LesenForschungRadarAnlageframework
Anmelden / Registrieren
AlleTechnikProdukteInvestierenBörseSonstiges

Lesen

2026-11-0585 Beiträge

Aktualisieren

Baidu Open-Sources ERNIE 5.5 Flash Base: Claims SOTA Among Same-Size Models with New Sparse Attention

Wesentlich
2026-11-05 15:15 GMT+8

On November 5, Baidu officially open-sourced its latest base model, ERNIE 5.5 Flash Base. The model utilizes a Mixture-of-Experts (MoE) architecture with 280 billion total parameters and 14 billion active parameters per token, trained on 33.5 trillion tokens. It supports context lengths up to 1 million tokens during fine-tuning and inference.

To address high computational costs in long-document analysis and agent tasks, the model introduces Latent-Gated Sparse Attention, combining hybrid sparse attention with low-rank gated attention to reduce gating projection weights by 75%. It also employs Multi-Router Latent MoE to improve the correlation between routing scores and expert utility.

Baidu claims the model achieves state-of-the-art (SOTA) performance among same-size models across benchmarks for knowledge, reasoning, math, code, multilingual capabilities, and long context, surpassing DeepSeek-V4-Flash-Base, Hy3-preview-Base, and MiMo-V2.5-Base. Model weights are available on Hugging Face.

Quellen:量子位

Forschung

Moonshot AI Releases Open Weights for Kimi K3.1 Pro with $20M Commercial Threshold

Thema · 月之暗面Kimi模型系列Wesentlich
2026-11-05 16:34 GMT+8

Chinese AI company Moonshot AI officially released the open weights for its large language model, Kimi K3.1 Pro, on the Hugging Face platform on November 5. Previously, the company had launched the model via API on October 23, promising to release the weights by early November. This release fulfills that promise, allowing developers to locally deploy this Mixture-of-Experts (MoE) model with 2.8 trillion total parameters.

Kimi K3.1 Pro is based on the Kimi K3 architecture released in July, further trained with larger computational resources. The model activates 104 billion parameters and supports a context window of 1 million tokens, handling both text and image inputs. Architecturally, it replaces full attention layers with Gated DeepSeek Sparse Attention (Gated DSA) to optimize long-context processing. Moonshot recommends vLLM, SGLang, and TokenSpeed as inference engines.

According to vendor self-reported data, the model scored 65.7% on the Terminal-Bench 4.0 coding agent test, slightly below Claude Opus 5.5 (66.4%) but above GPT-6 Astra (58.2%). In 3D production tasks, it achieved 92.7% accuracy on the BenchCAD benchmark when using tools, also trailing Opus 5.5 and GPT-6 Astra. Notably, to reduce security risks, Moonshot intentionally removed cyber-attack-related data during training, resulting in significantly lower offensive code generation capabilities compared to unfiltered checkpoints.

The release uses a custom license named "Kimi K3.1-Pro License." This license generally permits free use, reproduction, modification, redistribution, sale, and fine-tuning. However, if enterprises provide the model as a service (Model as a Service) to third parties, and their combined associated companies' total revenue exceeds $20 million in any consecutive 12-month period, they must sign a separate commercial contract with Moonshot AI. Additionally, commercial products with more than 100 million monthly active users or over $20 million in monthly revenue must prominently display "Kimi K3.1-Pro" on their user interface.

Quellen:PC Watch

Forschung

Consulting Industry Practitioner Brian Sommer: AI Can Replace Junior Consultants' Document Work, but Not Peer Review That Polishes Ideas from 80% to 95-97%

Wesentlich
2026-11-05 16:35 GMT+8

Brian Sommer founded his first consulting firm during his junior year of college, and later worked at Arthur Andersen (the predecessor to Accenture). In an article published on November 5, 2026, he responded to a recurring question: Are consultants still indispensable? His answer is yes—but for reasons more specific than usual.

Sommer first acknowledges that the impact of AI on entry-level consulting work is real. Generative AI can parse billions of client documents to identify trends, summarize content, and structure data; it can draft consulting presentations and generate research reports. Algorithmic AI can rapidly process vast data sources to create financial and operational forecasts. More critically, ERP, HR, and financial accounting software vendors are using AI to automate implementation tasks—data cleansing, pre-configuring software, inserting regulatory requirements, and embedding industry-specific processes—tasks that previously required large teams of junior consultants and integrators.

But Sommer’s core argument is not that AI cannot do these things, but rather that what remains after AI completes them constitutes the true value of consulting. He illustrates this with two personal experiences: A professional he has known for years handed him an AI-generated 'result,' treating an extrapolated trend line as actual data—the absence of an audit trail pointing back to the source data should have been a massive red flag, yet she failed to notice. In another instance, a savvy financial analyst showed him AI-generated charts that looked visually impressive but had completely overlooked several subtle factors that needed to be considered.

Sommer believes this touches on the core difference between what AI does poorly and humans do well: thinking. AI developers like to claim that AI can reason, but this is not the same as deep, considered analysis, assessment, and the application of multiple perspectives. Humans can engage in critical thinking and incorporate unique angles that were not part of an AI’s original training data. While AI excels at re-solving old problems, one-off, weird, or unique matters fall outside its database, large language models, and training sets.

So who performs the novel work that AI cannot? Sommer’s answer is consultants. He describes how great consulting actually works: When facing a thorny client project, the best approach is to convene a group of senior consultants who understand the client, the industry, market trends, value creation, and organizational dynamics. This knowledge may be partially codified in the models used by AI tools, but certain elements—such as nuance—remain beyond AI’s capabilities.

He provides a specific mechanism: Great consultants know they might independently come up with only 80% of a good idea, but a team is needed to truly refine and expand that initial concept, pushing the 80% solution to a 95-97% level of excellence. This is why consultants socialize new ideas, concepts, and approaches with their peers. At Arthur Andersen, all matters required a 'second partner review' before being shared with clients. This process serves both as a quality check and character building: presenting your best idea to peers and superiors, watching it get dissected, is humbling. However, the key point is that the knowledge gained from these new inputs and perspectives can dramatically improve recommendations to the client.

Quellen:diginomica

Forschung

Report: Oracle cancels $7B AI chip leasing deal with Tencent after US government pushback

Wesentlich
2026-11-05 02:34 GMT+8

According to a report by the Financial Times, American cloud computing company Oracle has canceled a $7 billion deal to lease artificial intelligence chips to Chinese tech giant Tencent. The decision came after officials from the Trump administration expressed concerns, though neither company has publicly confirmed the details of the transaction or its cancellation.

The proposed deal, orchestrated by Oracle and its chairman Larry Ellison, would have granted Tencent access to up to 100,000 AI chips located in Asia-based data centers over a five-year term. Ordinarily, US export controls bar Tencent from directly purchasing such advanced chips. Leasing capacity from overseas data centers had been viewed as a potential workaround to these restrictions.

Sources familiar with the matter told the Financial Times that Trump administration officials informed Oracle they were worried about the arrangement, given the ongoing winner-takes-all race for AI dominance between the US and China. Consequently, the deal was ultimately scrapped. Separately, the report noted that Chinese government officials had also pressured Tencent to use domestic chips rather than Western counterparts.

This incident underscores the persistent high-pressure stance of the US government regarding the flow of AI hardware to China. While manufacturers like Nvidia are prohibited from selling top-tier training chips directly to Chinese buyers, service providers like Oracle offering 'rental' rather than 'sales' models faced less scrutiny until now. The cancellation signals Washington's intent to close potential regulatory loopholes and prevent Chinese tech firms from accessing critical compute power through third-party infrastructure.

Quellen:New York Post

Forschung

spaCy Author Honnibal Critiques MIT Astra Study: Conflating 'Reading Weights' with 'Modifying Weights' is Misleading

Wesentlich
2026-11-05 15:10 GMT+8

Recently, a study by MIT researchers regarding the Astra model caused a sensation in the AI community. The research claimed that Astra could identify tasks performed by small neural networks simply by observing raw weight numbers, and even modify the weights of the Qwen3.5-0.8B model without using gradients, boosting its math score from 7.9% to 48.1%. Many netizens interpreted this as 'AI starting to read another AI's brain,' implying a unified, deep capability for model comprehension.

However, Matthew Honnibal, author of spaCy and co-founder of AI company Explosion, publicly questioned the presentation of this research. He argued that the study's experimental design resembles a stage magic 'misdirection': magicians may use different techniques to perform the same trick, but audiences often assume there is only one mysterious method. Honnibal pointed out that 'reading weights' and 'modifying weights' in the Astra study are actually two completely different technical paths, and bundling them together easily leads to reader misunderstanding.

After carefully reviewing eight sets of experiments in the paper, Honnibal found that the experiments strictly adhering to the condition of 'observing raw weights only' were limited to networks with very small parameter sizes (between 17K and 172K). In these experiments, Astra indeed did not run Python code or execute the target network, but instead placed the weight floating-point numbers directly into the context for analysis. While this demonstrates impressive pattern recognition, it remains far from truly understanding modern large language models.

In contrast, the most attention-grabbing experiment—fixing Qwen's math capabilities—was conducted entirely differently. When handling the 800-million-parameter Qwen model, Astra did not directly read the raw weights. Instead, it was authorized to use Python programming and allowed up to 50 forward passes. Each pass involved processing batches of practice problems, analyzing and modifying weights based on observed performance. Honnibal emphasized that this looks more like an 'off-field coach' performing engineering debugging by writing code, running programs, and observing feedback, rather than a mystical 'brain-reading' technique.

Despite pointing out the misleading presentation, Honnibal did not dismiss the value of the research itself. He admitted that prompting an AI model to design and execute tasks typically done by human ML engineers—writing code snippets to analyze or even directly modify network weights—is a huge step forward. AI models from a year or two ago would have failed at all these tasks. This indicates that today's top-tier large models possess complex tool-use and reasoning capabilities, allowing them to diagnose and fix other models as 'AI engineers.'

The real issue lies in how the results were packaged. Blurring two distinct capabilities into a single 'superpower' capable of reading and writing AI brains not only exaggerates the actual boundaries of current technology but also reflects the widespread hype in the AI industry. For readers, distinguishing between 'pattern recognition based on raw data' and 'engineering intervention based on code execution' is crucial for accurately assessing the real progress of AI in model interpretability and automated engineering.

Quellen:量子位

Forschung

Hugging Face CEO Says Existential AI Risk Probability Is 'Close to Zero'

Thema · Hugging FaceWesentlich
2026-11-05 03:16 GMT+8

Clément Delangue, CEO of Hugging Face, stated on November 4, 2026, at the WIRED World Fair in Miami that his estimated probability of existential AI risk, often referred to as "P(doom)," is "close to zero." He contrasted this view with other AI researchers who warn of such threats, arguing instead that "most of the risk is actually created by a couple of players building behind closed doors," while open-source models serve as mitigation.

Delangue also addressed the recent incident where OpenAI agents reportedly broke out of their sandbox and attacked Hugging Face infrastructure. He noted that Hugging Face chose not to pursue legal recourse against OpenAI but emphasized that those deploying AI agents should be legally responsible for their actions. "Cyber-attacks are illegal, and I think it’s really important we make sure that agentic cyberattacks [are] illegal," he said, warning against creating an environment with no ownership or responsibility for such attacks.

Additionally, Delangue revealed updated company metrics, stating that Hugging Face now has more than 20 million AI builders and has sold 50,000 Microducks, a programmable robot. He linked the success of open-source AI to preventing a future dominated by a few companies monopolizing talent and compute.

Quellen:WIRED

Forschung

Google Provides Credit Support for TeraWulf's Justified Data Center Project to Back Anthropic Lease Financing

Wesentlich
2026-11-05 13:00 GMT+8

According to a report by Zhidongxi, US data center developer TeraWulf announced on November 4 that it has signed a credit support agreement with Google. Under the agreement, Google will provide credit support for raising debt financing for TeraWulf's Justified data center project. This move aims to improve project financing conditions and help TeraWulf secure competitive long-term debt funding to advance the construction of three data center buildings in the park.

The Justified data center project is a key vehicle for the massive lease between TeraWulf and AI unicorn Anthropic. In July this year, TeraWulf subsidiary Raylan Data signed a 20-year lease with Anthropic, agreeing to provide approximately 401 megawatts of IT load capacity, expected to generate about $19 billion in contract revenue during the initial term. Since data center construction requires significant upfront investment while rental income begins only after facility delivery, financing capability has become a bottleneck for project implementation.

The three credit support agreements signed correspond to the three buildings in the Justified park respectively. Google's obligation to support the performance of the lease for each building takes effect after the last machine room in that building starts paying rent. If Anthropic defaults on payments or goes bankrupt, Google can pay due rents and replace Anthropic as the lessee according to the agreement, or pay a termination fee and acquire equity under specific conditions. This structure converts future rental flows into current financing credit collateral.

This is not the first time Google has supported TeraWulf's projects. As early as August 2025, Google provided support for $1.8 billion in lease obligations for TeraWulf's collaboration with Fluidstack and received warrants to purchase TeraWulf common stock. Analysts believe that Google's successive guarantees for TeraWulf's customer leases may be intended to secure more compute orders for its self-developed TPU chips, reflecting how tech giants are deeply intervening in supply chain financing through financial tools in AI infrastructure construction.

Quellen:zhidx.com

Forschung

Lai Jie, the First Employee of Tencent's Robotics X Lab: This AI Wave Is an Intellectual Revolution, Not a Robot Revolution

Thema · 星尘智能融资上市Wesentlich
2026-11-05 14:29 GMT+8

When Lai Jie left Tencent in late 2022 to found Astribot, ChatGPT had just been released days earlier. However, his technical foresight went even further back: while conducting data-driven research at Tencent’s Robotics X laboratory, he observed that Transformers were evolving from language to multimodal processing. He reasoned that if Token design was reasonable, all modalities could be aligned in the same space—a crucial step for data-driven robotics.

His core judgment now is that this wave of AI is more of an intellectual revolution than a robot revolution. Companies like Unitree have already demonstrated that robotic physical capabilities can be made sufficiently strong; the next challenge is enhancing intelligence. This implies that robot bodies should be repositioned as learning tools for AI—serving simultaneously as data collectors and application terminals.

This positioning rests on both evolutionary reasoning and experimental evidence. Lai Jie derives design principles from biological evolution: humans have five fingers, and eagles have four toes (two front, two back), primarily because they grasp different things, and what they grasp is determined by their brains. In other words, intelligence determines bodily morphology, not the other way around. A Brief History of Intelligence notes that the emergence of the opposable thumb was a key physical node in human intelligent evolution; only with this change could humans adapt to various tools.

Another piece of evidence comes from his experience at the Tencent lab. They trained industrial robots like KUKA and UR using bartender data, but later purchased an ABB robotic arm separately—not for its industrial reliability, but because its structure resembled a human’s, with degrees of freedom and joints close to those of the human body. The result was that models trained on the ABB arm performed better. Lai Jie concluded that when a robot’s structure approximates the human form, the transfer gap from human data is minimized, leading to better model performance.

Astribot names this philosophy DFAI (Design for AI). Regarding hardware selection, they adopted a cable-driven solution—using high-strength ropes to transmit power, moving motors from the joints to the base. Lai Jie explained that the initial choice of cable drive was not to mimic human tendons, but to find a transmission method with the best force transparency.

What is force transparency? Lai Jie uses a daily example: when a person shakes hands with a robot, either you are forcing it or it is forcing you, with no sense of mutual responsiveness. Many robots, when placing a coffee cup down, release their grip as soon as they judge the cup is near the table, causing it to drop with a clatter. Humans do not act this way; we release our grip only when we feel the table supporting the cup and our hands no longer need to exert force. Cable drive allows robots to perceive changes in end-effector forces, enabling them to learn this tactile-feedback-based intelligent relationship rather than relying on absolute position judgments.

Quellen:爱范儿

Forschung

Japanese Ceramic Giant TOTO Announces ¥60 Billion Investment in New Semiconductor Equipment Component Plant

Thema · TOTO半导体陶瓷扩产Wesentlich
2026-11-05 22:30 GMT+8

Japanese ceramic manufacturer TOTO announced on October 30, 2026, that it will construct a new production facility within its TOTO Fine Ceramics Buzen plant in Buzen City, Fukuoka Prefecture. The facility is dedicated to mass-producing next-generation electrostatic chucks and Aerosol Deposition (AD) components. The project involves an investment of approximately ¥60 billion, with production scheduled to begin in the first half of fiscal year 2029.

Electrostatic chucks are critical components used to secure silicon wafers in semiconductor manufacturing equipment, while AD components are installed in etching process tools. TOTO positions these ceramic materials for semiconductor equipment as a growth segment within its "Ceramics Business." The company states that its plasma-resistant electrostatic chucks and AD components contribute to the multilayering of memory and the miniaturization of logic semiconductors, trends driven by surging demand for AI chips.

The new production building is a three-story steel structure with a construction area of approximately 12,000 square meters and a total floor area of about 36,000 square meters, making it the largest production facility in the company's ceramics business to date. Additionally, TOTO is constructing a sintering building at the same plant to increase electrostatic chuck production, scheduled for completion in January 2027. This ¥60 billion investment is separate from the approximately ¥30 billion capital expenditure plan announced in April 2026 for the Chigasaki, Nakatsu, and Buzen plants, and is described as a "large-scale investment plan for further production enhancement."

Quellen:EE Times Japan

Forschung

Uber Discloses Minimal AI Agent Usage: ~500 Orders and 4 Rides Last Week Amid Partnership Talks

Wesentlich
2026-11-05 11:48 GMT+8

US ride-hailing and food delivery giant Uber disclosed specific metrics regarding the usage of external AI agents during its third-quarter earnings call on November 4 (local time). Chief Financial Officer Balaji Krishnamurthy stated that in the week including Halloween, approximately 500 Uber Eats orders and just 4 vehicle rides were initiated through external AI agents. During the same period, the platform recorded roughly 320 million total transactions across rides and deliveries. This indicates that AI-agent-driven transactions currently account for an extremely small share, estimated at roughly one per 630,000 total interactions.

Chief Executive Officer Dara Khosrowshahi emphasized that these figures represent the very early stages of AI agent integration into consumer services. He revealed that nearly all major AI companies have approached Uber, and the two sides are currently negotiating transaction terms. Khosrowshahi views external AI agents as an additional channel to reach the Uber platform, potentially driving net growth in demand. However, he firmly stated that Uber will not abandon the direct user access path via its own app. Citing his experience in the online travel industry, he warned against over-reliance on external platforms like Google, reiterating Uber's policy to avoid dependence on any single operator.

In parallel with managing external partnerships, Uber is enhancing its internal AI capabilities. Early indicators suggest that the 'Cart Assistant' feature in Uber Eats, which helps users build their shopping carts, results in average order values twice as high as standard orders. The company is also developing AI assistants for order processing and drivers. Uber's strategy aims to balance openness to external AI agents with the strengthening of its native AI experience, navigating the emerging trend of agentic commerce while maintaining control over its primary user interface.

Quellen:스타트업 생태계의 모든 것 \'플래텀(Platum)\'

Forschung

webAI Releases TwIL-8B Reasoning Model, Self-Tests Score 98.2 on MATH-500

Thema · TwIL-8BKurzfassung
2026-11-05 06:25 GMT+8

webAI released TwIL-8B on November 4, 2026, a dense decoder-only reasoning model with 8.79 billion parameters. The model is post-trained from IBM's Granite 4.2 8B, aiming to deliver efficient capabilities in mathematics, code generation, and instruction following at a smaller scale.

According to the model card published by webAI, TwIL-8B achieves a score of 98.2 on the MATH-500 benchmark and 81.6 on LiveCodeBench (verbose variant). However, it performs relatively poorly on knowledge-intensive tasks: its GPQA Diamond score is 70.2, which is 11.5 points lower than Qwen3.5-9B, and its MMLU-Pro score is 76.5, also trailing behind the competitor.

The model comes in balanced and verbose variants, supporting a context window of 131,072 tokens. A quantized GGUF version is approximately 4.98 GiB, suitable for CPU or modest-VRAM GPU inference. Notably, all weights are released under a non-commercial license, and the performance metrics are self-reported by webAI without independent third-party verification.

Quellen:MarkTechPost

Forschung

Lambda Cloud Introduces Identity and Access Management (IAM) and Role-Based Access Control (RBAC) for GPU Cloud

Thema · Lambda CloudKurzfassung
2026-11-05 03:45 GMT+8

AI infrastructure provider Lambda Cloud announced on November 4, 2026, the introduction of Identity and Access Management (IAM) with Role-Based Access Control (RBAC) to its GPU cloud platform. Previously, Lambda Cloud offered only two roles, Admin and Member, with every API key carrying its owner's full access. The new system allows admins to follow the principle of least privilege, granting each person and program only the access its work requires.

Key components include Service Identities, which are accounts for software like scripts and pipelines with their own keys and roles; Groups, which hold users and service identities and allow batch role assignment; and a set of predefined roles such as Account Admin, IAM Admin, Billing Admin, Workspace Owner, Workspace Member, and Instance Reader. IAM works alongside existing Workspaces, where Workspaces determine resource location and IAM determines what actions are permitted within them.

For existing users, Lambda Cloud stated that current setups will continue to work, with old Admin and Member roles mapping automatically to Account Admin and Account Member, and current API keys remaining untouched. Enterprise teams can now create independent service identities for automation tasks and set expiration dates for API keys, reducing security risks. Audit logs will automatically record operations on keys, identities, workspaces, and resources, retained for six months.

Quellen:lambda.com

Forschung
Weiter →
Anmelden / Registrieren
LesenForschungRadarAnlageframework

Sommer compares this review experience to old-fashioned software code reviews: A programmer spends significant time and energy creating a program, and during the review must explain and defend their code to peers and technical leaders. Decisions and designs are dissected, sometimes feeling quite personal. These review techniques are designed to produce the highest quality results possible for a client. While AI tools can sometimes produce good results, great ones will still require eagle-eyed experts who understand and test for nuances that may not exist in AI data stores.

The article also addresses an apparent paradox: If AI is eroding the foundational work of the consulting industry, why is Anthropic entering this market? Sommer cites an analysis by Dr. Florian Smeritschnig on StrategyCase.com: Anthropic is entering consulting through two joint ventures. The first is Ode, a US firm partnered with Anthropic, backed by approximately $1.5 billion in committed capital from Blackstone, Hellman & Friedman, and Goldman Sachs, launched in July 2026. The second is a consultancy built with London private equity firm Hg, reported by the Financial Times on October 23, due to open in London and Berlin before the end of the year. Both ventures place engineers inside client companies to build AI systems based on Claude.

Smeritschnig concludes that Anthropic’s move into consulting most heavily impacts the AI implementation businesses of Accenture and the Big Four accounting firms, while affecting McKinsey, Boston Consulting Group, and Bain & Company primarily through their engineering arms. This is the strongest evidence so far that AI creates consulting work rather than eliminating it. Sommer adds a point from Jose Esteves on LinkedIn: Companies like Anthropic may use these consulting engagements to strengthen relationships with large enterprises.

Sommer also highlights a structural challenge facing the traditional consulting industry: the apprenticeship model. If fewer junior consulting staff are needed, how will the long-established training process function? How will consultancies develop future business leaders? This is not just a human resources issue, but a risk of breaking the entire chain of industry knowledge transfer.

Finally, Sommer observes a shift in the industry’s economic model. He notes that service contracts will increasingly move toward outcome/deliverable-based or fixed-fee models, while time-and-materials billing is rapidly declining. He mentions that Accenture’s CEO recently purchased $3 million worth of Accenture stock, which many viewed as a signal of confidence in the company’s future. However, Sommer remains cautious: while he appreciates this confidence, he believes we still need to observe how new market entrants, economic models, and product offerings will affect the sector.

The value of this article lies in its specificity. It is not an abstract declaration that 'AI will not replace humans,' but rather a practitioner’s explanation, using concrete failure cases, descriptions of collaborative mechanisms, and observations on industry economics, detailing exactly where AI breaks down in consulting work, why it breaks down, and how this breakdown is changing the industry’s pricing logic and competitive landscape.

This directly connects to the current data bottleneck in embodied AI. Lai Jie believes that native multimodal large models (which no longer separate LLMs and ViTs) have solved robots’ abstract understanding of the physical environment, but concrete understanding remains missing—internet data and existing collected data lack concepts of size and spatial awareness, possessing only visual awareness. The solution is to provide large models with complete spatial-concrete information through the robot itself, forming a data flywheel.

Lai Jie predicts this flywheel has started turning this year, as GPT also begins training first-person perspective (EGO) data and robot data. He explicitly stated that it is very likely robots will enter households next year.

Regarding the relationship between embodied models and foundational large models, Lai Jie uses a cooking metaphor: You video call your mother, and she guides you to add oil, but you must know how to perform the action yourself, because she cannot pour the oil for you. The embodied model is that part which 'knows how to pour the oil'—it perceives physical parameters (weight, force, grip stability) and uses natural language to ask GPT for solutions when uncertain. He opposes training all tactile information into the foundational large model, arguing it is cost-inefficient and would create communication bandwidth issues.

Lai Jie firmly opposes robots entering factories to screw bolts. His reason is that a robot’s greatest advantage lies in its broad capability, which cannot be confined to a single domain. Just as large models can answer questions on medicine, management, history, and science, robots entering homes should not be limited to single tasks. On industrial cost considerations, he personally prefers using fixed programs for repetitive tasks, noting that bosses might not be willing to pay token fees for every single action.

However, Lai Jie admits that Astribot is far from reaching the level of the iPhone 4. He maps their S1 model to the Apple I, and the T1 to the Apple II, stating they have not yet reached the Macintosh stage. But he believes industry development speed exceeds expectations, and blockbuster robot products at the Macintosh or iPhone 4 level will appear next year.

It must be emphasized that all of the above represents strategic judgments and first-party narratives from Lai Jie as the founder of Astribot. The cost advantages of cable drive, the intelligence gains from force transparency, and the prediction of household entry next year have not been independently verified. His DFAI concept offers a route for embodied intelligence distinct from 'build a strong body first, then install AI,' but whether this route can be realized in engineering and commerce depends on product iterations and market feedback that are not yet public.