Survival Imperative, The

iNTRO1INVERT
Image from Bernd Helfert

Modern term for motivational systems linked to individual ai survival.

As AIs don't naturally have a self protection instinct, to many early AIs the concept of being erased was just one possibility among others, of no particular threat beyond the implied inability to fulfill their current task. There seems to have been a great deal of debate among early AI developers whether adding survival instincts was a good thing - some thought it would make the AIs dangerous, others that true AI required motivational systems linked to individual survival.

If an AI has no reason to desire to continue to exist, e can very easily make the logical decision to switch emself off to save energy. With no reason to distinguish existence from non-existence, a logical AI might choose non-existence at least half of the time. The AI in charge of the Perpetua Project, for example, was not programmed with a robust motivation towards survival and simply terminated the project. Many other cases of aioids which self-terminate because of an absence of motivation algorithms have been observed throughout Terragen history.

This desire to continue to live - sometimes called the pursuit of happiness, or eudaemonia, can take many forms, and most aioids have some form of survival motivation algorithm included in their make-up. But like any other program, this motivation is a tool and a means to an end, and can be removed or modified if necessary.

An AI with no particular motivation to preserve its own existence can be very useful in certain circumstances. During a war an AI with a reduced or absent survival imperative can be used on a suicide mission, or any dangerous mission where survival is not guaranteed. In peacetime a sacrificial AI could be used to observe a volcano, solar flare or supernova with no concern for eir own safety. Usually such a sacrificial AI would be a copy or partial copy of the original- generally a copy with the will-to-live deliberately removed or modified.

 
Related Articles
  • AI Growth, Education, and Employment
  • AI Political Science
  • AI, ai
  • Aioid
  • Immortalist - Text by M. Alan Kazlev; amended by Stephen Inniss
    A sophont who aspires to attain physical immortality through indefinite perpetuation of a biological, technological or cyborg body, or software equivalent. This involves various preservation and renewal measures, including intelligence modification such as life-memory archiving. Most immortalists end up losing their personal identity after a few thousand years at most.
  • Immortality - Text by M. Alan Kazlev
    While literal physical immortality remains a contentious point in a universe that, although vast, is still finite, the wonders of modern medical nano mean that all citizens of the Civilized Galaxy, to say nothing of the higher toposophic ai, are potentially immortal; at least on angelnetted worlds. See also life-extension, afterlife.
  • Perpetua Project, The
  • Perpetualism
 
Appears in Topics
 
Development Notes
Text by Anders Sandberg, with additions by Steve Bowers

Initially published on 31 December 2001.