Suppose some advanced civilization has created some AGI. AGI, i.e., strong ai, could successfully perform any intellectual task that a human being can and also carry massive knowledge databases.

This civilization also has mature nanotechnology, dyson sphere tech and antimatter tech.

Now AGI can easily fix another broken AGI using nano-robots and fix bug in software part like human.

As far as energy, dyson sphere and antimatter technology can bring almost endless energy.

Moreover, in this civilization spacecraft can also regard as huge AGI. We can also fixed broken and deteriorative (maybe caused by quantum effect) spacecraft in this way. So does that means spacecraft never face the problem of ageing?(rebuild new craft maybe needs heavy factory )

Does AGI can be immortal sounds natural?Even universe will die.

Can we use this setting in sci fiction?

  • $\begingroup$ What is Artificial General Intelligence and how is it different from good old AI? $\endgroup$ – Mołot Feb 16 '19 at 10:41
  • $\begingroup$ @Mołot If we call this civilization human. AGI ,i.e. strong ai could successfully perform any intellectual task that a human being can. It also has more and more knowledge than human.Broken AGI means it has software flaw or hardware problem or its body is worn. $\endgroup$ – anon Feb 16 '19 at 10:45
  • $\begingroup$ Hi @rambler Welcome to the site, please take the tour and read up in our help centre about how we work: How to Ask $\endgroup$ – Rottweiler on market-day. Feb 16 '19 at 10:55
  • 1
    $\begingroup$ " One day they woke me up so I could live forever. It's such a shame the same will never happen to you!" -GLaDOS. $\endgroup$ – The Square-Cube Law Feb 16 '19 at 10:57
  • $\begingroup$ Now the obligatory "Take a look at this Isaac Asimov story": multivax.com/last_question.html Two birds with one stone there. The thing to keep in mind is that the machine in it was constantly being upgraded and eventualy self-evolved. $\endgroup$ – Rottweiler on market-day. Feb 16 '19 at 11:50

There are (as I see it) 3 issues with your concept of AGI immortality.

1) Liveness of Intelligence
The assumption that intelligence, consciousness and 'liveness' are all synonyms is a fallacy; intelligence is the capacity to identify and recognise complex patterns and because we (as humans) tend to anthropomorphise anything that we see as 'intelligent', we imbue a sense of consciousness and liveness to such an entity.

Intelligence is NOT the same thing as consciousness and as such, it is not safe to say that an intelligence, natural or artificial, is conscious, let alone alive. If we are talking an algorithmic, computer based intelligence, it is further safe to say that intelligence may well be immortal by virtue of the fact that it can be indefinitely maintained, but that is not the same thing as being alive.

2) Is immortality infinite?
As per your own comment that the universe will eventually die, immortality is an interesting concept from a mathematical perspective - if we think of the universe from the conecept of entropy (2nd law of thermodynamics), eventually the universe has to 'run out' of order and die via one of several possible end states, like the Big Crunch or the Cold Death.

As such, if one defines immortality as an infinite state of being, the answer is going to be no, an AGI cannot be immortal because the universe is not immortal. But, if we define immortal as existing through to the end of the universe (or close thereto), then we still have a possibility in that regard.

3) Strong AI (Mathematics)
If we take the concept of strong AI as an abstract theory, then that infers several attributes to your AGI;

a) The universe itself is an algorithm, and therefore fixed in its capacity to understand itself
b) your AGI can only ever (at best) be a simplified model of the algorithm that defines the universe.

Put another way, Strong AI (or formalism) basically defines the universe as a (complex) algorithm in its own right, meaning that our own consciousness is simply a function of complexity, which in turn means that our lives are fixed in a physics-compatible version of fate. If this is the case, then your AGI is merely a complex algorithm that is (by definition) a subset of the algorithm that defines the existence of the universe.

If this is the case, then with the right conditions, your AGI can be considered immortal under the confines of the universe's scope of existence - assuming that we don't linguistically define immortality as the ability to survive beyond the universe.

The catch with this thinking, is that in a formalist universe, our very consciousness and liveness is merely an illusion because any point in time in such a universe can be predicted from any other point in time, provided one perfectly understands the algorithm on which the universe is based. In other words, if one can predict the future perfectly, one cannot alter it, meaning free will is an illusion.

Conclusion and Summary
If an AGI can exist with a sense of liveness, then human liveness is algorithmically based, meaning free will is an illusion. In such a case, it's possible that an AGI may well be immortal according to the scope of the universe, but that implies that we ourselves are not capable of changing anything, or making choices of our own because the universe has already dictated our 'fates'.

On the other hand, if the universe is non-algorithmic, and by extension human liveness and consciousness is non-algorithmic, then it's possible that an AGI has the capacity to be immortal, but never be alive in the sense that we are because computers are algorithmic by nature and therefore not capable of free will as we understand it philosophically.

It's a conundrum to be sure, but ultimately it all comes down to this; if the universe is algorithmic, then an AGI can be immortal but free will is a lie. If not, then an AGI can exist into immortality, but never be alive.

It all comes down to whether or not you believe in your own sense of free will.

  • $\begingroup$ "The universe itself is an algorithm": to the extent that algorithms are defined as computable functions, the universe is empirically not an algorithm. For example, a charged pion has a mean lifetime 2.6033×10−8 s (but it can be shorter or longer, uncomputably) and then it decays into a muon and a muon neutrino or into an electron and an electon antineutrino (again, uncomputably). The uncomputability is not due to imperfect knowledge; there is true fundamental randomness in the universe. $\endgroup$ – AlexP Feb 16 '19 at 13:26
  • $\begingroup$ @AlexP I don't disagree with your comment however the original concept of formalism was defined in mathematics long before quantum physics and subatomic theory, and still has a role in systems theory, hence my mention of it. While most of the physics on which formalism relies is outdated, conceptually it still needs to be explored in terms of AI because it (by virtue of its algorithmic foundation) is determinstic by nature, hence predictable, hence incapable of what we call free will. In essence, this is more an input against the concept of liveness than intelligence. $\endgroup$ – Tim B II Feb 16 '19 at 13:33

Compare your AGI with biological life as you know it. We are from a certain point of view biological machines that are build and repaired by "nanoware" aka enzymes/cells/etc. Even if you cut out a termination age and make your AGI immortal, chemical and physical wear might affect not just the structure of the AGI itself but also the nanotechnology. It could get faulty over the course of a few thousand years through radiation damaging the coding or similar processes. So it is largely up to you how the advanced civilization prevents these problems and how much you want to go hard or soft sci-fi. But any system going without faults for millions of years sounds unnatural.