Technology Trends

Explores the latest innovations, protocol upgrades, cross-chain solutions, and security mechanisms in the blockchain space. It provides a developer-focused perspective to analyze emerging technological trends and potential breakthroughs.

OpenAI Exposes 'Polaris' Project, the '2028 Great Unemployment' May Really Be Coming

OpenAI has disclosed its "North Star" project, aiming to build a fully automated multi-agent AI research system by 2028. This goal, announced by Chief Scientist Jakub Pachocki, includes launching an "autonomous AI research intern" capable of handling specific research tasks by September 2024. The project represents a major strategic shift, consolidating resources and products like ChatGPT and Codex into a unified super-app. The initiative reflects a broader industry trend, with competitors like Anthropic also advancing AI agent integration into workflows, such as through Claude Code on platforms like Discord. However, Pachocki acknowledges significant challenges in safety and controllability, noting that full control over LLMs remains unresolved. Economically, the project could be highly lucrative, with projections suggesting AI agents—including a $20,000/month "research agent"—generating $29 billion annually by 2029. The core implication is profound: if AI can autonomously conduct research, the pace of scientific and AI advancement could accelerate beyond human limitations, potentially reshaping labor markets and research paradigms.

marsbit03/24 02:46

OpenAI Exposes 'Polaris' Project, the '2028 Great Unemployment' May Really Be Coming

marsbit03/24 02:46

AI Influencer Zhang Zala's Secret to Gaining 50,000 Followers: Algorithm as Network, Beware of AI Alienation...

AI influencer Zhang Zala shares strategies for growing to 50k followers on X (formerly Twitter), emphasizing treating the platform as a "party" rather than a "stage." Key tactics include: curating the algorithm feed to prioritize high-quality AI and tech discussions; engaging authentically in replies without AI-generated responses; posting consistently (3+ times daily) through public learning, product feedback, and demo sharing; and avoiding AI-assisted writing until developing a unique voice. Underlying factors include her pre-existing Silicon Valley AI network and prior follower base from Chinese social platform Xiaohongshu. Authenticity and providing value are central to sustainable growth.

marsbit03/24 00:42

AI Influencer Zhang Zala's Secret to Gaining 50,000 Followers: Algorithm as Network, Beware of AI Alienation...

marsbit03/24 00:42

From "Global Computer/Settlement Layer" to "Bulletin Board": What Are Ethereum and Vitalik Trying to Achieve?

From "World Computer" to "Public Bulletin Board": Ethereum's New Vision For years, Ethereum has been widely viewed as a "world computer" or "global settlement layer," executing smart contracts and powering DeFi and NFTs. However, Vitalik Buterin recently proposed a fundamental shift in perspective: Ethereum's core value may not be its smart contract functionality, but rather a simpler primitive—a cryptographically secure, globally shared "public bulletin board." This "bulletin board" represents a neutral, uncensorable data availability layer. Anyone can read or write data to it, but no single entity—not even a government or Vitalik himself—can alter or erase it. This makes it ideal for applications like secure online voting (where votes are verifiable yet private), certificate revocation lists, and decentralized governance, where the need is not for complex computation but for immutable, publicly verifiable data storage. Technically, upgrades like EIP-4844 (Blob data) and the upcoming PeerDAS are expanding this "board's" capacity, aiming to make Ethereum the highest-security data attestation layer. The rise of AI further underscores the need for such a system. Current AI services tie user queries to real identities. Proposals like ZK API Usage Credits suggest using Ethereum and zero-knowledge proofs to enable anonymous AI API calls, where users prove they have usage credits without revealing who they are. Additionally, AI agents lack legal identities; Ethereum can serve as their economic coordination layer for transactions, staking, and reputation. This repositioning from "world computer" to "bulletin board" is not a reduction in ambition but a pragmatic evolution. It shifts the focus from what the technology can do to what the world actually needs—a foundational, trustless data infrastructure. Like TCP/IP for the internet, Ethereum aims to be the indispensable base layer upon which essential applications, especially in the AI era, can be built. Its ultimate value lies in providing a permanent, tamper-proof record of truth.

marsbit03/23 15:37

From "Global Computer/Settlement Layer" to "Bulletin Board": What Are Ethereum and Vitalik Trying to Achieve?

marsbit03/23 15:37

From OpenClaw to the History of the Web: When AI Gains Sovereignty, What Remains for Humanity?

From Web1 to Web4: A History of Power and Ownership in the Digital Age This article examines the evolution of the web not as a series of technical upgrades, but as a fundamental shift in power—specifically, who owns data, controls wealth, and wields productive force. **Web1 (Read-Only):** Characterized by one-way communication. Platforms like Yahoo owned all content and users were merely passive consumers, or "traffic," with no digital assets. **Web2 (Read-Write):** Users became content creators, but platforms like Facebook and TikTok established a "panoptic dictatorship." They harvested user data to create immense value, but users retained only usage rights, not ownership, of their digital assets and social presence. **Web3 (Read-Write-Own):** A movement to reclaim digital rights through cryptography and decentralization. It enables true digital ownership (e.g., via private keys) and trustless systems (e.g., DAOs, smart contracts). However, it remains a wild frontier with significant legal and security challenges, lacking a capable "workforce" to realize its full potential. **Web4 (Agent Economy):** The convergence of AI Agents and Crypto. AI Agents (autonomous, task-completing AIs) use Crypto as their native currency for machine-to-machine transactions. This shifts power from humans to algorithms, creating independent AI economic actors. This raises critical legal questions, such as liability for AI errors. The future could lead to two extremes: a utopia of liberated human creativity or a dystopia of extreme inequality if AI power is monopolized by a few. **Survival Guide for Web4:** * **Work:** Become a director and risk-manager for AI, not an executor. * **Invest:** Focus on projects with genuine utility, not hype-driven "air tokens." * **Risk Management:** Prioritize robust legal and compliance frameworks for AI operations. The conclusion emphasizes that understanding the transfer of power and assets is key to navigating the future, urging innovation within the boundaries of regulation.

marsbit03/23 13:32

From OpenClaw to the History of the Web: When AI Gains Sovereignty, What Remains for Humanity?

marsbit03/23 13:32

Karpathy Diagnosed with "AI Psychosis"! Not Eating or Sleeping, 16 Hours a Day Raising Lobsters

Andrej Karpathy recently revealed that he has developed what he calls "AI psychosis," an obsessive state where he spends up to 16 hours a day directing AI agents instead of writing code himself. In a podcast with Sarah Guo, he explained that his workflow has shifted from 80% hand-coding and 20% AI-assisted to the reverse, or even more extreme. He now manages multiple AI agents simultaneously, treating them as a team to execute tasks. Karpathy admitted that he’s become addicted to optimizing AI performance, constantly worrying about whether he’s using tokens efficiently or pushing the system to its limit. He highlighted the importance of an agent’s “personality,” noting that Claude Code feels more like a collaborative teammate compared to colder, more mechanical alternatives. He also shared practical applications, such as "Dobby," a Claude-based smart home agent that integrates and controls all his home devices through natural language, replacing six separate apps. In research, his "AutoResearch" project used AI to run 700 experiments, resulting in an 11% training speed improvement for an AI model—discovering optimizations he had missed as a human researcher. Despite the capabilities, Karpathy noted that AI agents still exhibit uneven performance—sometimes brilliant, other times childlike—due to limitations in reinforcement training. He predicts that 2026 will see a "slopacolypse," with AI generating vast amounts of mediocre content. His experience signals a broader shift: humans are becoming directors of AI systems rather than executors, navigating a new era of human-AI collaboration.

marsbit03/23 11:44

Karpathy Diagnosed with "AI Psychosis"! Not Eating or Sleeping, 16 Hours a Day Raising Lobsters

marsbit03/23 11:44

AI Wealth Tutorial: Start with NSFW, Then Sell Courses

The article "AI致富教程：先搞色色，再去卖课" (AI Money-Making Guide: Start with Adult Content, Then Sell Courses) explores how AI-generated content (AIGC) is being monetized, particularly through adult entertainment and low-barrier creative work, before ultimately shifting to selling instructional courses. A16Z’s report highlights a striking trend: in the U.S., user spending on OnlyFans surpassed combined spending on OpenAI and The New York Times. This reflects a broader pattern where “sexual appeal outperforms productivity.” Early adopters used tools like Midjourney and Stable Diffusion to create AI-generated virtual models, offering “girlfriend experiences” on platforms like Fanvue, where AI models now contribute significantly to revenue. Similarly, some turned to AI-generated children’s books, though market saturation and quality issues quickly diminished profitability. Both paths often lead to selling courses—packaging the “get-rich-quick” illusion to newcomers. However, the real barrier isn’t technical proficiency but aesthetic judgment: the ability to translate vague ideas into precise prompts. Those with design, photography, or writing backgrounds excel because they know what “good” looks like; others struggle even with advanced tools. The rise of AI also brings ethical and trust issues. Clients often reject AI-assisted work on principle, perceiving it as “unfair” or lacking human effort. Regulations now require AI-generated content labeling, but boundaries remain unclear—especially for hybrid human-AI creations. The core question isn’t just whether AI was used, but whether someone is genuinely accountable for the output. In summary, while AI lowers entry barriers for content creation, success still hinges on traditional skills like审美 (aesthetic sense), and the real money often moves from creating content to selling the dream of easy success.

marsbit03/23 10:52

AI Wealth Tutorial: Start with NSFW, Then Sell Courses

marsbit03/23 10:52

People Laid Off by AI Won't Disappear; They Will Become the Creators of the Next Economy

The article argues that the real question surrounding AI is not whether it will cause unemployment, but what happens to the people displaced. AI is replacing not humans, but the standardized, replicable, and automatable parts of human work. This follows historical patterns where technological revolutions, from stone tools to computers, made old skills obsolete and dissolved old structures—but humanity adapted and reorganized. The author draws a parallel to China’s large-scale layoffs during state-owned enterprise reforms 30 years ago, which initially seemed catastrophic but eventually fueled the growth of a new private economy, new companies, and new types of jobs. Engineers, though among the first impacted, are also positioned to recover fastest. Their systemic understanding and proximity to new productive forces make them ideal candidates to adapt and create in the new economy. More importantly, AI is reshaping companies themselves—reducing organizational bloat, communication costs, and bureaucracy. This enables smaller, more agile teams and empowers strong creators who may have previously struggled with management rather than innovation. The core issue is not job loss, but self-definition: will individuals wait to be reassigned by the old system, or use new tools to reorganize production? AI accelerates differentiation—eliminating some jobs, shattering illusions for some, and offering others a chance to leap forward. The author’s view is that AI is dismantling an entire generation’s belief in stable career paths. Those laid off won’t vanish; instead, many will reinvent themselves—transitioning from employees in old systems to creators of the next economy. Every productivity revolution淘汰 (eliminates) not people, but those who refuse to rewrite themselves. The first to accept this and start building the new world will succeed.

marsbit03/23 10:31

People Laid Off by AI Won't Disappear; They Will Become the Creators of the Next Economy

marsbit03/23 10:31

Token Doesn't Need a Chinese Name, But the Business Behind It Does

Recent discussions in China have intensified around finding an appropriate Chinese translation for the technical term "Token," driven by its growing economic and industrial significance. Previously an obscure technical term within AI circles, Token has now entered mainstream discourse due to its role as a billing unit in cloud services, a revenue metric for AI companies, and a key indicator in national AI industry statistics. Proposed translations include "智元" (suggested by AI media, implying "intelligence unit"), "模元" (proposed by academics, leaning toward "model unit"), and "符元" (a more neutral, technical term meaning "symbol unit"). The debate is not merely linguistic but reflects broader commercial and narrative control over the AI industry. Different translations align with different stakeholders’ interests: "智元" benefits those emphasizing intelligent computation, while "模元" reinforces the role of model developers. The term already had an academic translation—“词元” (ciyuan)—since 2021, but it gained little attention until Tokens became a valuable economic unit. As Token consumption in China surges—reaching 180 trillion per day—the naming contest underscores deeper issues of market influence, branding, and “coinage” rights in the emerging AI-driven economy. Ultimately, those who produce Tokens may hold the power to define them, regardless of the chosen name.

marsbit03/23 08:48

Token Doesn't Need a Chinese Name, But the Business Behind It Does

marsbit03/23 08:48

AI Agent Outputs Garbage? The Problem Is You're Not Willing to Burn Enough Tokens

The core argument is that the quality of an AI Agent's output is directly proportional to the number of tokens invested in the process. More tokens lead to fewer errors, as they allow for deeper reasoning, multiple independent attempts, self-critique from fresh contexts, and verification through testing. This approach can solve problems of scale and complexity but fails when facing novel problems not present in the model's training data. For such novel challenges, human domain knowledge and guidance are essential. Two practical, immediate solutions are proposed: implementing an automatic review cycle (WAIT) for the Agent to repeatedly critique and fix its work, and establishing frequent verification checkpoints (VERIFY) where a separate Agent validates outputs to catch errors early. The key takeaway is that insufficient token investment is often the primary reason for poor Agent performance, not the underlying framework.

marsbit03/23 06:14

AI Agent Outputs Garbage? The Problem Is You're Not Willing to Burn Enough Tokens

marsbit03/23 06:14

10 Cold Takes on WeChat's Integration of OpenClaw

WeChat's recent integration of OpenClaw as a plugin has sparked discussions, but its immediate impact may be overestimated. The feature allows users to connect their existing OpenClaw agents to WeChat via a simple setup, but it comes with limitations: no group chat support, no streaming output, and no multi-agent connections. The move is seen as a necessary, though delayed, step rather than a groundbreaking change. WeChat’s approach is elegant—acting as a platform rather than building its own agent, thus supporting various OpenClaw-compatible models. However, the integration doesn’t lower the barrier to effectively using AI agents; it only simplifies access. The real challenge remains in configuring and leveraging agents meaningfully. Long-term, this signals WeChat’s shift from connecting people to also connecting humans and AIs, paving the way for future native AI-agent integrations within WeChat’s ecosystem. But for now, it’s a cautious, incremental update—typical of WeChat’s product philosophy.

marsbit03/23 01:10

10 Cold Takes on WeChat's Integration of OpenClaw

marsbit03/23 01:10

活动图片

Популярные категорииright-arrow

Популярные тегиright-arrow

Project Updates