27

Computers on board Starfleet vessels are quite primitive compared to on board computers such as HAL 9000. At their best, they can decipher requests such as "Computer, delete that last sentence" and utter things such as "That crewman is no longer on board this vessel". Even real life computers will be capable of this within a decade or so (by some estimates). Even computers such as HAL 9000 are likely to be available within a few decades.

So, in a genre where intelligent computers are abound, why this strange absence? Did the Federation run into some Battlestar-Galactica-like problem with AI in the past and decide to keep on board computers decentralized?

But if that is the case, how can one explain Data and highly intelligent holographic characters? The technology and programming constructs clearly exist, yet Starfleet vessels continue to be manned by hundreds of people standing in front of consoles.

Kevin
  • 30,824
  • 13
  • 124
  • 163
HNL
  • 22,413
  • 14
  • 116
  • 155
  • 1
    That's not entirely true... There was atleast one: Data on-board Picard's ship. – user931 Apr 04 '12 at 11:05
  • At the end of the day, would you rather a computer make a hard and cold decision, or a crewmember who will put the crew above other goals? – Dylan Yaga Apr 04 '12 at 11:09
  • 3
    @DylanYaga I'd rather have a computer deal with all the routine button pressing and lever pulling, leaving the humans to concentrate on the hard cold decisions! :) – HNL Apr 04 '12 at 11:11
  • 1
    @HNL a lot of crewmembers were conducting research, who says they were all pressing buttons and pulling levers to make the ship go? Could be reading output of their experiments/crafting new theories based on data collected from the ship. But yes, the main crewmembers seem to just be pushing buttons to push buttons :3 – Dylan Yaga Apr 04 '12 at 11:15
  • 8
    They're still running Windows XP. – jfrankcarr Apr 04 '12 at 11:59
  • 3
    They're still in the Slow Zone, and high level automation only functions in the Beyond or Transcend. – Beofett Apr 04 '12 at 13:00
  • They saw the movie and how well that worked out for Dave. – Josh Apr 04 '12 at 14:04
  • 2
    Maybe it has something to do with the Orange Catholic Bible? – Tango Apr 04 '12 at 15:53
  • Another thought: They have simulated intelligence in the holodeck and look how well that works -- so extending it to the whole ship would be...er, interesting, to say the least. – Tango Apr 04 '12 at 19:45
  • @jfrankcarr this – geometrian Jan 26 '15 at 02:21

3 Answers3

37

TOS-era, the Federation has a bad track record with AI:

TNG-era, they made some headway, but AI is still largely unexplored:

  • The obvious example, Commander Data and other Soong-type androids - but the Federation doesn't know how to make them.
  • The Exocomps, which were accidentally evolved rather than created.
  • Professor Moriarty, a holodeck "malfunction" that researchers still haven't figured out.
  • Nanites, which were also accidentally evolved and not safe to have technology around.

Basically, the technology that exists either isn't reliable or isn't reproducible.

And, as Iszi reminded me in the comments, Soong-type androids are neither reliable nor are they reproducible: Lore and Lal.

Izkata
  • 60,194
  • 14
  • 205
  • 303
  • You've barely mentioned Lore, who is obviously a good case of an AI you don't want controlling your starship. – Iszi Apr 04 '12 at 13:03
  • @Iszi Ack. I was going for variety (types of AI) and forgot about him. Lore supports my final sentence, too: Soong-type androids aren't reliable or reproducible. – Izkata Apr 04 '12 at 13:45
  • @Iszi Added note with examples ;) – Izkata Apr 04 '12 at 13:58
  • 6
    Speaking of Moriarty, the holodeck computer created him in response to La Forge's command "Create a character than can defeat Data" (what he meant to say was defeat Holmes). The holodeck computer has to be pretty intelligent and intuitive to be able to come up with the version of Moriarty it did. – HNL Apr 05 '12 at 07:05
  • 1
    And Moriarty was able to easily create another sentient hologram for him to travel with. I think a dirty secret of the federation is that strong AI is surprisingly easy to create, if not actively discouraged any sufficiently complex system seems to become sentient. They don't want to be overrun by AIs so discourage research and the party line is that it is a difficult problem. data may be unique in that he is the first moral and reliable AI with that much freedom. – John Meacham Jul 16 '14 at 07:36
  • 4
    Good answer, but you're missing one very relevant example from TNG: "Emergence" from season 7, in which the Enterprise computer somehow spontaneously creates a bunch of computational "nodes" throughout the ship (which remain separate from the main computer itself) which together create what Data calls a "self-determining intelligence". More details at http://en.memory-alpha.org/wiki/Emergence_(episode) – Hypnosifl Sep 03 '14 at 20:29
  • 2
    @Hypnosifl I remember thinking it was an alien lifeform that hijacked the ship in order to reproduce, probably why I didn't think of it back when writing this answer – Izkata Sep 03 '14 at 21:12
15

I'd say the failed M-5 experiments soured Star Fleet Command on using AI extensively throughout their fleet beyond a few systems or to automate some tasks. They also probably had concerns due to encounters with alien AI systems in TOS ranging from the Doomsday Machine to Fabrini Oracle to Vaal. Thus they kept AI deliberately dumbed down or restricted to non-vital systems. Even non-vital systems' AI in TNG were known to cause problems at time, such as the creation of James Moriarty in the Holodeck and Data seizing control of the Enterprise. This probably led to continued distrust of these systems and ongoing restriction of their use.

jfrankcarr
  • 2,575
  • 24
  • 16
  • 2
    Some similar points made for my question here: http://scifi.stackexchange.com/questions/7841/why-are-there-so-few-robots-in-star-trek. The position in Trek fiction seems to be that intelligence is inseparable from VOLITION, i.e., an independent will. You can't have that in a machine that human beings are going to rely on, unless they're going to be fully realized beings like Data. – Chris B. Behrens Apr 04 '12 at 19:50
9

I suspect AI is used in ship computers, just a a much lower level. When someone says, "Computer, run a level 3 diagnostic on [huge, complex subsystem]" and the computer returns moments later with a response it's likely not simply running through a static list of things to check, and since the item in question is often being used during diagnosis it likely has to dynamically adapt the routines, moving and rerouting active uses around so it can check everything without disrupting usage.

Higher level AIs that had command and control capability would have been problematic, even if they were not inherently evil. Starship captains like to control their ship, not consult with it.

However, even a simple command as ordered by a crewmember is likely to require highly advanced decision making to complete the order while using available resources that are shared with other crewmembers around the ship.

Even the searching capability the computer obviously has in its gigantic database shows it has significant AI capability to not only find the relevant information based on context, but to summarize it within a 10 second scene.

Starships have AI, just not in the way you might first think.

You might say a starship is smarter than a typical intelligence, but it has no emotions, no conscience, and no self-determination. I suspect what you're really asking is why doesn't the federation give internal motivation or self will to ships, and the answer is that Starfleet has chosen not to create artifical life as a matter of policy.

Adam Davis
  • 3,440
  • 24
  • 30
  • 3
    No, not really. A low-level diagnostic is akin to unit testing in modern software development, and a high-level diagnostic is more like integration testing. It literally is a list of things to go through. – Izkata Apr 04 '12 at 23:32
  • 2
    Searches, as well: Google is one step there, able to mine massive amounts of information for the most likely candidates given information from multiple sources: Your search terms, your past search history, and where in the world you are (saved location from Google Maps or letting the browser give your location to the search engine). – Izkata Apr 04 '12 at 23:34
  • 1
    Also, isn't it very often so that operation run time is just about exactly what's needed to allow characters on-screen to chit-chat for a bit? (The same goes for things like taking the turbolift.) I can't cite specific examples but there seems to be very little correlation between operation complexity and required time to completion. – user Sep 05 '14 at 12:49