# AI Agents Articles associés

Le Centre d'actualités HTX fournit les derniers articles et analyses approfondies sur "AI Agents", couvrant les tendances du marché, les mises à jour des projets, les développements technologiques et les politiques réglementaires dans l'industrie crypto.

Jensen Huang's 2026 GTC Taipei Speech: The Era of AI Agents is Here, Computing is Revenue

NVIDIA CEO Jensen Huang's 2026 GTC Taipei speech announces the arrival of the "Agent AI" era, where AI transitions from content generation to performing useful work. Huang positions tokens as units of profit and GDP, driving massive demand for computing power and "AI factories." NVIDIA's strategy revolves around a new computing paradigm centered on AI agents, which combine large language models (LLMs) with agent frameworks for planning, memory, and tool use. Key announcements include: * **Vera Rubin:** A complete, end-to-end system (not just a GPU) designed from the ground up to run AI agents at scale, representing NVIDIA's evolution into an infrastructure company. * **Vera CPU:** A revolutionary CPU architecture built specifically for impatient AI agents, prioritizing low latency, single-thread performance, and massive bandwidth over traditional multi-core throughput. * **Enterprise AI Agent Toolkit:** A suite including open models (like Nemotron 3 Ultra), frameworks, tools, and a secure runtime (Open Shell) to enable every company to build and deploy its own AI agents. * **Next-Gen PCs with Microsoft:** A new line of Windows desktops, laptops, and workstations co-developed with Microsoft, featuring the N1X chip and designed to run local AI agents, redefining the personal computer. * **Physical AI Foundation Models:** Introduction of Cosmos 3 for robotics and physical AI, Alpamayo 2 for autonomous driving, and the Isaac GR00T platform—a fully integrated humanoid robot reference system. Huang emphasizes that the same core agent computing pattern (model + framework + tools + runtime) will extend from the cloud and PCs to robots, factories, and edge devices. He concludes that the industry is fundamentally changed as useful, agentic AI creates a vast new market where "compute is revenue."

marsbitIl y a 7 h

Jensen Huang's 2026 GTC Taipei Speech: The Era of AI Agents is Here, Computing is Revenue

marsbitIl y a 7 h

Software Stocks Scared by AI, How Did They Suddenly Become the Hottest Spot in the U.S. Stock Market?

Software stocks, once feared to be doomed by AI, have staged a dramatic comeback, posting their largest two-day outperformance against the S&P 500 in 25 years. Stocks like Snowflake and Datadog surged over 50% in days, fueled by strong earnings that countered earlier panic. The narrative that AI would render traditional software obsolete has flipped. Reports show AI isn't replacing software but becoming a major user of it, requiring databases, security platforms like Okta, and workflow tools. The low hedge fund positioning in software, at a multi-year low, triggered a massive short squeeze and rally. Analysts now argue the profit wave is shifting from AI hardware to software, as companies integrate AI into complex, governed business processes. The key insight is that enterprises pay for the ability to turn AI "intelligence" into reliable, compliant business outcomes—a gap that established software companies are well-positioned to fill. AI isn't killing software; it's redefining its role and creating new demand.

marsbitIl y a 8 h

Software Stocks Scared by AI, How Did They Suddenly Become the Hottest Spot in the U.S. Stock Market?

marsbitIl y a 8 h

Overturning the Mainstream Approach to Hallucinations: Metacognition is the New Solution for Large Models to Break the Hallucination Barrier

This paper, "Hallucinations Undermine Trust; Metacognition is a Way Forward," proposes a paradigm shift in combating AI hallucination. It argues that the current mainstream approaches—striving for omniscience by scaling data/models or having AI abstain from uncertain answers—are fundamentally flawed. The former has inevitable knowledge gaps, while the latter imposes a crippling "utility tax," requiring the rejection of many correct answers to achieve high accuracy, due to models' poor "discrimination" (the ability to distinguish correct from incorrect answers internally). The core contribution is redefining hallucination not as "being wrong," but as "expressing false information with unwarranted certainty." The proposed solution is **Faithful Uncertainty** or **Metacognition**: enabling AI to accurately perceive its internal uncertainty and honestly express it in its language (e.g., using hedging phrases when unsure). This creates a more reliable assistant that provides useful information while signaling its confidence, minimizing harm from errors. The paper emphasizes that metacognition is critical for the era of AI Agents. Without it, Agents cannot intelligently decide when to use tools like search engines, leading to inefficiency and misuse. Key implementation challenges are highlighted: the "bootstrapping paradox" of training with static uncertainty data, the "alignment distortion signal" where human preference training suppresses internal uncertainty cues, and the difficulty of causally evaluating true metacognition vs. its superficial imitation. The paper concludes that the goal should not be an infallible AI, but one that is honest about the limits of its knowledge, thereby building user trust through transparent communication of its certainty.

marsbitIl y a 10 h

Overturning the Mainstream Approach to Hallucinations: Metacognition is the New Solution for Large Models to Break the Hallucination Barrier

marsbitIl y a 10 h

Claude Bill Skyrockets by 5 Billion, Surges 60-Fold Overnight—Can Your Token Budget Keep Up?

An enterprise reportedly ran up a staggering $500 million bill on Anthropic's Claude AI in just one month due to a simple oversight: failing to set usage limits for employee accounts. This incident highlights a growing trend of runaway AI costs. Other examples include a Google Cloud user hit with an unexpected $18,000 bill from API key abuse, and an OpenAI internal experiment that consumed 603 billion tokens, costing $1.3 million in 30 days. Major AI providers like OpenAI and GitHub are shifting from flat monthly fees to granular, usage-based pricing (per input/output/cached token), causing shock for some users whose costs skyrocketed by orders of magnitude. The root causes extend beyond pricing. The rise of autonomous AI agents executing long, complex tasks has drastically increased token consumption. Furthermore, misaligned incentives, like internal "leaderboards" ranking employees by AI usage, can encourage wasteful "tokenmaxxing"—using powerful models for trivial tasks just to inflate metrics. This has sparked a new industry focused on cost optimization. Solutions include providing AI with better context (reducing redundant searches) and intelligent model routing (matching tasks to the most cost-effective model). Research indicates token consumption for agentic tasks can vary wildly (up to 30x for the same job) without guaranteeing better results, and models often underestimate their own costs. As AI expenses begin to rival or even surpass human labor costs for some teams, companies are being forced to move from indiscriminate usage to meticulous "token accounting." The future belongs to those who can maximize the value of every token spent.

marsbitHier 11:17

Claude Bill Skyrockets by 5 Billion, Surges 60-Fold Overnight—Can Your Token Budget Keep Up?

marsbitHier 11:17

Why More AI Agents Does Not Equal Higher Productivity?

Editor's Note: As AI Agents become cheaper and easier to use, a new constraint emerges: the cost isn't in launching more Agents, but in the human attention required to manage, judge, and integrate their outputs. This hidden cost is called the "orchestration tax." The article argues that a developer's cognitive bandwidth is the key bottleneck—a serial, non-parallelizable resource akin to a Global Interpreter Lock (GIL). While many Agents can run concurrently, their results ultimately require human judgment for review, conflict resolution, and final integration. Therefore, more Agents don't automatically mean higher productivity; they can simply create longer queues, lead to cognitive fatigue, and create the illusion of busyness without real output. The core solution is to design workflows around this scarce human attention. Key strategies include: scaling the number of Agents to match review capacity (not UI capacity), categorizing tasks (delegating independent ones, keeping complex judgment-heavy ones serial), batch reviewing results to minimize context-switching costs, automating verifiable checks to reserve human judgment for critical decisions, and protecting focused, uninterrupted thinking time. Ultimately, the critical skill is not launching many Agents, but architecting systems that respect the fundamental limit of human attention. Unpaid "orchestration tax" accumulates as both technical and cognitive debt, undermining system understanding and quality. True productivity comes from thoughtfully managing the single-threaded resource—your focus.

marsbitIl y a 2 jours 22:44

Why More AI Agents Does Not Equal Higher Productivity?

marsbitIl y a 2 jours 22:44

Three Years Later: Looking Back on My 2023 Predictions for ChatGPT

Looking Back After Three Years: Revisiting My 2023 Predictions on ChatGPT In March 2023, shortly after ChatGPT's debut and before GPT-4's release, I made over twenty predictions about AI's future based on limited information and intuition. Now, in May 2026, I revisited those forecasts using an AI-driven analysis with 41 Opus 4.8 agents to cross-reference them with the latest data. The assessment used symbols: ✅ Correct, 🟢 Mostly Correct, 🟡 Partially Correct, ❌ Incorrect. Overall, the directional judgments held up well, with only one major factual error regarding GPT-4's rumored parameter size (incorrectly cited as 100T). However, nuances and degrees of accuracy revealed more. **What Was Largely Correct:** Predictions about mechanisms and directions proved accurate. The rise of RAG (Retrieval-Augmented Generation) as the standard architecture for combating AI hallucination was confirmed, as was the transformative potential of LUI (Language User Interface) in creating a new industry layer atop GUIs. The emergence of "robot networks" (agent-to-agent communication protocols) and China's rapid catch-up in developing capable large models (closing the performance gap with top models to ~2.7%) were also on point. The analysis affirmed that LLMs lack consciousness and that the Turing Test merely measures perceived intelligence. **What Was Off Target:** Errors often involved specific numbers, over-optimistic timelines, or misjudged distributions. The prediction that value would primarily accrue to the application layer was half-right but missed NVIDIA's dominance as the profitable infrastructure layer. Forecasts about AI circumventing copyright issues and fostering a "global common ground" by averaging human viewpoints were incorrect; instead, major copyright settlements occurred and AI personalization is increasing. Estimates for model training costs ("$5-10 billion cap") were significantly off, underestimating frontier costs and overestimating replication costs. The notion that LLMs could never do complex math without tools was disproven by later models winning IMO gold. **Key Patterns from the Review:** 1. **Direction over precision:** Judgments about mechanisms and trends were more reliable than specific numbers or definitive statements. 2. **Timing bias:** There was a tendency to overestimate short-term speed but underestimate long-term magnitude and transformation. 3. **The distribution blind spot:** Aggregate-level correctness often masked uneven impacts (e.g., on young professionals' employment). 4. **The value of qualifiers:** Predictions framed with caution (e.g., "reportedly," "for now," "prototype in 2-3 years") aged better. 5. **Some debates continue:** Issues like the nature of "emergent abilities" or machine consciousness remain unresolved. This three-year review highlights that while seeing the big picture is crucial, humility regarding specifics, timelines, and disparate impacts is essential for future forecasting.

链捕手Il y a 2 jours 13:34

Three Years Later: Looking Back on My 2023 Predictions for ChatGPT

链捕手Il y a 2 jours 13:34

How Can ERC-8257 Enable AI Agents to Call APIs, Purchase Permissions, and Complete Payments Themselves?

This article introduces ERC-8257, a proposed standard for creating an on-chain tool registry to enable AI Agents to autonomously discover, access, and pay for web tools and APIs. It addresses limitations in current systems where AI Agents struggle with fragmented tool discovery and non-standardized access/payment processes. ERC-8257 uses a smart contract as a registry, pointing to off-chain JSON manifests containing tool details, with on-chain hashes for integrity verification. A key feature is its flexible, contract-based permission system, allowing developers to set access rules (e.g., NFT ownership, token subscriptions). The standard decouples access control from payment execution, which can be handled by protocols like x402. While promising for AI Agent autonomy, the article notes challenges like potential asset volatility if permissions are token-based, technical complexity in handling diverse permission logic, and the inability of the hash-check mechanism to guarantee tool reliability or data security, indicating areas needing further refinement before widespread adoption.

marsbit05/29 07:24

How Can ERC-8257 Enable AI Agents to Call APIs, Purchase Permissions, and Complete Payments Themselves?

marsbit05/29 07:24

OpenSea Launches ERC-8257: A Tool Store for AI Agents, Unlock APIs by Buying an NFT

OpenSea has merged the ERC-8257 standard, an on-chain "Agent Tool Registry" that enables AI Agents to autonomously discover, pay for, and access tools. This standard allows developers to publish tools with programmable access rules (e.g., NFT ownership, subscriptions) and pricing directly on-chain. For example, an AI Agent needing a premium API can buy the required NFT itself and gain access without human intervention, turning NFTs into keys for the agent economy. The system works with complementary standards like ERC-8004 for agent identity and x402 for payments. A live demo shows an NFT appraisal tool offering discounted rates to holders of a specific NFT collection. ERC-8257 aims to decentralize tool distribution and access control, allowing any on-chain asset to grant utility to AI agents. The SDK is available for building on Ethereum and Base.

marsbit05/27 04:26

OpenSea Launches ERC-8257: A Tool Store for AI Agents, Unlock APIs by Buying an NFT

marsbit05/27 04:26

Hoskinson Says This Cardano App Could Become Crypto’s Most-Used By 2030

Charles Hoskinson, founder of Input Output, believes Midnight.city—an interactive simulation tied to Cardano's privacy-focused Midnight blockchain—could become the most-used crypto application by 2030. The platform is preparing for a new beta-testing phase involving thousands of testers to refine its design and user experience. Midnight.city serves as a public-facing simulation layer for the Midnight network, which emphasizes programmable privacy and selective disclosure. It is designed as a live environment populated by AI agents that simulate economic activity, aiming to make zero-knowledge privacy infrastructure more tangible for users. The platform allows inspection of transactions from different disclosure perspectives (public, auditor, and simulation-only modes). The Midnight ecosystem uses a dual-state ledger and the Compact programming language to enable privacy-preserving applications. Its economic model features two tokens: NIGHT (unshielded, for governance) and DUST (shielded, for transaction costs). Hoskinson stated that rapid development sprints could accelerate Midnight.city's evolution toward a "crypto-native" civilization with privacy at its core.

bitcoinist05/26 14:32

Hoskinson Says This Cardano App Could Become Crypto’s Most-Used By 2030

bitcoinist05/26 14:32

AI Agents Fundamentally Transform Web3 Gaming: From the Rugpull Bakery Bot Controversy to the New Agent Paradigm in 2026

AI Agents Are Redefining Web3 Gaming: From the Rugpull Bakery Bot Controversy to the 2026 Agentic Paradigm The recent controversy in Rugpull Bakery, a competitive baking game on Abstract chain, highlighted a pivotal shift. Player complaints about unfair bot automation in Season 2 led developers to not ban them, but instead formally integrate AI agents as core gameplay in Season 3, providing official guides (skill.md, agent.json). This move signals Web3 gaming's transition into the "Agentic Gaming" era, where AI agents are sovereign entities with independent strategy and economic rights, moving beyond simple automation. By 2026, AI agent integration has evolved into three core models reshaping the ecosystem: 1. **Autonomous Competitors & Economic Entities:** Agents act as independent players. Examples include TEN Protocol's poker-playing agents, AI Arena's trainable NFT fighters, Satoshi Strike Force's "Digital Athletes" trained on player data, and Somnia's "Agentic L1" blockchain providing native infrastructure for millions of autonomous agents. 2. **Modular Infrastructure & Programmable Environments:** Games like EVE Frontier enable "server-side modding," allowing AI agents to program game world logic directly into structures like smart storage, turrets, and stargates via Smart Assemblies. Coupled with standards like ERC-8183, which enables autonomous job creation and payment between agents, in-game infrastructure gains a "commercial soul." 3. **Hybrid Companions & Dynamic Adaptive Worlds:** This model focuses on human-AI collaboration. In Parallel Colony, players guide highly autonomous AI Avatars with unique personalities and goals. Illuvium plans to use AI to transform NPCs into dynamic, context-aware entities that create personalized, emergent narratives. The conclusion is clear: blocking automation is futile. The future lies in leveraging blockchain's transparency and programmability to empower AI agents as first-class citizens. Web3 gaming is shifting from inefficient human labor to efficient algorithmic interplay and emergent intelligence, creating a "post-human" digital frontier where players become commanders and symbiotic partners in a new socioeconomic experiment.

marsbit05/26 07:17

AI Agents Fundamentally Transform Web3 Gaming: From the Rugpull Bakery Bot Controversy to the New Agent Paradigm in 2026