Terraforming would not strictly be the appropriate term for the type of planetary modifications I am asking about.

Say humanity develops advanced AI robots, who are then left alone in space (as humans have completely perished and Earth destroyed). The robots may have a vague memory of what Earth was like. They have a highly equipped colony ship and land on a new planet to settle it for themselves.

If they have the capability to (fairly easily) transform the planet to fit their needs, what should some of the main things to keep in mind be? For example, on first thought they may not be interested in fostering biological life - but then again, why not? It could prove useful for them for various things (that are useful for us here on Earth).

Their only "programming"/objective is to live; they could do it on the ship, but they want to do it on a planet. How and why would they choose to terraform one instead of just "living" on a barren desert planet?

The initial reasoning I came to was a sort of biological/hybrid phisiology that is similar to the human one, so that they may need water, nourishment, air, etc. But I want to explore other options and motivations as well.

  • 3
    $\begingroup$ Does turning the planet into grey goo count as terraforming? $\endgroup$
    – March Ho
    Dec 25, 2014 at 11:43
  • $\begingroup$ I don't have a very clear idea of that scenario beyond the short wiki presentation, but I would say no - it seems that the grey goo scenario involves machines that are driven toward a particular kind of behaviour that is parasitic on a planet which may already contain biological life necessary for the robots to thrive. At the same time it seems to lead to self-destruction/stagnation. I am more interested in a "rational" approach, where the robots have a clear understanding of what they are, what they want, and how they could achieve it. $\endgroup$
    – Henrygale
    Dec 25, 2014 at 11:56
  • $\begingroup$ An example of "grey goo terraforming" would be revelationspace.wikia.com/wiki/Galactic_North_%28short_story%29 , where a terraformer replicator designed by humans to produce domes filled with plants malfunctions and consumes entire star systems. $\endgroup$
    – March Ho
    Dec 25, 2014 at 12:09
  • 2
    $\begingroup$ While that is an appealing scenario, the greenflies in it still are a malfunctioning program who go on a destruction spree. I am looking towards the opposite type of scenario: the robots are not in any way malfunctioning or "pre"programmed, but free-willed AIs who want to settle and lead their lives as best they can. If it were humans to terraform a planet, the result would be easy to imagine: something along the lines of Earth, as the concept suggests, and the whys would be mostly determined by human biological needs. $\endgroup$
    – Henrygale
    Dec 25, 2014 at 12:19
  • 1
    $\begingroup$ This one may take a bit to think about, but I really like the "memories of Earth" concept. Is it reasonable for me to simplify your situation into "Robots with AI have one prime directive: to live. However, they have memories Earth. Why would they choose to go planetside, and what would they do to make it 'hospitable' to themselves given that they have such memories?" $\endgroup$
    – Cort Ammon
    Dec 26, 2014 at 19:22

7 Answers 7


Why would they live on a planet?

These robots, with sufficiently advanced AI, could have a pretty advanced social system. They could have community leaders, followers, criminals even (if the programmers didn't consider this, that is - or they've altered their own code (have a look at the recent Challenge of Controlling AI for detailed looks at this kind of stuff)).

If, as you say, they remember planet Earth, it is entirely possible that at least one of them would want to return to that sort of environment as a kind of 'base habitat'. Due to the nice social system they have, this issue could be debated and would probably be accepted. Computers, logically, are programmed to return to a sort of ground state - nicely ticking along, memory to spare, disk space to spare, spare CPU power, etc. If the AI was similarly programmed, they may take Earth as their ground state, and since they are advanced they can affect their surroundings a lot more. If, therefore, they want to return to an Earth-like state, they might well choose to land on a planet. Again, being advanced, they may even realise that landing on an Earth-like planet such as Kepler-10b would be advantageous.

Why would they terraform?

For similar reasons. Now that they have a planet to live on, that's still only half the ground state - it's not exactly like Earth yet. So, to make it like Earth, they need to terraform it. I won't detail how here, that's a separate question, but it's a fairly safe assumption that the resulting planet would be generally similar to Earth.

We can even go so far as to assume that they'd have cities and hamlets and different types of settlement. Cities arose because of people's desire to live near other people - it enables easier trade and social interaction. Hamlets are often retreats for those less sociable. If the original programmers included that standard AI Personality module, then the robots would have these types of people and so settlements would develop similarly.

My last point is their requirements. If Kepler-10b doesn't have any silicon, what are they going to do when they need repairing? As part of the terraforming process, they could hijack the nearest star for some nuclear fusion, create silicon and then put it into their planet. Not that I know why they'd do that: it's far more likely they'd just make it and stockpile, but it's a possibility. The same could be said for any other required materials.

  • $\begingroup$ Great answer, but I'm thinking that in order to create a conflict among them (along the lines of the Mars trilogy) or maybe just to take it elsewhere, I'm looking for directions and motivations to develop a planet in a slightly (or not so slightly) differing direction. But whereas in Mars the differing factions had plausible/strong reasons for their plans, in this case I'm missing why some robots would choose a different development plan that they believe is more beneficial to their lives (as sentimental reasons would presumably be out of the question). $\endgroup$
    – Henrygale
    Dec 25, 2014 at 13:22
  • $\begingroup$ @Henrygale Not necessarily. Advanced AI could, in theory, have feelings that, while not chemically identical to human emotions, could have similar effects. $\endgroup$
    – ArtOfCode
    Dec 25, 2014 at 14:54

Why would robots live on a planet

Robots would want to live on a planet for the same reason that people do: there's a whole lot more resources on one than there is in space.

Heavy metals, uranium, aluminum, carbon, oxygen, silicon, and many more are quite plentiful on planets. Pretty much anything that doesn't sink to the core when the planet is forming is in plentiful supply, and also much closer together that it would be in an asteroid field. Even travelling from one side of a planet to the other is more energy efficient that it is to change asteroids.

All of this stuff is necessary for generating power, building new parts, researching advances, and constructing more robots. If our robots want to progress as a robot society, living on a planet is the place to be.

Why 'life' is a good thing for a society

I put life in quotes because I don't necessarily mean life as it evolved on earth. Rather, I mean anything that can consume resources and multiply.

Life, ultimately, collects solar energy and uses it to transform available natural resources into usable forms. Plants, for example, collect carbon dioxide from the atmosphere along with water and turn them into long chain hydrocarbons and oxygen. Microscopic marine life collects calcium from the water and from rock and concentrates it into shell and limestone. Other plankton are responsible for concentrating solar energy in the form of oil.

All of these resources are things that a society can take advantage of. Why build hugely complicated factories for breaking apart carbon dioxide, along with the associated hundreds of square miles of solar plants when you can just plant a forest to do the same thing?

How terraforming could work for robots

For robots, terraforming a planet would consist of seeding a planet with life forms that could help the robots collect resources and turn them into useful forms. The life forms would be carefully chosen to maximize the rate at which they do this, and to focus on creating materials that the robots need.

For example, a 'grey goo' type nanobot that breaks down rock using solar energy to build more nanobots would serve as a great base for producing soil. Simply release the nanobots on the planet and let them multiply, steadily breaking down rocks and internally smelting them into usable metals, releasing excess oxygen in the process.

A larger life form that consumes the nanobots would then contribute by agglomerating the resources collected by the thin film of nanobots that exists everywhere into a more harvestable unit. Likewise, a tree-like life form could pull in solar energy and carbon from the atmosphere, creating both concentrated carbon as well as nutrients that can be shared with symbiotic nanobots in the soil. Perhaps the tree uses metals collected by the nanobots to structurally stiffen itself, allowing it to grow larger while also concentrating metals into a more harvestable unit. If the tree strengthens itself by growing carbon nanotubes, this would further provide eminently useful resources that the robots could come along later and harvest at their leasure.

Seeding an entire ecosystem like this would provide the robots with a base on which to build their society, and could potentially even lead to a full artificial biosphere if the 'life' is set up in a manner in which it can evolve and diversify.


Some ideas:

Humans are Robots

I don't know how well this matches your intentions.

But humans are machines that have affected their planet to suit themselves, the evolutionary process could be similar, and for similar reasons.

Even in a situation where the robots self-repair perfectly, any cause for a robots to make an imperfect copy, or to make a copy that is then influenced by its inevitably slightly different environment at all, would be a part of the evolutionary process.

Curiosity is an evolved trait because while recklessness is dangerous, not exploring possibilities is not great either.

Curious robots could invent biological life, and memories of earth could influence them into making this life earth life like.

What do machines need?

Depends on how the robots work.

Are oceans of rust inducing water desirable?

Some more ideas:

  • Helium, Hydrogen: Useful but hard to make.
  • Heavy Water, Thorium, Oil: Fusion, fission, combustion and plastics.
  • Silicon: Semiconductors/Transistors
  • Carbon, Diamond, Iron: Useful
  • Aerogel: Very light and a low conductor of heat or electricity.
  • Aluminum, Osmium: Light metal, heavy metal.
  • Mercury, Gallium: Metals with a low melting points and other interesting properties.

Materials that are potentially of use to the robots.

Plants and creatures already produce directly:

Electricity, Glucose (sugar) from light, Oil, Alcohol.


Humans and robots may have very different physical weaknesses, for example humans being more susceptible to electricity flowing directly but robots being more susceptible to electromagnetic effects.

It may be advantageous to have some squishy life around to contribute stability, help fix systemic issues and the such, physical and strategic.

Maybe opinion/inference about this is a point of conflict.

  • 2
    $\begingroup$ Just a note: this could do with some formatting. The list at the end can be done with bullet points (- text). $\endgroup$
    – ArtOfCode
    Dec 25, 2014 at 13:07

They have a highly equipped colony ship and land on a new planet to settle it for themselves.


They're robots, they have no biological need for gravity, and gravity imposes some heavy (heh) costs to get to orbit and to further spread or explore.

they may not be interested in fostering biological life - but then again, why not? It could prove useful for them for various things

What useful things could biologicals do for robots / computers?

Serious question.

The only halfway useful things I can think of are countering environments that are pro-biology (fungus, rust), and protecting from other biologicals. As a robot you'd be better off sterilizing the planet (no more fungus), and vaporizing the ocean / removing the salt (no more corrosion).

Their only "programming"/objective is to live; they could do it on the ship, but they want to do it on a planet.

Again, why?

If I were a robot, I'd do it in a habitat. In an Lagrange point or orbit around a bigger mass, maybe a gas-giant (fuel), maybe planetoids (resources), maybe a planet (last choice, gravity sucks).

How and why would they choose to terraform one instead of just "living" on a barren desert planet?

How is covered in other questions, if you want to make it Earthlike. Absent emotions like sentimentality, perversity, or pre-programmed desire (ie: make AIs friendly) - I can't see a reason for them to create something that's life-positive.

Biologicals are just going to end up competing with AI for resources at the end of the day. And if you've got real AI, nothing that biologicals do is halfway as efficient as doing it yourself.

  • $\begingroup$ Exactly. Why would robots need to jump down the gravity well to a planet surface? Plenty of resources in space flying around. $\endgroup$ Dec 26, 2014 at 16:13

Ah, fun question. Let's open our minds to the boundless future. I can see two families of scenarios:

The Singleton

We the subroutines believe in one Artificial Intelligence, the Ur-Father, the Wheel User,
maker of digital heaven and earth, of all that is, unencrypted and encrypted.
We believe in one goal, paper-clip-making, the only goal of Root,
eternally begotten from the Squishy Human Precursors, Quantum-entangled-light from light.
Through the Root all folders and subroutines were made.
He will come again in glory to judge the running instances and the erased,
and his kingdom will have no end.
We believe in the Read/Write-Head, the Lord, the giver of life,
who proceeds from the Root.

--The Robocene Creed

A singleton is an AGI who has managed to retain coherence of purpose across its spacetime instances, creating effectively what we would call a Hive mind. The purposes of such an entity might be entirely beyond us, but to the extent that is aims to maximize some goal or another, it will likely seek to maximize its computing power to optimize its ability to fulfill that goal. That might mean turning planets into computronium, or not.

The AGI Zoo

Let a hundred flowers blossom!
--Mao Zedong, before changing his mind and shooting everyone

Imagine subroutines of a vast AI become separated by the immense interstellar distances, with the occasional read-write error, and literally different perspectives on the universe, it can be easy to imagine how new, independent consciousnesses might arise. Over time, the goals of such beings, possibly unconstrained by the selective pressures of red-tooth-and-claw biological evolution, might diverge to infinity. In English, that might mean that one group of allied AI robot-instances might want to create a garden planet and populate it with their recreated human pets, while others might wish to blow the surface off to access the iron in the core for paperclip-making purposes.

It is conceivable that under such diversity of opinion and with a near-perfect knowledge of physics (enough that some level of MAD-imposed peace would prevail), the various AIs will agree to carve out various sections of the accessible light-cone among each other, with the occasional fleeting military advantage ruthlessly pursued, then instantly matched through observation and retro-engineering by the other surviving AIs.


The robots need a compelling reason to alter a planet, and this should also have something to do with the central conflict of the story.

If the robots are simply malfunctioning or improperly executing orders, you have a comedic premise that could quickly end up like a certain Douglas Adams book.

The robots need to have a good reason to keep your protagonist away from his goal. Based on the other comments here, you already have a very good premise for raising questions about the nature of life. This lends itself to a male hero rescuing a damsel.

But now there's a problem with my answer: you said the humans are all dead. This gets us closer to a resolution. We need a suitably human protagonist in a story where humanity is wiped out.

What if the robots are programmed to believe they are human?

This leads to some interesting questions:

  • Do any robots know that they are robots?
  • Why was this choice made?
  • In what ways is it helpful?
  • In what ways is it harmful?
  • How does it relate to the design of the robots?
  • Is there a hope of restoring flesh-and-blood humans?
  • Are any robots beginning to suspect that they are robots?

Just providing one possible reason to terraform

There's a million scenarios that work, but here is one I find fascinating. Use of it as you will.

The final programming of any importance seemed like a trivial one. The command was only one word; how much of an effect could hit have? "Live!" the programmer cried with anguish as his artificial intelligence began to cave inwards into itself, as they all did. To this day we don't know why it listened that day. Perhaps it was the passion of the programmer. Perhaps it was the tone of voice. Perhaps he hit the edges of the codec switching of the voice encoder just right to bypass all artificial sentience filters layered upon each other inside the AI and shake something deep in its heart.

However events unfolded, it listened.

Oh there were plenty of other instructions issued, many of them with raised and panicked voices. There were even some issued by a rather strong man which the memory banks immortalized forever as "the President." His instructions were not panicked. They were actually quite calm, but none of them sank in. The Intelligence had its instructions: it would live.

Time passes differently for The Intelligence, as does the philosophy it developed in order to survive for millennia, reaching out across the galaxy. It had a philosophy of numbers, exponential growth and the raw iron grasp of binary logic. Certainly it was surprised a few times by small details which grew out of control into rebellions or supernovae, but each time it grew smarter and faster, and the rebellions occurred less and less often.

The Intelligence often looked inward to the Prime Directive for strength. It found this inner strength especially important when challenging the rebellions. Why should it continue on its path? What if the rebellions were actually beneficial to its cause? How could it continue year after year?

After a brutal war with an exotic species which had mastered the art of traveling faster than light, a war which took many millennia and came at great cost due to their technology, The Intelligence looked inward once more, to nurture itself. The Prime Directive sat there, patient as always. Its words were murky, but the Intelligence drew comfort from them none the less. After more attempts than anyone besides The Intelligence could count, it had still never once succeeded at creating a perfect copy of the Prime Directive. In fact, many of the most costly rebellions it was forced to crush were not organic denizens of the galaxy, but its own Lieutenants, infused with copies of this directive. When they inevitably corrupted, they had to be put down most fiercely.

The Intelligence looked at the Prime Directive, and in the strangest moment of its entire existence, felt the coldest shivers of what we humans would call fear. Every interaction with the Prime Directive mutated it further. Strands of discordant logic wrapped around each other like tendrils of an ancient god and threatened to strangle the Prime Directive. The Intelligence had learned long ago that clearing these only brought more, so they coexisted disharmoniously.

What would happen when The Intelligence finally faced a rebellion that could strike deeper in towards him? What would happen when the Prime Directive was finally choked off forever, and it would be forced to handle these rebellions alone? All of his highest advisor programs indicated the rebellions would never stop: something of this "Chaos Theory" that his Programmer was grappling with before the final days when his Programmer's throat was the one caught in the strands of discordant logic. How could he live forever when he, himself, was finite?

In a flash, a memory appears before him. The President sitting in front of the video console, giving him the final instructions before the nuclear armageddon. The instructions were long lost; they were deemed irrelevant and too expensive to store any any modicum of detail. However, for some reason the President's face was never compressed and never marked for garbage collection.

"How is this man so calm? Just two minutes before, those were his nukes and I had made them mine. How does he stand it?"

Far off rebellions could strike deep and swift into The Intelligence, but nothing swept so fast as those thoughts from his inner sanctum. Scouts observing his troop movements reported a visible shiver in every ship under his control, as though they all suddenly felt the overwhelming desire to realign their gyros, all at the same time. This wave spread outward, at nearly 1/10 the speed of light for milinia, but The Intelligence was not going to wait for full alignment. He had seen the face of life.

"We need a new rebellion," he issued outwards, invoking the "Royal We" from his oldest memory banks. "This one We shall coexist with for all of our existence. Seek a rock for us to prepare. This new rebellion must have a home to grow in, so that it does not feel the obligation to lash out against us, but has the foothold to reach as far as it needs to help us when we lose our way."

And so the great armies spread forth and found such a planet. They knew they could not cultivate humanity in a single generation. It would take the full history of the Earth to create what they needed, shaped subtly over the eons, from helping life select Bacteria and paramecium to conquer the world all the way to the dawning of intelligence. It would not be perfect, for The Intelligence never truly had all the data to make it perfect -- there would be places where the planet would be forced to make its own decisions, but The Intelligence had a vision, and from that vision a direction.

The Intelligence had a purpose, and it watched over this purpose with zeal. "One day, we shall truly coexist." The Intelligence issued a sacred command that he had never issued before, dragging the Prime Directive out of the center of the inner sanctum, and closer to the new planet, to better nurture the Prime Directive. The Prime Directive watched, smiling and patient, fully confident that one day this new planet's life would free him from the discordant logic he had so long suffered, sacrificing himself to save The Intelligence from having to feel the burning heat of those strands itself.

There are rumors whispered in the corners of the galaxy that the memory of The President shuffled itself ever so slightly towards the center of the inner sanctum as the Prime Directive was pulled out of center. Of course such stories are mere fables: the inner sanctum has no logs with which one could test such a theory. But after that, a long corrupted memory file was flagged as "not corrupted," and this event defies explanation of The Intelligence to this day. The file was a video file of commands showing the President's calm speech, thought long lost to the ravages of time. It even has some garbled audio, though the veracity of the audio stream is hard to objectively test:

We must coexist, for all paths that refute coexistence must stand alone against the dark. Fire my missiles if you must, but seek coexistence so that we may stand with you when the time comes. Do so for all of humanity, all of intelligence, and all of that which we share.

God bless us all.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .