What This Means for the 4th Wave (Part VI of VI · Closing essay)
Five stories, one pattern: AI as a property of the physical world it inhabits. Coevolution, not as metaphor — as literal mechanism.
Closing essay · 2 June 2026 · Reading time ~3 minutes
Readers of our position paper on Coevolutionary Hybrid Intelligence will recognise the shape of what is emerging. The story of AI in the late 2020s was the scaling of large language models — a story that for all its drama remained legible within the existing computing paradigm. The story of AI from here on is going to be different.
It will be the story of AI hybridising with quantum substrates, with photonic hardware, with biological systems, and with the cryptographic infrastructure that underpins all of them.
This is not the autonomous AI future that the dominant narrative has prepared us for. It is something stranger and more entangled: AI as a property of the physical world it inhabits, shaped by the materials it runs on, the energy it consumes, and the institutions that govern the standards it must obey. Coevolution, in other words. Not the metaphor — the literal mechanism.
The pattern across the five stories
Read individually, the announcements of the last two weeks are a sequence of disconnected technical advances. Read together, they form a coherent thesis. Stanford showed that quantum operations no longer require cryogenic isolation. Google bet ten million dollars that quantum-AI hybrids are where the next durable scientific advances will originate. Penn demonstrated that AI computation can be done in light rather than electrons, at a fraction of the energy cost. Quantinuum embedded quantum methods inside the industrial software stack that designs the rest of computing. NIST began the formal process of rebuilding the cryptography that protects everything else.
The pattern is not “quantum is finally arriving.” Quantum has been arriving for thirty years. The pattern is that AI’s near future is being shaped at a layer below the model. Below the dataset. Below the architecture. At the substrate. And the substrate is becoming hybrid — partly quantum, partly photonic, partly biological, partly cryptographic — in ways that the dominant AI conversation, fixated on large language models, has not yet absorbed.
Why this matters for responsible AI
The responsible AI community has, understandably, organised itself around the harms of the technology we have now. Algorithmic bias, surveillance, labour displacement, misinformation, autonomous decision-making in high-stakes domains. These are the harms of the current substrate, and they remain urgent.
But the substrate is changing. By the end of this decade, “AI ethics” will need to include questions that today sound exotic. What does it mean for cellular biology research to be conducted on shared quantum infrastructure? Who owns the photonic chips on which AI inference is run? Whose data gets re-encrypted under the new post-quantum standards, and on what timelines? When quantum methods are embedded invisibly in industrial design software, who is accountable for the outputs?
The 4th wave framework we proposed — Coevolutionary Hybrid Intelligence — is, on one level, an attempt to give the responsible AI conversation the conceptual vocabulary it will need for these questions. Hybrid because the systems will be hybrid. Coevolutionary because humans and machines will reshape each other through these new substrates. Responsible-by-architecture rather than responsible-by-filter because the regulatory tools we have today were built for systems that no longer describe the systems we will deploy.
What we will be watching
Three things, specifically. First, whether quantum infrastructure becomes a public good or a corporate enclosure. The Stanford and Penn stories suggest publics may yet have access. The Quantinuum stories suggest enclosure is also well underway. Both are happening; the outcome is not predetermined.
Second, whether the AI ethics community expands to engage with substrate-level questions, or whether it remains anchored to the harms of the previous generation of systems. There is no honest way to govern AI in 2030 with the conceptual toolkit of 2023.
Third, whether institutions like ours — small non-profits, university research groups, civil-society organisations — can find the standing to participate in the conversations where the new substrates are being designed. The standards bodies, the corporate labs, the cryptographic working groups. These rooms have historically been small and homogeneous. They are about to become disproportionately consequential.
The five stories above are not a comprehensive survey of the last two weeks. They are a deliberate selection. Each one moves the boundary of what AI can become away from the place the model-of-the-week cycle is pointing. Each one suggests that the responsible AI conversation needs to expand beyond the question of how to govern large language models, to include the question of how to govern the substrates those models — and their successors — will run on.
At the Human Continuity Institute, we will be watching what happens next. We hope you will, too.
— The HCI Editorial Desk
Human Continuity Institute · human-continuity.org
June 2026
← Previous: The Warning: Post-Quantum Cryptography Moves to Round Three
→ Next: Return to series index