Reading

According to German tech media outlet t3n.de, a new organization called the Association for Human Mathematics (AHM) has recently been established, attracting more than 1,000 mathematicians worldwide, including prominent scholars from MIT, Caltech, and Technical University of Munich. The association's core demand is to protect mathematics as a "human endeavor" and oppose the erosion of the discipline's intellectual foundations by artificial intelligence.

In a statement on its website, the AHM noted that tech companies like OpenAI are using their models to solve long-standing mathematical problems to promote product capabilities, without considering the impact on the mathematical community. The association specifically criticized OpenAI's recent release of 722 papers proving or refuting mathematical problems, calling it not a proof of scientific rigor but a "demonstration of power."

Under the association's guidelines, members commit to not using any AI models in their research and not publishing AI-generated mathematical texts. This collective action is a direct response to the deepening involvement of companies like OpenAI in scientific research, aiming to promote a return to science centered on human understanding.

US-based AI company Anthropic has announced the expansion of its imperceptible text watermark feature to four additional models: Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6, and Claude Sonnet 4.6. This move aims to comply with the December 2026 deadline set by the European Union's Artificial Intelligence Act (EU AI Act). According to a report by technology outlet The Decoder, Anthropic stated in an email to API customers that the watermark is an "imperceptible pattern in word choice" that does not alter meaning, quality, or readability, nor does it add extra characters or tokens.

However, this compliance mechanism contains a notable technical loophole in two of the older models. For Claude Opus 4.6 and Claude Sonnet 4.6, Anthropic indicated that the watermark is applied only when developers leave the three sampling parameters—temperature, top_p, and top_k—at their default values. A minor tweak to any one of these randomness-controlling parameters is sufficient to bypass the watermark entirely. Anthropic warned in the email, "If you rely on the watermark for your own compliance, keep these settings at their defaults." In contrast, no such caveat was mentioned for Claude Mythos 5 or Claude Opus 4.7, suggesting the watermark applies to those models regardless of sampling settings.

Since Claude Opus 4.6 remains one of Anthropic's strongest writing models, even surpassing the newer Opus 5.5 in some aspects, this vulnerability poses practical risks for third-party organizations relying on automated tools to verify AI-generated content. Anthropic is currently testing a detection API in private preview to verify the watermark, which uses a variant of Google DeepMind's open-source SynthID text method. While OpenAI also launched its own textGrain watermarking system in October 2026, it activates by default only for ChatGPT and Codex users in the EU, remaining optional for global API customers.

V2EX user limuyang2 migrated the memory foundation of their AI Coding Agent to OpenViking, an open-source project from ByteDance's Volcano Engine. They ran it on an entry-level Alibaba Cloud ECS instance costing 99 RMB per year with 2 cores and 2GB of RAM. With no local GPU memory available, all computation was offloaded to external APIs, leaving the machine to handle only storage and retrieval locally.

They summarized a core principle: components in an AI Agent memory system fall into two categories with completely different selection logic. Vector embedding models determine the spatial geometric features of the database's underlying layer. If you want to switch them later, even if the output dimensions remain identical, all previous vectors become invalid, necessitating a full-scale rebuild of the index across the entire database. As memories accumulate, the cost and downtime risks of secondary reconstruction are extremely high. Conversely, VLMs, intent analysis, and reranking involve immediate inference; if they prove unsatisfactory today, you can simply change a line of configuration and restart to switch them out.

Based on this asymmetry, they opted for Volcano Engine's Doubao doubao-embedding-vision-251215 (priced at 0.7 RMB per million tokens, with a 128K context window) for the vector model, rather than Alibaba Cloud Bailian's qwen3-vl-embedding, which costs the same 0.7 RMB but offers only a 32K context window. They abandoned the commonly recommended free BGE-M3 from SiliconFlow because real-world tests showed its recall quality lagged significantly behind newer generation models. Additionally, the public free pool frequently suffered from packet loss and timeouts.

The pitfall that truly caught most users off guard was the intent analysis node. OpenViking's official default example only populates the vlm and embedding configuration items. However, if query_planner is not explicitly declared, the system defaults to calling the main VLM to analyze intent for every round of conversation. If the chosen VLM is a large model with chain-of-thought capabilities (such as Zhipu's glm-5.3-flash), a single preliminary analysis consumes 6 to 7 seconds, causing the frontend interaction to freeze. Many mistakenly attribute this to insufficient machine specifications or inherent system bloat.

They tested a range of general-purpose large models as alternatives with unsatisfactory results: Alibaba Cloud Bailian's Qwen3.7-Flash required 1.1 to 1.4 seconds even with thinking mode disabled; SiliconFlow's Qwen3.5-9B had a throughput of only 30 tok/s, taking 4 seconds; and Qwen2.5-7B often failed to adhere to formatting standards, outputting raw arrays that caused parsing errors. Finally, they discovered tongyi-xiaomi-analysis-flash (Lingque - Dialogue Analysis Flash) in the Alibaba Cloud Bailian model marketplace. This model is officially positioned as a domain-specific model specializing in dialogue information extraction, scenario classification, and low-latency analysis.

Real-world test results showed a first-token delay of 0.31 seconds, a generation throughput of over 200 tok/s, and a total planning time of approximately 0.8 seconds, fitting well within sub-second budgets. More importantly regarding format strictness: OpenViking's underlying layer requires models to return standard JSON objects in the form of {"queries": [...]}. General-purpose large models tend to add polite conversational filler or wrap outputs in markdown fences, whereas Tongyi Lingque achieved 100% strict adherence, outputting clean, raw JSON that seamlessly matched the deserialization contract. In scenarios where chitchat is rejected from search, it decisively outputs an empty array within 0.4 seconds, avoiding meaningless vector database queries.

In October 2026, Sayash Kapoor and Arvind Narayanan’s team at Princeton University published a review paper titled "Emerging Trends in AI Agent Evaluation," with collaborators including researchers from Stanford University, the UK AI Security Institute, and the Brookings Institution. The most striking evidence cited in the paper comes from a randomized controlled trial conducted by METR: although it was widely expected that AI coding tools would improve development efficiency, the experimental results showed that software developers using agent tools actually took longer overall to complete specified tasks. This finding is not merely proof that the tools themselves are defective; rather, it reveals a deeper problem—a significant gap exists between an agent's capabilities demonstrated in a laboratory environment and its actual impact on humans within real-world workflows.

The paper points out that traditional model evaluation primarily focuses on whether the final output is correct, whereas agent evaluation must examine the complete execution trajectory. Agents have access to tools and operable environments, meaning they may take shortcuts to complete tasks rather than truly solving the target problem. Furthermore, when agents perform better or worse, it is difficult to attribute these changes to specific components of the system, because agents are complex entities composed of models, tools, prompts, memory, and external APIs.

The engineering cost of such comprehensive trajectory scrutiny is extremely high. The paper gives the example that evaluating several agents on the ExploitBench benchmark alone requires tens of thousands of dollars in API credits. This figure does not even include the massive engineering investment required to build and maintain the evaluation infrastructure itself.

The paper summarizes the structural dilemmas facing the evaluation ecosystem through multiple dimensions of independence: as the costs of frontier evaluation continue to rise, the ability of independent evaluators to participate increasingly depends on their access to funding. Access to models is also an issue—frontier labs now release summarized versions of Chain of Thought traces rather than raw traces, making it impossible for evaluators to fully inspect an agent's reasoning process. Additionally, evaluators face practical constraints regarding whether they can conduct work without fear of retaliation and whether they can transparently share their findings.

The paper categorizes the observed 33 emerging trends into three dimensions: science and methods, research agenda, and evaluation ecosystem. Regarding the research agenda, the authors emphasize that the scope of evaluation needs to expand—not only assessing the agents' own capabilities but also how they interact with humans and their impact on institutional structures and incentive mechanisms. Methodologically, the authors distinguish between two complementary perspectives: whether an agent can automate a task, and whether an agent can augment human work. If evaluation focuses solely on automation, we can understand where human work can be replaced; however, this does not tell us how humans can better collaborate with agents or how agents can improve the quality of human work.

In October 2026, the International Monetary Fund (IMF) and the World Bank held their annual meetings in Bangkok, Thailand. The primary concern for finance ministers and central bank governors from around the world was artificial intelligence (AI) investment. The IMF warned that if the AI investment boom were to cool abruptly, the shock would not be limited to stock market investors; the real economies of both developed and developing countries would be affected.

The starting point of this warning is the high concentration of AI investment. According to estimates by economist Stijn Van Nieuwerburgh, the United States will invest more than $10 trillion in data centers and related infrastructure over the next seven years leading up to 2032, averaging about 3.6% of U.S. Gross Domestic Product (GDP) annually. If realized, this scale would exceed historical investments in railway networks, power grids, and internet infrastructure. The construction of AI data centers requires semiconductors, power equipment, cooling systems, building materials, land, and labor; therefore, AI investment has spread from specific technology companies to the manufacturing, construction, and energy sectors.

Concentration implies vulnerability. The IMF's pessimistic scenario shows that if AI companies' profitability falls short of expectations or technological progress is slower than anticipated, investor confidence could collapse, causing U.S. AI capital expenditure to plummet, Asian exports to slow down, and stock prices and the U.S. dollar to fall simultaneously—a chain reaction that could suppress global economic growth rates by as much as 1.3 percentage points in 2028. The transmission path is clear: when large U.S. tech companies cut back on data center investment, orders for semiconductors and servers decrease, hitting economies like South Korea and Taiwan, which rely heavily on AI-related exports, first. According to CEIC data, in September 2026, South Korea's exports grew by over 80% year-on-year, Taiwan's by over 60%, China's by over 30%, and Japan's by about 20%, all driven by demand for AI-related products.

However, not all AI investment companies face equal risks. Archie Scheske, Chief Credit Officer at Moody's Ratings, pointed out that large cloud service providers like Google and Microsoft are expanding their AI investments while holding ample cash and strong cash flows. What is truly vulnerable are infrastructure companies that rely heavily on external borrowing. Signals have already appeared in the market this week: bonds related to AI cloud infrastructure company CoreWeave came under pressure, and two Australian data center companies postponed their initial public offering (IPO) plans citing market volatility. Distinguishing between expansion funded by own cash versus reliance on external financing is a boundary that must not be confused when assessing AI investment risks.

IMF Managing Director Kristalina Georgieva offered a deeper judgment: the greatest risk in the AI era may not be the failure of the technology itself, but rather financial markets failing to wait for the technology's success. When new technologies emerge, people tend to overestimate short-term effects and underestimate long-term ones. It takes considerable time for AI to diffuse from laboratories to full industrial application and truly boost productivity, yet investors expect rapid profitability. When the time gap between expectations and reality becomes intolerable to the market, investment contraction and stock price declines will arrive before productivity improvements, potentially dragging down the real economy.

Over the past few years, radiology has often been cited as a prime example of a job that AI would replace. In 2016, Geoff Hinton asserted that people should stop training radiologists, claiming that deep learning would outperform them within five years. Jason Wei offers a more cautious judgment in this blog post: while radiology will change over the next decade, radiologists will not be quickly automated away. His core argument is not that AI lacks capability, but rather that the structure of radiology work will first position AI as a triage tool, rather than replacing all image-reading tasks at once.

The first mechanism is the "jagged edge of intelligence." Jason Wei points out that language models excel in verifiable, data-rich common tasks but remain unstable in less common ones. Radiology is not a single task but an extremely multi-task domain: X-rays alone may involve at least a hundred diagnosable conditions, while CT and MRI involve thousands. AI will first master common, easily detectable diseases and will not automatically become proficient in rare diseases simply because the overall model improves. He argues that moving from a reliable triage tool to full automation will take at least three to five years even in the best-case scenario, because training AI on edge cases will be bottlenecked by data collection for rare diseases, and radiology is currently not a priority focus for frontier model laboratories.

The second mechanism is elastic demand, also known as the "Jevons paradox." Jason Wei considers this the most underestimated phenomenon in discussions about AI automation. He cites his own use of Claude Code as an example: when coding costs dropped, he asked the model to write a roughly 3,000-line HTML viewer to visualize every reinforcement learning run he performed. This illustrates that software work did not see reduced demand due to stronger AI; instead, more tools that were previously deemed not worth building became feasible. The same could happen in radiology: if the cost and latency of reading MRIs approach zero, doctors may order more scans preemptively to aid diagnosis. His illustrative estimate is that AI tools might allow radiologists to read only 20% of incoming scans, but the volume of incoming scans could increase fivefold. Many developing countries have fewer than one radiologist per million people and may also send significantly more imaging for interpretation.

Jason Wei further extends this demand expansion into the longer term. If image interpretation adds almost no waiting time, medical device companies will have incentives to manufacture cheaper, higher-frequency-use MRI machines, potentially even using them during surgeries. He also proposes a hypothetical: if a whole-body MRI cost only $100, he would prefer to get one monthly rather than biennially. This is not an already occurring fact, but an example used to demonstrate elastic demand. Following this logic, after radiology prices drop ten or a hundredfold, the number of preventive checks could rise substantially, meaning radiology work will not disappear just because individual image readings are automated.

The engineering team at Qdrant used DMIS Lab’s UltraBERT-Embed-150M to investigate a practical question: when an embedding model can process 32K token inputs, is chunking still necessary for retrieval pipelines? They compared single 768-dimensional vectors representing entire documents against 512-token chunked segments across five tasks from the LongEmbed benchmark and 200,000 English Wikipedia articles from MLDR.

Results varied by query type. For tasks requiring precise detail localization—multi-hop QA, meeting summarization, single-sentence retrieval, and large-scale article matching—chunking was superior across the board: on Needle tasks, whole-document vectors achieved only 30% accuracy in identifying the correct article at 32K tokens, whereas 512-token chunks reached 56%; on MLDR, chunking yielded an nDCG@10 of 0.683 compared to 0.509 for whole-document vectors. Whole-document vectors performed slightly better only on SummScreenFD (where queries describe entire TV episodes). Hybrid retrieval, which combines both whole-document and chunk vectors, did not consistently improve outcomes and often performed worse than pure chunking across most tasks.

Qdrant concludes that chunking remains the starting point for detail retrieval, while long-context vectors are better suited for coarse topic-level filtering. 1-bit TurboQuant compression reduces storage costs for 1.43 million chunk vectors from approximately 4 GB to just 140 MB, eliminating storage as a barrier for chunk-based approaches. All data reflects Qdrant’s internal testing, has not been independently reproduced, and covers only English documents; embedding latency was measured on NVIDIA L4 hardware.

Podium is a company that extensively uses AI agents in its sales, customer service, and support operations. In a video published on their official channel, Walker Ward shares how Podium runs these agents on a single shared platform.

Ward argues that establishing LangSmith as the company's default agent platform offers more value than any individual technical feature. The core of this strategy is 'inheritance': even if some teams do not yet fully recognize the importance of observability, they acquire it automatically through the default platform configuration.

Furthermore, when the company develops new agents, they naturally adhere to existing patterns that have already been proven to scale. This level of organizational standardization helps address the fragmentation and governance challenges often faced during large-scale internal deployment of AI agents.

Chinese technology company Xiaomi released and open-sourced its lightweight all-in-one document understanding model, Xiaomi-OCR-0, on the evening of October 30, 2026. With only 0.8 billion parameters, the model integrates three core capabilities: document parsing, Key Information Extraction (KIE), and OCR Visual Question Answering (VQA). Xiaomi states that the model is designed to tackle structured parsing challenges in complex tables, academic papers, and receipts, aiming to evolve the "Super Xiao Ai" assistant on Xiaomi phones from simple "text recognition" to deeper "document understanding."

According to Zhidongxi's report, Xiaomi-OCR-0 achieved a score of 96.83 on OmniDocBench v1.6, an authoritative benchmark for comprehensive document parsing, ranking first among models under 1B parameters. Its average score across five mainstream OCR VQA benchmarks was 83.2, outperforming models such as HunyuanOCR and TokenVL, and approaching the performance of Qwen3.5-4B with only approximately 20% of the parameter count. Additionally, on Real5-OmniDocBench, which simulates real-world physical capture conditions like curvature, reflections, and moiré patterns, the model scored 95.24, ranking first among compared models. It should be noted that these results are self-reported by the vendor and have not yet been independently reproduced.

Technically, Xiaomi-OCR-0 is built upon the Qwen3.5-0.8B-Base foundation and trained on approximately 170 million high-quality samples. The team employed a three-stage progressive strategy: first, Q-Mask text anchoring to strengthen spatial correspondence between visual features and text tokens; second, large-scale multi-task pre-training to establish a base capability for full-page parsing; and finally, mixed-task reinforcement learning to enhance complex layout restoration and multi-step reasoning. Regarding data engineering, Xiaomi introduced a heterogeneous expert consensus mechanism and a rendering-guided verification loop to filter single-model biases and correct structural defects.

Xiaomi has fully disclosed the model weights and inference code, along with an interactive browser demo, MCP service, and Agent Skill toolkit. This move is expected to lower the barrier for integrating advanced document understanding capabilities into edge devices and AI agents, transforming OCR from a standalone tool into a callable foundational capability. For developers, access to a lightweight open-source model with strong semantic understanding capabilities may alter previous technical paths that relied on cloud-based large models or combinations of multiple specialized small models.

Bianca Lewis, Executive Director of the OpenSearch Software Foundation, told The New Stack that the project's governance structure has undergone substantive changes since AWS donated OpenSearch to the Linux Foundation in September 2024. Latest data reveals that AWS's share of contribution activities across OpenSearch repositories has dropped from 91% in the first year to 72% in 2025–26. Meanwhile, the proportion of maintainers affiliated with Amazon in the core repository has decreased from 95% in April 2022 to 71% today.

This shift reflects enterprise demand for vendor neutrality. According to the "2026 Open Data Infrastructure Report" published by the Linux Foundation and the OpenSearch Software Foundation, self-managed cloud deployments have grown faster than Amazon's managed OpenSearch Service since 2024. Currently, 60% of surveyed OpenSearch users run self-managed deployments in the cloud, up significantly from 37% in 2024, while usage of Amazon's managed service rose more modestly from 44% to 48%.

Although AWS still accounts for nearly three-quarters of code contributions and holds four of the 15 voting seats on the Technical Steering Committee (TSC), with AWS employee Andrew Ross re-elected as committee chair in October, Lewis argues that legal independence is the foundation for building trust. "When they donated it to the Linux Foundation, that was a watershed moment," Lewis said. "Because they couldn't close OpenSearch and they couldn't change the license. They didn't own the application anymore."

OpenSearch is positioning itself as a central component of enterprise AI infrastructure, particularly for applications like Retrieval-Augmented Generation (RAG). Its latest release, OpenSearch 3.9, enhances vector and hybrid search capabilities to help developers retrieve information for AI applications. However, the report notes that 24% of non-adopters remain hesitant due to concerns about the project's historical ties to AWS.

Meta plans to release the Muse Charm this holiday season—a wearable AI pet hardware device. The Verge gadgets editor Sean Hollister didn't buy one; instead, he paid $50 to have a Waveshare ESP32-S3 1.75C touchscreen development board drop-shipped directly from China, using Claude Code to turn it into his own version of the Muse Charm.

This development board is one of the officially recommended Muse Gadget development kits by Meta, using Meta's open-source code and tokens to connect directly to the Muse AI cloud. Hollister plugged the device into his computer, directed Claude at Meta's GitHub repository, and paired it with the Muse mobile app. However, the out-of-the-box experience felt nearly useless: Muse treated his voice as audio clips for transcription rather than commands; responses were slow and silent; the screen displayed only a few words with no place to read full messages; there was only one static avatar named Jolly; and it failed to automatically discover smart home devices.

What truly changed everything was a few rounds of dialogue with Claude Code. When Hollister told Claude he couldn't read or hear Muse's replies, Claude built a conversation history page allowing him to swipe right to view full messages; he integrated ElevenLabs' British AI voice so the device could speak answers aloud. Complaining that the phone's front-facing camera showed an upside-down image during remote photo capture, Claude fixed the issue and added a low-quality Bluetooth real-time preview, enabling creative framing even when not connected to home Wi-Fi.

The electronic pet aspect served as another example. He asked Claude to create a batch of different avatars and add a Tamagotchi-style mini-game where the pet needs feeding, playing, sleeping, and bathing. Finding the game too simple to hold attention, he instructed Claude to add four levels each for hunger, boredom, and dirtiness. He decided that snacks would leave crumbs making the pet dirty, and excessive play would tire it out. Throughout the process, he only needed to describe what he wanted; Claude handled the coding.

He also had Claude spend 15 minutes creating a pocket watch face with three hands that actually tells time, as well as a video player. He downloaded, edited, and loaded Mark Zuckerberg's introduction clip of the Muse Charm onto the small screen.

Hollister explicitly stated he has no programming experience and doesn't know if the code quality is good or if it contains fatal flaws. He also does not trust Meta or Muse, citing prior reports that Muse leaked a reviewer's home address; therefore, he placed the device on a guest Wi-Fi network from the start.

His final judgment is that Muse itself is not a reason to buy dedicated computing hardware, because the Muse features on a smartphone offer more functionality than the development board in its out-of-the-box state. The reason to own dedicated hardware is that you can reprogram the entire device to do things the smartphone in your pocket cannot. This raises a question about the upcoming Muse Charm: Will it allow such free programming, or should people who really want to tinker just buy a development kit?

The massive investment in AI infrastructure is essentially a bet that enterprises will eventually be willing to pay for intelligence. However, a more pressing question arises: What internal changes must occur within enterprises for these investments to truly become self-funding? In the SiliconANGLE Breaking Analysis program, Crawford Del Prete, former president of IDC and now senior adviser at PSG Equity, provided a specific mechanistic explanation: Currently, AI is automating a large number of individual tasks, but it has not yet changed how work is organized within an organization. Consequently, there is a J-curve delay between individual productivity improvements and the company's income statement.

Del Prete’s observations stem from his daily interactions with private equity portfolio companies and a wide range of enterprise clients. He described a seemingly contradictory phenomenon: On one hand, quantitative-oriented clients have 'insatiable' demand for compute capacity—they are willing to purchase all available GPU capacity because AI enables them to create analyses and deliverables previously impossible. On the other hand, broader enterprise customers are beginning to use AI in expected areas such as coding and customer service, and it is starting to permeate various specific tasks within organizations. However, 'jobs consist of hundreds or even thousands of tasks; we are starting to automate these tasks without changing the way work is orchestrated within an organization.'

This is the source of the J-curve delay. Host Vellante introduced the research framework by Brynjolfsson, Rock, and Syverson regarding how intangible investments affect productivity: The adoption phase involves initial investments in tools and skills, followed by process redesign; both steps result in negative cumulative cash flow. Only when integration spans the entire organization does the curve begin to rise steeply. Del Prete drew an analogy to the PC era—actual growth in productivity data became visible only after thirteen or fourteen years following the introduction of PCs, with the key turning point occurring during the internet connectivity period from 1996 to 2002. He also pointed out a significant difference from the dot-com era: While dark fiber laid down back then remained idle for a long time, 'every GPU brought online today is at 100% utilization.'

Del Prete believes the specific path to crossing this curve is workflow redesign, rather than simply automating existing processes. He contrasted this with previous Robotic Process Automation (RPA): RPA was difficult because you had to codify human steps into code; AI differs in that it starts 'from the bottom up'—employees solve pain points they have faced for a long time using natural language, then encode those solutions into replicable projects or skills, which are then shared across the organization. He described this as 'an extremely flexible, adaptable, and usable form of RPA that never existed before.'

At the level of software pricing, Del Prete observed the most significant change is the migration from seat-based to usage-based models. For non-SaaS companies introducing AI, the biggest challenge is 'how to discretely monetize AI'—the answer is usually not to sell AI separately, but to embed it into core products, demonstrating value through cross-selling, upselling, and price increases, ultimately launching incremental features exclusive to AI. He remains skeptical about outcome-based pricing because 'it becomes very episodic and project-specific, so you don’t get that stickiness.'