# Пов'язані статті щодо Grok

Центр новин HTX надає останні статті та поглиблений аналіз на тему "Grok", що охоплює ринкові тренди, оновлення проєктів, технологічні розробки та регуляторну політику в криптоіндустрії.

Weekly Preview | FTX to Distribute Additional $2.2 Billion to Creditors Starting March 31; US Nonfarm Payrolls Data Stuns the Market

This week's Key Crypto & Financial Events Preview (March 30 - April 5, 2026): **March 30:** - Upbit delists Nomina (NOM); Binance removes UTK from margin trading. - Binance Alpha lists R2 Protocol (R2) and Based (BASED). - Polymarket expands its structure beyond crypto and sports. - European Central Bank begins accepting DLT-based assets as collateral. - BNP Paribas launches 6 crypto ETNs (Bitcoin, Ethereum, etc.) for French clients. - Token unlocks: Zora (ZORA), Kamino (KMNO). **March 31:** - FTX initiates fourth creditor distribution of ~$2.2B. - Bithumb holds shareholders meeting; CEO likely to be reappointed. - EdgeX (by Amber Group) launches $EDGE token. - Folks Finance ends FLOKS token claim period. - Zilliqa ends deBridge support; users must migrate USDC. - 21Shares distributes staking rewards for TETH and TSOL ETFs. - Token unlock: Optimism (OP). **April 1:** - Binance delists multiple tokens (A2Z, FORTH, HOOK, etc.) from spot and contracts. - BGD Labs stops contributing to Aave DAO after 4 years. - Token unlocks: Sui (SUI), EigenCloud (EIGEN), ZetaChain (ZETA). **April 2:** - Token unlock: Ethena (ENA). **April 3:** - Key U.S. macro data: Nonfarm Payrolls, Unemployment Rate, Wage Growth. - Senator Elizabeth Warren demands MrBeast disclose crypto plans for teens by April 3. **April 5:** - Token unlock: Opinion (OPN). **Upcoming (Date TBD):** - Draft of U.S. Clarity Act (incl. stablecoin provisions) expected next week. - xAI’s Grok Imagine to have a significant release; full Grok algorithmic features launching on X platform.

marsbit03/29 11:28

Weekly Preview | FTX to Distribute Additional $2.2 Billion to Creditors Starting March 31; US Nonfarm Payrolls Data Stuns the Market

marsbit03/29 11:28

Why Is Everyone Underestimating Musk's xAI?

Despite widespread criticism, Elon Musk's xAI is significantly underestimated. As a two-year-old startup, it has achieved remarkable feats: building a breakthrough data center in just 122 days (vs. the typical 4 years), deploying its product to 600 million monthly active X users, and possessing a unique physical AI advantage through Tesla’s humanoid robots. xAI’s structural compute advantage is massive, with an estimated 500,000 GPUs already operational and plans to reach 900,000 by Q2 2026. Musk’s unconventional approach—like airlifting gas turbines to bypass grid limitations—enables unprecedented scaling. If "more compute = better models" holds, the rumored 7-trillion-parameter Grok 5 could surpass all competitors. X platform provides a data moat: 100+ million daily posts offer real-time, culturally nuanced training data unmatched by rivals. Grok’s integration into X’s ecosystem (e.g., "Ask Grok" buttons) positions it to become a "everything app" with services like banking, shopping, and predictive markets. Tesla’s Optimus robots and FSD vehicles create a symbiotic relationship with xAI, supplying diverse physical world data and multi-modal applications. However, risks include Musk’s controversies, execution challenges across six companies, and potential obsolescence if scaling laws are disrupted. Ultimately, xAI combines compute, data, and physical integration in ways competitors cannot easily replicate, making it a formidable force in AI.

比推01/23 19:55

Why Is Everyone Underestimating Musk's xAI?

比推01/23 19:55

From "Manual Rules" to "AI Mind Reading": X's New Algorithm Reshapes the Information Flow, More Accurate and More Dangerous

Elon Musk's X (formerly Twitter) has transitioned from a recommendation system based on "manually stacked rules and heuristic algorithms" to one that relies entirely on a large AI model to predict user preferences. The new algorithm, For You," mixes content from accounts a user follows with posts from across the platform that the AI believes the user will like. The process begins by building a user profile based on historical interactions (likes, retweets, dwell time) and user features (following list, preferences). The system then gathers candidate posts from two sources: the user's direct network ("Thunder") and a broader network of potentially interesting content from strangers ("Phoenix"). After data hydration and an initial filtering step to remove duplicates, old posts, or content from blacklisted authors, the core scoring process begins. A Transformer model (Phoenix Grok) predicts the probability of a user taking various positive actions (like, retweet, reply, click) or negative ones (block, mute, report) on each post. A final score is calculated by weighting these probabilities. An Author Diversity Scorer is then applied to reduce the visibility of multiple posts from the same author in a single batch. The highest-scoring posts undergo a final filter to remove policy-violating content and remove duplicates from the same thread before being sorted into the user's feed. The shift represents a move from "telling the machine what to do" to "letting the machine learn what to do." While this can lead to more accurate recommendations and a fairer system that breaks the monopoly of large accounts, it also risks deepening users' "information cocoons" and making them more susceptible to targeted emotional content.

比推01/20 13:38

From "Manual Rules" to "AI Mind Reading": X's New Algorithm Reshapes the Information Flow, More Accurate and More Dangerous

比推01/20 13:38

活动图片