# AI Research İlgili Makaleler

HTX Haber Merkezi, kripto endüstrisindeki piyasa trendleri, proje güncellemeleri, teknoloji gelişmeleri ve düzenleyici politikaları kapsayan "AI Research" hakkında en son makaleleri ve derinlemesine analizleri sunmaktadır.

Weekly Token Unlock: XPL to Release Millions in Token Value

Weekly Token Unlocks: XPL and SoSoValue Tokens Worth Over $10 Million Set for Release This week sees significant token unlocks for two projects. SoSoValue, an AI-powered crypto investment research platform, will unlock 13.03 million tokens, valued at approximately $3.87 million. The project aims to merge CeFi efficiency with DeFi transparency. Separately, Plasma, a Layer 1 blockchain designed for global stablecoin payments, will unlock 110 million tokens worth about $10.42 million. The network focuses on high throughput, stablecoin-native features, and full EVM compatibility. Both projects have provided their detailed token release schedules.

marsbit13 saat önce

Weekly Token Unlock: XPL to Release Millions in Token Value

marsbit13 saat önce

Large Language Models Ace All Exams, Yet Move Farther from AGI: What Does This Paper Reveal?

The article discusses the ongoing challenge of defining and achieving Artificial General Intelligence (AGI). It notes that industry leaders have set vague, often profit- or time-based benchmarks for AGI, while the concept itself lacks a consensus definition—a situation the article compares to a "Rorschach test." It highlights a recent 2025 paper by researcher Michael Timothy Bennett, who proposes a new, measurable definition. Bennett frames AGI not as mimicking human performance on tests, which current large language models (LLMs) have already mastered, but as an "artificial scientist." A true AGI, according to this view, should be able to widely and efficiently adapt to new environments and tasks within real-world constraints (like computational and energy limits), focusing on the *discovery of new knowledge* rather than the replication of existing data. The author contrasts this with the current dominant approach of "scale-maxing"—massively scaling up data, parameters, and compute. While powerful, this method leads to models that fail on out-of-distribution problems and lack core intelligent abilities: they are passive learners, cannot reason causally, and cannot actively experiment or balance exploration with exploitation. The article argues that Bennett's framework offers a crucial shift. It makes AGI a quantifiable engineering problem and proposes new evaluation "adaptation benchmarks" that test an AI's ability to actively learn in novel scenarios. The conclusion is that achieving AGI will require a fundamental reset—a fusion of multiple methodologies beyond simple scaling, moving AI from mimicking patterns to embodying the scientific spirit of inquiry and discovery.

marsbit05/28 00:24

Large Language Models Ace All Exams, Yet Move Farther from AGI: What Does This Paper Reveal?

marsbit05/28 00:24

AlphaGo's Creator Puts AI into a 23-Year-Old Artificial Society: All Three Toughest Challenges for AI Agents Are Here

Demis Hassabis, CEO of DeepMind, has embarked on a new AI research venture by partnering with the long-running space MMO, EVE Online. This collaboration, announced in early May, aims to use the game's 23-year-old, player-driven persistent universe as a testbed for tackling three core challenges in AI agent research: long-horizon planning, memory, and continual learning. Unlike previous DeepMind environments like AlphaGo (Go) or AlphaStar (StarCraft II), EVE Online features no fixed end state. Its single-shard universe has fostered complex, emergent player societies with real economies, political alliances, and wars that can span months or years. These conditions naturally demand the very skills—long-term strategic planning, maintaining memories over extended periods, and adapting to constant change—that are hardest for current AI agents to master. The research will initially use an offline version of EVE, providing a controlled, complex sandbox without interfering with the live player server. This move continues DeepMind's trajectory of using increasingly complex and open-ended virtual worlds for AI training, from Atari games and Go to StarCraft II and the SIMA project. The EVE environment represents a significant step towards testing AI in a persistent, socially complex, and continuously evolving world shaped by human behavior over decades.

marsbit05/25 00:08

AlphaGo's Creator Puts AI into a 23-Year-Old Artificial Society: All Three Toughest Challenges for AI Agents Are Here

marsbit05/25 00:08

A Western Scholar's Field Notes from Visiting Chinese AI Labs: Humility, Openness, No Philosophy, Just Wanting to Train Better Models

An Impressed Western Scholar's Observations from a Tour of Chinese AI Labs: Humility, Openness, and Pragmatism A visiting Western academic's key takeaways from touring numerous Chinese AI companies highlight striking differences from the US ecosystem. Researchers in China displayed deep humility, frequently praising competitors like DeepSeek with genuine admiration. They operate in a remarkably open and collaborative atmosphere, often sharing research publicly, contrasting with the more guarded, zero-sum competition perceived in the West. The primary focus is intensely pragmatic: training better models. Philosophical debates about AI's societal impact or consciousness are largely absent; the drive is purely technical. Researchers are young, often PhD students working in industry, and are highly online and tool-savvy. Their passion is palpable when users engage with their models. Public sentiment also differs, with Chinese citizens showing greater optimism about AI's benefits compared to Western skepticism. While major closed-source players exist, the prevalence of open-source models fosters a unique cooperative spirit. The visit left the author optimistic about the future of open AI research and hopeful for increased international collaboration.

marsbit05/08 03:19

A Western Scholar's Field Notes from Visiting Chinese AI Labs: Humility, Openness, No Philosophy, Just Wanting to Train Better Models

marsbit05/08 03:19

Exploring Physical World AGI with "Visual Reasoning", ElorianAI Raises $55 Million

ElorianAI, co-founded by ex-Google AI expert Andrew Dai and former AI specialist Yinfei Yang, has raised $55 million in early funding to develop next-generation AI systems with advanced visual reasoning capabilities. While current large models excel in text-based tasks like programming and math, they perform poorly in visual reasoning—even top models like Gemini only match a 3-year-old’s ability in basic visual benchmarks. The key limitation lies in the architecture of current vision-language models (VLMs), which first convert visual inputs into text before reasoning, losing critical spatial and structural information. ElorianAI aims to build a native multimodal model that processes and reasons directly in visual space, enabling deeper understanding of physical relationships, constraints, and environments. The company plans to release a state-of-the-art visual reasoning model by 2026, with potential applications in robotics, disaster management, engineering, healthcare, and AI hardware. By using high-quality, diverse, and synthetically generated data, ElorianAI intends to create models that don’t just perceive but truly understand and reason about the physical world—bringing us closer to visual AGI.

marsbit04/23 10:10

Exploring Physical World AGI with "Visual Reasoning", ElorianAI Raises $55 Million

marsbit04/23 10:10

a16z Outlook: AI Will Reshape Research, Identity, and the Foundation of the Network Economy by 2026

a16z Crypto experts predict three major AI trends for 2026. First, AI will increasingly perform substantive research tasks, moving from basic assistance to handling abstract, multi-step reasoning. This will enable a new "generalist" research style that explores conceptual relationships and leverages AI "hallucinations" for creative breakthroughs, though it requires better model interoperability and compensation mechanisms—potentially solvable via blockchain. Second, identity verification will shift from "Know Your Customer" (KYC) to "Know Your Agent" (KYA). With non-human agents vastly outnumbering humans in financial services, there is a critical need for infrastructure that issues cryptographically signed credentials to link agents to entities, constraints, and liabilities. Without KYA, agents will remain excluded from economic systems. Third, AI agents impose a "hidden tax" on the open web by disrupting its economic model. They extract data from ad-supported websites (the context layer) while bypassing revenue sources like ads and subscriptions. To sustain the open internet, new techno-economic solutions are needed—such as real-time usage-based compensation, nano-payments, and attribution systems—to automatically reward content contributors and replace static licensing models.

比推01/12 14:48

a16z Outlook: AI Will Reshape Research, Identity, and the Foundation of the Network Economy by 2026

比推01/12 14:48

a16z: Three Major AI Trends for 2026

a16z: Three AI Trends for 2026 1. AI Takes on Substantial Research Tasks: AI models are evolving to handle complex, abstract instructions and assist in research, particularly in reasoning. They are beginning to solve difficult problems and foster a new "generalist" research style that focuses on connecting ideas and making inferences from hypothetical answers. This requires new "nested agent" workflows where models collaborate and refine each other's outputs, though better model interoperability and compensation methods (potentially via blockchain) are needed. 2. The Shift from KYC to KYA (Know Your Agent): The bottleneck in the agent economy is shifting from intelligence to identity verification. With non-human identities vastly outnumbering humans in finance, there's a critical need for a "Know Your Agent" infrastructure. Agents require cryptographically signed credentials to transact, linking them to their principals, constraints, and liabilities. 3. Solving the Open Web's "Invisible Tax": AI agents are disrupting the economic foundation of the open web by extracting data from ad-supported sites (the context layer) while bypassing their revenue models. This creates an "invisible tax" that threatens content creation. Solutions are needed to automatically reward content creators, moving from static licensing to real-time compensation systems using technologies like nanopayments and attribution standards, potentially blockchain-enabled.

marsbit01/12 08:09

a16z: Three Major AI Trends for 2026

marsbit01/12 08:09

Five Great Crypto Gifts for the Holiday Season

Five Crypto-Friendly Gifts for the Holiday Season Instead of traditional gift cards, consider these five practical and thoughtful cryptocurrency-themed gifts for friends interested in crypto. Each is easy to give, straightforward to use, and designed to last through multiple market cycles. 1. **AskSurf.ai** – Offers AI-powered crypto research, on-chain insights, and sentiment analysis. Various pricing plans are available. It’s a year-long information advantage for traders and creators. 2. **Ledger Nano S Plus** – Priced at $59, this hardware wallet supports over 5,500 assets and is ideal for beginners entering self-custody. A perfect gift to show you care about their crypto security. 3. **Fireproof Metal Seed Phrase Backup Plate** – Ranging from $6.99 to $8,000 (for a gold version), this essential tool protects recovery phrases from fire, water, and accidents. Often overlooked, it makes for a highly considerate gift. 4. **Gold Bar Bourbon Limited Edition (750ml)** – A $109.99 bourbon blended from 5- to 14-year-old whiskies, limited to 21,000 bottles. Each includes a metal Bitcoin commemorative coin, nodding to Bitcoin’s 21 million supply cap. Must be gifted to those 21+. 5. **One Shot Miner PRO (2025 Edition)** – At $59.99, this plug-and-play miner offers a fun, low-power way to participate in Bitcoin. It runs 24/7 and gives users a chance to earn Bitcoin rewards every ten minutes.

深潮12/24 07:11

Five Great Crypto Gifts for the Holiday Season

深潮12/24 07:11

活动图片

Industry Newsright-arrow

Regulatory Policyright-arrow