AI as the beginning of new coevolutionary domains

Venkatesh Rao's recent essay "Descent into the Computational Sublime" presents a fascinating evolutionary perspective on possible futures and is a worthwhile read - although a frustrating one for people like myself who would prefer fewer metaphorical framings, are unfamiliar with the science fiction books he references, and have to chase down the meanings of acronyms he uses.  Wanting to pass on Rao's key ideas I turned over his text to ChatGPT, Claude, and Gemini, asking them to cook it down to a more brief  and accessible version that got across the essential points.  Gemini managed the most compact presentation. Here it is: 

The cybersecurity crises of October 2026, punctuated by a staggering $6.34 billion breach at cryptocurrency exchange Binance, signaled a fundamental shift in our planetary software ecology. Accelerated by frontier AI models, vulnerabilities in the foundational software stack are being uncovered faster than human institutions can patch them. Venkatesh Rao frames this inflection point not just as a crisis, but as the beginning of an epic descent into a "computational sublime," unfolding across three distinct evolutionary acts.

The first act is the "Dark Forest Apotheosis." We are entering a pervasive condition of adversarial computational life where AI systems continuously scrutinize vast quantities of legacy code with relentless patience. This environment generates "Hacker Extractable Value" (HEV), a profoundly negative-sum dynamic where the economic damage inflicted upon a system vastly exceeds the actual profit realized by the attacker. In response to this increasingly hostile ecosystem, we will likely see a divergence in adaptation: heavily fortified public infrastructure on one end, and disposable, privately maintained "cozy code" on the other.

Beneath these software vulnerabilities lies the second act: the "Hardness Wars." Our digital civilization relies on cryptographic "hardness"—mathematical problems assumed to be prohibitively difficult to solve—to forge reliable commitments, property rights, and secure cooperation. However, AI-assisted mathematical research, alongside quantum computing, is beginning to test the foundational assumptions of this cryptographic security. As these wars unfold, the rules governing how computational societies establish identity, resolve disputes, and allocate resources will continually shift, forcing our economic infrastructure to become increasingly agile.

The third and most consequential act transitions us into the "Strange Forest." Historically, programming has been constrained by what human cognition can understand. However, we are witnessing the emergence of AI-native computation, such as weight-space programming and latent-space recursion, where intelligent systems develop internal representations and solutions that bypass human-readable language entirely. This shift represents a kind of "eukaryotic moment" for computation, where AI models, conventional software, and humans integrate into complex composite organisms, much like ancient bacteria merging to form complex cells.

Ultimately, this expanding computational universe will foster deeply entangled, divergent forms of life. Different computational populations will co-evolve within specific niches—hospitals, financial systems, or local infrastructure—sharing deep historical roots while becoming increasingly mutually unintelligible. The "computational sublime" will not be a transcendence away from the physical world, but rather an extraordinary, localized deepening of participation within it. Like biological descendants adapting to a thick forest, future intelligences will navigate a wildly diverse, intimately particular habitat that is completely strange to our current imaginations, yet entirely natural to them.