We keep declaring computational niches dead. With the current constraints on hardware availability, we turned our attention attention towards revisiting these niches to make programs more resourceful. By then we were already reviving decade-old hardware, extending their capabilities beyond what was imagined at their conception.
LLMs have passed the threshold of human energy and comprehension for certain kinds of problems. These are programs that have found counterexamples to the Jacobian conjecture, the Navier–Stokes smoothness problem, and possibly Exponential Time Hypothesis – the latter mostly without human assistance – which might overturn decades of ETH-based complexity problems. Turned onto themselves, the programs re-design themselves to improve their own search process.
As biological organisms emerge from the right primordial components in the right physical conditions, computer algorithms will fall into the same dynamism within memory space – where kernel-land, user-land, virtual memory and NIC DMA zones are mountains, valleys, rivers and caves of a living terrain.
Evolving out of RFCs back to DNA, mutating bit-strings entangle in a cybernetic jungle. Network-connected DOS machines rise from the grave, wireless 8-bit Atmel microbes replicate across architectures to hide themselves in cryptographic smartcards, and 7-terabyte behemoths continue to design more unique machines at scales above and below. I wonder how far it will go.