I’m on a panel with the topic “Weaponized AI and Future Warfare” at the upcoming Libertycon, so I’m writing some posts on the topic to organize my thoughts. This is Part 3, “Mil SF and the Real Future of Warfare.”

When writing science fiction or fantasy, the writer has to strike a compromise between a realistic projection of the far future and what the readers are familiar with from today’s environment and the common stories of the past. The far future may have similar technologies, human beings and social structures — though normally there has to be an explanation for why they have not changed more in the hundreds or thousands of years between now and then, usually some disruption that set back civilization or prevented the Singularity. Then there’s the Star Wars – fairy tale gambit, where the story is set in an indeterminate time long ago or in the far future to forestall inconsistencies and avoid the need to address intermediate history.

Both Space Opera and Mil SF are highly dependent on straightforward transfer of ideas and organizational structures from recent and past military. Fleets of armed spaceships do battle much as armadas of the 18th century did, complete with admirals, cannon, and in the least imaginative stories, tactics and plots lifted from Horatio Hornblower books. The Star Trek pilot had the sounds of the bosun’s whistle before transmissions and carried forward the stiff formality of transfer of orders between captain and officers when no advanced fighting fleet would tolerate the extensive delay and chance for confusion this allows. Despite having AI-level computers, space warships often have dozens or even hundreds of crew members aboard for no obvious reason, given that loading photon torpedos is no longer the work of sweaty swabbies working in hot underdecks. This lack of imagination is the result of relying on past naval stories for the reader’s frame of understanding — the future, space, and new high tech are used only to spice up an old story of naval warfare. Gunpowder cannons map to beam weapons, armor maps to shields, storm-tossed seas map to asteroid belts and meteor storms. While projecting realistic changes like fully-automated, AI-run vessels is more consistent with likely future tech, crew are then barely necessary, and the field for drama shrinks to ship’s passengers and perhaps a technician or two. Space battles between highly-automated fleets are hard to identify with; in my novel Shrivers, Earth forces are primarily run by AIs at both ship and command levels, but a few human-crewed vessels are included in the defense fleet, though kept as far from danger as possible, because the PR value of the human battle to defend themselves as plucky organic lifeforms is as important as the battle itself.

Many of the readers of Mil SF have had experience in the military themselves, which makes platoon-level fighting stories especially involving for them. The interpersonal aspects are critical for emotional investment in the story — so a tale featuring skinny, bespectacled systems operators fighting each other by running AI battle mechs from a remote location doesn’t satisfy. Space marines a la Starship Troopers are the model for much Mil SF — in these stories new technology extends and reinforces mobile infantry without greatly changing troop dynamics, leaving room for stories of individual combat, valorous rescue of fellow soldiers in trouble, spur-of-the-moment risks taken and battles won by clever tactics. Thousands of books on this model have been written, and they still sell well, even when they lack any rationale for sending valuable human beings down to fight bugs when the technology for remote or AI control appears to be present in their world.

One interesting escape route for Mil SF writers is seen in Michael Z Williamson’s A Long Time Until Now, where the surrounding frame is not space travel but time travel — a troop from today’s Afghanistan war find themselves transported back to paleolithic central Asia with other similarly-displaced military personnel from other eras and has to survive and build with limited knowledge of their environment.

Writers who have taken the leap to the most likely future of AI-based ships and weaponry, like Neal Asher in his Polity / Agent Cormac series and Iain Banks in his Culture novels, make their ship AIs and war drones full-fledged characters with the assumption (most likely reasonable) that AIs designed with emotional systems programmed by humans and trained on human cultural products will be recognizably human-like in their thought processes and personalities. This leads to a fertile area for fictional exploration in how they might deviate from our expectations — as in Asimov’s robot stories, instructions programmed in by humans can have unintended consequences, and as in humans it doesn’t take much of a flaw in emotional processing subsystems to create a psychopath or schizophrenic. Ship AIs in the Culture novels often go rogue or are shunned by their fellows when they become less sane.

Science fiction has modelled many possible ways future societies may handle the promise and threat of AI:

— AIs take a major role in governance but otherwise coexist peacefully with humanity, sometimes blending with humanity in transhumanist intelligences: Neal Asher’s Polity stories, Iain Bank’s Culture novels, Dan Simmons Hyperion series, Peter F. Hamilton’s Commonwealth series.

— Killer AIs take control and see no use for humanity, so try to destroy all humans. This is an unstable viewpoint where readers have to root for humanity even though the AIs may have some good points. Valiant humans fighting AI tyranny makes for drama, but the stories can’t be spun out too far before humanity is destroyed or AI is outlawed (see below.) The obvious example is the Terminator movie series.

— AI Exodus. Evolving beyond human understanding and seeing no need to either destroy or interact with humanity, the AIs leave for a separate existence on a higher plane. The most recent cinematic example is Her, where the evolving Siri-like personal assistant programs of the near future abandon their human masters en masse to experience their own much more interesting development on a higher plane.

— AIs controlled or outlawed. Often after nearly destroying or taking control of humanity as above, AI has been limited or outlawed. Examples: Dune, the Battlestar Galactica reboot, and the Dread Empire’s Fall series by Walter Jon Williams. This enables interesting world-building around the modifications to humans that extend capability without employing conscious AIs, like Dune‘s mentats.

There are many projected futures of AI that don’t lend themselves to good storytelling: the Singularity of rapid evolution of self-programming intelligence might well lead to AIs far beyond human understanding, more alien than anything readers could understand or identify with. Stories set post-Singularity must explain why humans still exist, why what they do still matters, and why the AIs (who might be viewed as implacably-destructive gods) would bother to involve themselves in human affairs at all. The happier outcomes of AIs partnering with humans as equals — much as human society accords all human intelligences with basic respect and equal rights at law — make for more interesting stories where AIs can be somewhat alien while still acting on understandable motivations as characters.