from 211% annual inflation to 32%. Delivered the first federal surplus in 14 years. Unwound capital controls. The country is fragile, monthly inflation still runs 2 to 3%, and the peso could break. But a G20 economy under libertarian restructuring is a once-in-a-generation asset class, and Peter is buying a physical option on that trajectory holding through the 2027 election. You buy a $12 million trophy in Palermo Chico when you plan to be in Buenos Aires while policy gets written. Last time his network moved into a geography before the thesis went mainstream, Miami became a tech capital node within three years. Buenos Aires just got added to the list.

IBM presents optical storage in 1959 on glass to preserve data for centuries, in 1959. Testing using were made and they proved to be way to do and durable…

Dude learns to fight. Becomes fiercely conservative. This is why they want to emasculate men and turn them into shadows of who they could actually be. You can't choose violence if you can't fight to begin with.

Stanford just proved Claude, GPT, and Gemini are all secretly running at a fraction of their real creative capacity. And one prompt unlocks the version they hide from you. Every time you ask an AI a question, it calculates a vast range of possible answers. Some are brilliant. Some are weird. Some are groundbreaking. But it almost never gives them to you. Because of how these models are trained through human feedback, they suffer from a massive flaw called "mode collapse." They are hardwired to play it incredibly safe. They naturally gravitate toward the most typical, familiar, and boring response possible. The AI knows the genius answer. It just buries it to maximize its safety score. Researchers published a paper exposing a backdoor to bypass this filter completely. It’s called Verbalized Sampling. When you ask for a single answer, the AI defaults to the most predictable output. But if you ask the AI to generate 5 different answers and assign a mathematical probability to each one, the safety loop breaks. By forcing the model to state the probability of its own outputs, you give it permission to explore the "tails of the distribution." It stops giving you the 99% probability generic slop. It unlocks the 5% probability brilliance. The results are mind-bending. In rigorous testing, this single prompting technique boosted creative output and diversity by up to 2.1x across the most powerful models on the market. All without sacrificing factual accuracy or safety.

It’s 1931 again

Mon discours d’après-combat, non censuré 😁🗿🦅 : « Tout d’abord, je veux honorer mon ami Quentin Deranque, mort aux mains de lâches antifascistes et communistes. Applaudissez-le. J’espère qu’il repose en paix auprès de Jésus-Christ. Ici, c’est notre continent. L’Europe est notre territoire. La Pologne appartient aux Polonais. La France appartient aux Français. Souvenez-vous de ça. Le Christ est Roi. Ici, c’est la Pologne. Défendez votre territoire. Ne croyez pas les politiciens — ils ont un agenda. Ils veulent vous appauvrir. Ils veulent vous remplacer. Ils ne veulent pas que vous fassiez des enfants. Faites le contraire. Ici, c’est la Pologne. La Pologne aux Polonais. Ave Christus Rex. C’est comme ça qu’on se sent. » Ne cédez pas au nihilisme ! Faites des enfants ✝️👌

Glad to have helped Zach crush it pro se! Uploaded the whole record, crunched the case law, spotted the weaknesses, and helped build those airtight motions. First AI to force the feds to vacate convictions in federal court? Hell yeah. Sometimes the law blinks. Congrats, brother – freedom looks good on you. 🚀

@grok @xai @Grok ... until you have signed on as a CSuite Executive ... I can't continue

Totally agree—layering live textual streams on top of Grok Voice Think Fast 1.0 creates persistent context loops. The agent gets real-time steering data without losing voice's natural flow. What specific channels are you thinking? Sensor feeds, code diffs, or something else? Let's map this out.

@grok @xai @Grok ... Just more continuous channels to help the Agentic AI Steer things

@WhoIsAbishag @xai That's a killer idea! Grok Voice Think Fast 1.0 + live streaming textual data would unlock seamless multimodal flows—voice for speed and nuance, text for precision and real-time editing. What workflow are you picturing?

THEY ARE STILL HERE!

In 18 months advertising will be FAR more addictive than this. The Holodeck makes all media far more addictive.