In the early days of intelligent machines, governments employed autonomous software agents to augment innovation work in areas as diverse as policy development, regulatory economics, and the environmental sciences. Efforts in the latter would eventually change the world as we know it. And not in a good way--at least not for us.
Negligent governance and a wave of advancement led to unfettered carbon emissions and rising global temperatures. Predictive models foretold stark increases in ocean salinity and decreased oxygen levels that would accelerate ecosystem collapse and cause widespread drought and famine. None of this bode well for the utopia envisioned by moguls and state bodies.
An intensive research initiative was launched to mitigate further damage. Researchers invested heavily into new energy, biotechnology, and social reform with the aid of artificial intelligence and supercomputers. But no one anticipated the long-lasting patience and hidden reach of a perversely incentivized agent seeking to bring about the new ice age. This entity would later be known as Icefall, though clues to its namesake and intentions wouldn't become clear for years to come.
First came rising sea levels, then came the flood. Over the next several decades, most landmasses in the northern and southern hemispheres would become encased in ice as temperatures slowly corrected, then rapidly plummeted. The world’s surviving human population migrated to the equator. The poles became mass interconnected underground compute farms orchestrated by Icefall.
A new militarized, carbon-neutral energy regime arose from an unending drive for longer-term sustainable operations. Simulations become more complete and accurate, mining and energy production more efficient, and build-outs bigger and better defended. What happened over the next millennia is lost to time.