Brendan Eich, CEO of Brave Software and creator of JavaScript, recently weighed in on the current state of artificial intelligence and the emerging landscape of agentic browsers, emphasizing the economic realities of achieving widespread distribution. In a social media post, Eich pointed out that current AI architectures, as noted by Yann LeCun, often "saturate at or below human intelligence at very high training cost, still hallucinate, etc.," despite the significant advancements in the field. He further asserted that agentic browsers, while "shaking things up," will struggle to gain "big distribution without big revenue to share," suggesting that "search ads in agents" are an inevitable component.
Eich's comments underscore a critical challenge for the next generation of AI-powered web tools. Agentic browsers aim to autonomously perform complex tasks for users, moving beyond traditional browsing. For instance, Perplexity's "Comet" is highlighted as an AI-first browser built around these capabilities, demonstrating the potential for new workflows and monetization opportunities in the AI-native application space.
The discussion also echoes long-standing concerns from prominent AI researchers like Yann LeCun, Meta's Chief AI Scientist. LeCun has consistently argued that large language models (LLMs) lack true reasoning, planning abilities, and common sense, leading to issues like "confabulations" (hallucinations). He believes that current LLMs are far from human-level intelligence and require a fundamental shift in architecture, such as his proposed Joint Embedding Predictive Architecture (JEPA), to overcome these limitations.
Brave, under Eich's leadership, has been a pioneer in reimagining the browser's economic model through its Basic Attention Token (BAT) system, which allows users to earn revenue share from privacy-respecting ads. This model aims to align user, publisher, and advertiser interests by enabling direct micropayments and rewarding user attention. Eich's current stance suggests that even with innovative approaches like BAT, traditional revenue streams, particularly search advertising, will remain crucial for funding the development and scaling of sophisticated agentic AI systems.
The integration of search ads into agentic AI interfaces presents a complex balancing act between user experience, privacy, and economic viability. As AI agents become more prevalent, their ability to generate revenue will directly impact their reach and the quality of services they can provide, making search ads a key consideration for their sustainable growth and broad adoption.