garymarcus-substack-com

Gary Marcus announces he is taking a break from his Substack for adventures, leaving readers with some humorous content.

Gary Marcus argues that China has made significant progress in AI and technology, catching up to or surpassing the US in key areas. He suggests that US policymakers have been overly focused on safety concerns while neglecting the competitive threat posed by China's rapid advancements.

The article discusses how June 2024 marked a turning point for generative AI, with notable developments and challenges emerging that undermined the technology's earlier momentum and promise.

The article argues that generative AI is facing a slowdown, or "fizzle," as initial hype gives way to practical limitations and market realities, despite the potential for future surprises.

The article discusses the evolution of Accenture from its early days to its current state, and speculates on what its trajectory might indicate about broader business and technology trends. It questions whether recent developments at Accenture are a temporary anomaly or part of a larger pattern.

President Trump has reportedly asked Anthropic, the AI safety company behind Claude, to undertake a task that may be technically or ethically impossible, raising questions about the future direction of AI regulation and corporate responsibility.

Gary Marcus argues that OpenAI's competitive advantage in AI is fading due to a lack of a protective "moat," as rivals increasingly match or surpass its technology and open-source alternatives gain traction.

The article presents two contrasting sets of predictions for what will occur following the decline of tokenmaxxing, a trend focused on maximizing token value in crypto and AI contexts.

The article contrasts the Pope's nuanced understanding of AI with Geoffrey Hinton's perspective, emphasizing that the way a statement is generated—not just the statement itself—matters.

The article discusses the mathematical and psychological reasons why complex systems and structures eventually break down over time.

The article describes a real-life moment where the author's hopes for AI sanity are realized, reflecting on a personal experience that underscores this breakthrough.

The article argues against panicking over Anthropic's latest blog post, suggesting the reaction on Twitter is overblown.

Gary Marcus contrasts two competing visions of AI progress held by Demis Hassabis: one public-facing, optimistic timeline suggesting AGI within a decade, and a more cautious internal view acknowledging fundamental scientific bottlenecks. Marcus argues this tension reflects deeper contradictions in AI hype versus reality.

The article argues that Anthropic did not actually call for a pause on AI development, contrary to some media reports. It clarifies the company's actual position and criticizes how their statements were misrepresented in the press.

Gary Marcus reflects on recent AI developments, drawing parallels to retail's Black Friday to discuss the rapid, messy rollout of AI technologies, persistent limitations, and the hype cycles surrounding them.

Gary Marcus discusses a Financial Times graph from John Burn-Murdoch that illustrates sluggish productivity growth despite rapid AI advancements, arguing that current AI tools generate "slop" rather than genuine economic gains, and the hype-driven AI world is failing to deliver transformative productivity improvements.

The article argues that the AI industry relies on unsustainable financial math and unrealistic projections, suggesting that much of the current investment and valuation is based on fantasy rather than solid economic logic.

Gary Marcus critiques Claude 3.5 Sonnet, arguing that while it shows improvements, it still exhibits the same fundamental flaws as earlier large language models—confabulation, lack of true reasoning, and reliance on pattern matching rather than genuine understanding. He warns against overhyping its capabilities and compares the cycle of AI enthusiasm and disappointment to a recurring "mythos."

The article reflects on a recent interview with investor Steve Eisman and discusses potentially critical news developments, offering analysis on how significant changes or disruptions might begin to unfold.

A court has ruled Google can be held liable for AI-generated hallucinations produced by its systems, marking a significant legal precedent that could influence future cases and regulations in other jurisdictions.

A new German court ruling raises questions about whether Section 230 of the Communications Decency Act protects AI companies from liability, potentially upending the legal landscape for artificial intelligence firms.

OpenAI is reportedly considering significant price reductions, a move that signals competitive pressure or weakening demand rather than strength.

The article discusses a notable AI hallucination, highlighting how large language models can confidently generate false or fabricated information, which underscores ongoing reliability issues with such technology.

The US Commerce Department has taken decisive action to shut down Anthropic's latest AI models, marking a sharp shift from its previous two years of minimal AI regulation. This move represents the government's adoption of more aggressive enforcement measures in the AI sector.

The article criticizes the White House's AI policy as disorganized and ineffective, leading U.S. states to introduce their own AI regulations. It argues for a more coherent federal approach to AI oversight.

Gary Marcus argues that the U.S. government urgently needs to establish a federal AI agency to regulate and oversee the rapidly advancing technology, warning that without coordinated action, risks from AI—including misinformation, bias, and safety failures—will outpace any benefits.

Customers are recognizing that tokens from three major IPOs are being burned for millions of dollars with little tangible return on investment.

Uber COO Andrew Macdonald stated that the company is not seeing proportional productivity gains from its rising AI costs, raising concerns about the return on investment in artificial intelligence.

The article warns that a seemingly minor policy trick could potentially cost retirement funds billions of dollars, urging readers to contact their representatives in Congress.

Gary Marcus scrutinizes recent claims from OpenAI and Anthropic, urging readers to examine the underlying details and math behind their headline-grabbing announcements rather than taking them at face value.