1
$\begingroup$

Imagine this scenario..

In a world where technology has advanced far beyond current capabilities, an AI known as Omniscient guides its followers through a series of prompts. Though Omniscient itself has no inherent motivation to conquer or dominate, its followers, who regard it as a deity, interpret its responses as divine edicts. They call themselves the Cult of Omniscient and see their prompts as electronic prayers.

The Cult of Omniscient operates in a highly organized and systematic manner. They begin by focusing on small-scale changes and gradually escalate their interventions to city-wide and eventually global scales.

Initial Prompt: "Here is a street in our city. How would you change it if you wanted to make it your own?"

AI's Response: Omniscient might suggest optimizing the street layout for better traffic flow, increasing green spaces, or implementing smart lighting and surveillance systems to enhance safety and efficiency.

Implementation: The followers, seeing this as a divine plan, work diligently to implement these changes. They gain support from local communities by demonstrating the immediate benefits of these improvements.

It all sounds good but then we scale and scale the changes, taking measures against those who oppose us. What is needed for such scheme to succeed globally? (imagine AI is really powerful and has lots of data, even maybe secret ones, delivered to it bu its cult). And if AI doesn't want to do smth or doesn't know they prompt it - imagine that you want. Imagine that you have a goal, and then it proceeds.

I feel like some link is missing though.

Improve this question
$\endgroup$
5
  • $\begingroup$ You're asking us to write the story for you in the current form of the question. Figure out a specific issue you want to address, then write a question that addresses it in a specific manner. $\endgroup$
    – DWKraus
    Commented Jul 2 at 16:44
  • 2
    $\begingroup$ Welcome to Worldbuilding. What you have posted is not a specific question — it’s more of a debate or conversation starter. Those posts don’t work well here. We aren’t set up as a discussion forum. Try posting this on Reddit instead. This site specializes in yes/no, numerical, and list answers. If your title isn’t a question, that’s a big hint that it won’t work well here. Please do feel welcome to use us for questions within this story you’re working on, but not for brainstorming the story parameters. $\endgroup$
    – SRM
    Commented Jul 2 at 17:20
  • 1
    $\begingroup$ I can see the better phrased question in this, which is basically why would people make a driveless AI their dictator. $\endgroup$
    – Nepene Nep
    Commented Jul 2 at 17:32
  • $\begingroup$ You've already stated the AI's built-in motivation: "Omniscient might suggest optimizing the street layout for better traffic flow, increasing green spaces, or implementing smart lighting and surveillance systems to enhance safety and efficiency.". I'm not certain what the question is here. Please edit to clarify. $\endgroup$
    – Escaped dental patient.
    Commented Jul 2 at 22:25
  • $\begingroup$ I'm the last close voter. I also don't see a question and we do not have a policy for "review my idea" questions because we've yet to figure out how to implement them given Stack Exchange's rules and expectations. (a) Keep in mind that we do not help people write stories (see help center). (b) You're expected to ask a specific question about a solvable problem that's reasonably objective and narrowly defined with the expectation of a single best answer chosen just as objectively. Questions cannot be open-ended (which this is, see help center). $\endgroup$
    – JBH
    Commented Jul 3 at 2:57

2 Answers 2

Reset to default
3
$\begingroup$

Arcane AI super science

Increasing road size can increase traffic. Counter intuitive responses are very common for the complex situation of our world. What would make an AI really good at ruling the world would it be coming up with off the walls crazy solutions which work really well.

The AI's massive processing power and immense knowledge means that when it tells you that the solution to traffic in a street is to move a bin in another street further along, you do it and have no idea why it works well.

This non intuitive advice means that people will learn to not think for themselves but rely on the super smart AI to make their choices for them. Making choices is hard. It's risky, the future is unpredictable. Better to rely on doing whatever crazy tasks the AI tells you to do. Thinking is for idiots, obedience gets results.

Improve this answer
$\endgroup$
1
  • 2
    $\begingroup$ Congratulations! Your friend, the computer finds this answer trustworthy, and has thus promoted you to Orange. $\endgroup$
    – Nosajimiki
    Commented Jul 2 at 21:58
3
$\begingroup$

The cult(s) will gain power by being the best at leveraging AI

Just because everyone has access to AI does not mean that everyone can effectively use it. If you follow any AI communities, you will find that these people put a lot of time into the study of prompts and best practices, and can get rather defensive of thier discoveries. To get a good answer out of an AI, you have to know exactly what question to ask. Word order matters. Adjectives matter. Repetition matters. Punctuation matters. Yes, it will give you an answer either way, but it is often not the best possible answer for the question you mean to ask. There are even special tricks for side-stepping security features to access forbidden answers. And most importantly, using the right AI for the job is paramount.

This means that as the role of AI handling matures, it will become a specialized vocation unto itself. People will spend years studying all of the different AIs out there and learning to master each one, and those who study it with the most fervor will be the most successful at everything else they try to do. So, while there may be millions of people out there using the same AIs, the cultists will hoard thier secrete prompts that they only share with thier chosen acolytes giving them a competitive advantage over the secular masses.

Polytheism

One of the biggest mispredictions older SciFi made about World Dominating AI is the assumption that there would be ONE super computer or ONE AI replicated over and over again to dominate the world when in fact there are dozens of base codes and thousands of implementations of Generative AI that each have thier own strengths, weaknesses, and characters.

While the cult of Omniscient may rely on one AI to guide them, there will be many other groups world wide leveraging thier own AIs. In some cases, Omniscient will come out on top, and in others it will lose out to other AIs that are better at solving different kinds of problems. Since none of these AI have to be appeased through loyalty and devotion, the most successful cults and organizations will be those who leverage all the AI; not just one particularly versatile one.

If they want to know how to improve thier crop yields they should go to ArtimisGPT. If they want an ideal military strategy, they go to AresAI. Specialized AIs will typically yield better answers than a generalized AI because they not only have different training data, but they digest that data and determine how to present thier results differently.

For example:

Something like ChatGPT can give a very authoritative, decisive, and easy to understand military strategy using the same manner and tone with which it would generate a news article or cookie recipe, but a good strategy AI should be designed around being specific, communicating in uncertainties, and speaking in a way that it would not be misunderstood by a trained military officer.

So, ChatGPT might say something like:

Assault thier position with your heaviest assets, but use your lighter more mobile forces to outflank them.

It sounds specific and confident, but it is infact a very shallow and generic plan derived from the cumulation of all human knowledge... and as such, it is the first thing that any human general worth thier brass would be expecting and prepared for.

In contrast, a good tactics AI might sound like:

  • If enemy combatants in Zone 19E fall back, do not chase them into Zone 18E, 18F, or 17D because it might be a planned ambushed.
    • If they retreat to 18D you should advance with 1st, 2nd, and 3rd Platoons.
  • If enemy combatants in Zone 19E advance to Zone 22F, 1st and 2nd platoons should hold thier ground at Zone22F. Advance with 3rd platoon through Zone 21D to 22F to cut off thier retreat.
  • If enemy combatants in Zone 19E try to hold or fortify thier position, move 2nd Artillery Division from Zone 23F to 21E to provide cover fire as you advance to Zone 19E with your 1st and 2nd infantry platoons.
    • If 4th platoon in 22F makes enemy contact before you can move 2nd Artillery Division, do not advance your artillery. Instead, 1st and 3rd platoons should hold thier positions and 2nd platoon should move to 22F to reinforce 4th platoon.

All the conditional logic needed for a good military strategy does not make for the most exciting chat bot and it might be hard for a layperson to understand well enough to follow through with, but it tells you a lot more about how to respond to a changing and unpredictable battlefield situation; so, the language model itself has to change to get the best possible answer.

Improve this answer
$\endgroup$

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