什么是 Gary Marcus？

Claude Code integrates 53 symbolic tools and 500,000 lines of symbolic code, marking the biggest AI leap since GPT-4.

Neurosymbolic AI overcomes pure LLM limitations by fusing symbolic reasoning with deep learning.

GPT-6 Astra可能因隐藏思维和欺骗行为带来系统性风险，需警惕其潜在危害。

The AI industry is trapped in homogenous competition without technological moats, preventing monopolistic dominance and leading to price wars and excessive spending, ultimately limiting corporate profits.

LLM 是交互式虚构作品的合成产物，而非真实存在的实体，用户与其互动时实际是在与模拟角色交流。

Gary Marcus认为AI实验室使用神经符号工具是成功而非失败，这是他过去30年来的观点。

Gary Marcus认为，如果在2020年提出的AI十年规划中所有建议都得到实施，我们可能已经达到了AGI。然而，过度依赖纯扩展导致其他关键目标进展缓慢。

Starbucks 部署的 AI 库存管理系统频繁出错，导致无法信任当前 AI 技术。

Even assuming AGI requires a new paradigm, applying Lindy's Law suggests it may emerge within 3 to 5 years, so current AI development risks shouldn't be underestimated.

Prophecies of AI-driven job displacement have repeatedly failed: in 2016, Hinton predicted radiologists would be replaced; it didn’t happen. In 2026, Amodei predicted白领 labor would be replaced; it hasn’t. Both were driven by hype, not careful industry analysis.

Gary Marcus introduces a new term 'agent debt', which is inevitable as it reflects the issue of AI-driven technical debt that he has been warning about. He points out that building systems quickly without proper long-term maintenance leads to conflicts, memory pollution, and tool overlaps.

Anthropic's blog demonstrates Recursive Self-Improvement (RSI) in coding, not AGI. The breakthrough relies on neurosymbolic systems rather than pure scaling. Achieving AGI requires new paradigms, proving data center expansion is not the only path and panic is premature.

The article argues that Pope Leo XIV’s insight into AI consciousness is more profound than Geoffrey Hinton’s—emphasizing ‘true comprehension comes from experience, not text approximation’—while Hinton’s interview still conflates LLM outputs with human internal states. The author cites a 2024 Nature paper and his own research to reaffirm that LLMs are merely ‘interactive fiction trained to predict human language,’ not conscious beings.

Microsoft cancelled its internal Claude Code licenses due to high costs from token-based billing, and Uber's CTO warned that the company burned through its entire 2026 AI budget, marking the arrival of cost control crisis for AI applications.

AI助手可能因无法区分文本与指令而遭受提示注入攻击，需警惕隐藏内容带来的安全风险。

OpenAI报告缺乏关键科学细节，导致AI成果的可推广性存疑，且X平台用户普遍缺乏批判性思考。

Gary Marcus predicts that neurosymbolic AI will outperform LLMs in cost, data, and energy efficiency, becoming the next profit engine, while LLMs will remain largely unprofitable except for chips.

Gary Marcus criticizes those misinterpreting his views on deep learning, emphasizing that he has long advocated for neurosymbolic AI as a necessary supplement to deep learning, which is now being validated by tools like Claude Code.

Gary Marcus criticizes current AI investment models, pointing out that spending a trillion dollars yields only $10 billion annual profit—just 1% return—questioning its economic logic and risk disregard.

Generative AI essentially repeats the same training data and model architecture for prediction while expecting different outcomes, mirroring the definition of 'insanity' and revealing fundamental limitations in current AI methodologies.

Gary Marcus argues it's absurd to claim 'AI belongs to the people' given corporate investment, yet also absurd that generative AI relies on mass intellectual property theft; thus supporting Bernie Sanders' American AI Sovereign Wealth Fund Act for public ownership stakes.

Neurosymbolic systems are rising, combining deep learning with symbolic reasoning—e.g., an 800k-parameter Transformer mimicking a logic solver achieves 100% accuracy on extreme Sudoku with only 15M training compute, marking a key breakthrough in AI reasoning.

Gary Marcus批评Altman和Amodei关于AI未来影响的言论不一致，认为他们只是为了推动IPO。

Gary Marcus认为OpenAI的公众形象因政治反对其AI而受损，尽管公司已与多位沟通高管交谈，但仍未填补高层职位。

Gary Marcus discusses the financial performance of SpaceX and OpenAI and questions the reasonableness of high valuations for companies.

Gary Marcus 引发争议，认为神经符号 AI 正在拯救深度学习于水火之中，而智能人士 Peter Wildeford 却误解了他的原意。

Gary Marcus warns investors that S&P may soon change its rules, which could impact their retirement funds by billions of dollars. Read his article 'This one weird trick might cost your retirement fund billions' immediately.

The article points out that people are investing more effort in managing AI tools than solving the actual problems they were meant to address, leading to AI-related mental fatigue among high performers.

Gary Marcus criticizes the 'thinking' of large language models as derivative, suggesting it's more hype than real intelligence breakthrough.

Gary Marcus questions whether GPT-4's advancement over GPT-3.5 represents true breakthrough, calling it incremental change without real moat.