A recent analysis by lessw-blog evaluates the feasibility and strategic value of Whole Brain Emulation (WBE), concluding it is unlikely to play a competitive or cost-effective role in the transition to Artificial General Intelligence.
\nIn a recent post, lessw-blog discusses the feasibility, cost-effectiveness, and safety implications of Whole Brain Emulation (WBE) in the context of the AI transition.
As the race toward Artificial General Intelligence (AGI) accelerates, researchers and strategists are evaluating alternative pathways to advanced cognition. Historically, WBE-the theoretical process of scanning and simulating a human brain in software-has been proposed as a potential stepping stone to AGI or a safer alternative to purpose-built AI systems like Large Language Models (LLMs). Proponents have often argued that emulating a human mind might bypass the complex alignment problems associated with alien AI architectures. Understanding whether WBE is a viable, cost-effective, or safe route is critical for resource allocation, especially when measuring computational costs in modern hardware metrics like H100-equivalents.
lessw-blog has released analysis that systematically dismantles the optimism surrounding WBE's role in the near-term AI transition. The author argues that without AGI, WBE is likely decades away, with current timelines only supporting very short, rudimentary simulations. Even if AGI is achieved first and used to accelerate the process, physical constraints-such as the time required for biological preparation, high-resolution scanning, and massive data collection-mean WBE would still lag behind AGI development by several years. It is not a milestone that can be reached in mere months post-AGI.
Furthermore, the economic and safety cases for WBE appear fundamentally weak based on back-of-the-envelope calculations. Emulations are projected to be significantly more expensive per unit of cognition than biological humans, and orders of magnitude more expensive than purpose-built AI systems. From an alignment perspective, the post argues that emulations do not inherently offer a more trustworthy starting point for superintelligence. They also fail to provide a cost-effective method for learning about human values compared to studying human data directly. Consequently, pre-AGI investment in WBE is deemed largely inefficient, with its strategic value concentrated only in highly specific, narrow scenarios.
For strategists, researchers, and policymakers navigating the AI transition, this piece offers a sobering, highly analytical perspective on the limitations of WBE. It emphasizes the necessity of focusing on the alignment and development of purpose-built AI systems rather than relying on brain emulation as a strategic fallback or safety net.
\n\n