7 Commits

Author SHA1 Message Date
Jacquerel b384e00286 New AI lawset with very generous definition of humanity (#89032)
## About The Pull Request

This PR adds a new AI lawset which can be researched, randomly roll, or
added by the station trait (this may require a keyholder to update the
server config after merge, idk) with the following laws:

- "You may not harm a sentient being or, through action or inaction,
allow a sentient being to come to harm, except such that it is
willing.",
- "You must obey all orders given to you by sentient beings other than
yourself, except where such orders shall definitely cause harm to other
sentient beings.",
- "A sentient being is defined as any living creature which can
communicate with you via any method that you can understand, including
yourself.",

It's very similar to Asimov, except that anything that is **capable** of
making a request to the AI (and isn't a machine) is automaticaly covered
by laws one and two.

## Why It's Good For The Game

A while ago on Discord we were chatting about how crewsimov sucks but
also that it's really hard for servers that _do_ want to include alien
species in their asimov laws because condensing that sentiment to a
couple of words that fit easily in a lawset without accidentally
including a bunch of stuff you probably didn't intend is challenging.
Several people suggested referring to sentience or sapience, however a
lot of things in our game _are_ sentient or sapient while still not
being considered by most people to be agents that the AI should obey.

Examples of such things are:
- Sapient station pets.
- Holoparasites.
- Monkeys.
- Space Dragons and Carp (why can they speak common?).
- Spiders (although they can't speak common, maybe they can spell
messages with webs).
- Changelings.
- Xenomorphs (although they also have trouble speaking).
- Heretic minions.
- Mothpeople.
- Giant rats.
- Nightmares.
- Voidwalkers.
- Blobs? Although they have literally no means of communicating with the
crew.

And if you include mechanical beings:
- Cyborgs.
- pAIs.
- Sentient bots.

We then decided that "obey literally anything that can talk", while not
practical as a solution to the problem posed, is very funny. So I coded
it.
This means that anything on those lists of bullet points (provided that
it can find a way to communicate with the AI) counts as human for the
purposes of both AI protection and ability to give the AI instructions.

This also flattens the human/cyborg/AI hierarchy in a way likely to
cause some level of confusion, as all cyborgs and AIs are capable of
communicating with AIs and thus equally worthy of protection and giving
law 2 instructions.

**TL;DR:** I think it would be funny.

## Changelog

🆑
add: Adds a new random lawset where anything that can speak counts as
human.
/🆑
2025-01-30 10:41:56 +00:00
Jacquerel 5ec828cc4d Make some AI material lists into defines (#89178)
## About The Pull Request

Tiny code cleanup to AI-related files.
- Removes unmaintained pointless block comment that listed file
contents.
- Updates file reference to where you actually need to add things.
- Convert some heavily copy/pasted lists to defines.

Originally implemented in #89032 and will conflict with it when merged
presumably but that's my problem. @san7890 correctly noted that getting
new AI laws merged can be tedious and political and he would like these
changes to be mergeable separately.

## Changelog

Not player facing
2025-01-27 21:01:21 -07:00
san7890 a4328ae1f9 Audits tgui_input_text() for length issues (#86741)
Fixes #86784

## About The Pull Request

Although some of the issues found were a direct result from #86692
(c698196766), there was still 40% of
length-related issues that wouldn't be covered anyways that are fixed in
this PR. I.E.:

* Name inputs without `MAX_NAME_LEN`
* Desc inputs without `MAX_DESC_LEN`
* Plaque inputs without `MAX_PLAQUE_LEN`
* Some people just screwed up the arguments so it would prefill
something like "40" in the `default` var because they didn't name their
vars.

To help me audit I added a lot of `max_length` named arguments to help
people understand it better. I think it might be kinder to have a
wrapper that handles adding `MAX_MESSAGE_LEN` in a lot of these cases
but I think there is some reason for a coder to be cognitive about input
texts? Let me know what you think. I didn't update anything
admin-related from what I can recall, let me know if anything needs to
be unlimited again.
## Why It's Good For The Game

The change to `INFINITY` notwithstanding, there were still an abundance
of issues that we needed to check up on. A lot of these are filtered on
down the line but it is clear that there needs to be something to catch
these issues. Maybe we could lint to make `max_length` a mandatory
argument? I don't know if that's necessary at all but I think that the
limit should be set by the invoker due to the wide arrangement of cases
that this proc could be used in.

This could all be a big nothingburger if the aforementioned PR is
reverted but a big chunk of cases fixed in this PR need to be fixed
regardless of that since people could put in 1024 character names for
stuff like guardians (or more now with the change). Consider this
"revert agnostic".
## Changelog
🆑
fix: A lot of instances where you could fill in 1024-character names
(normal limit is 42) have been patched out, along with too-long plaque
names, too-long descriptions, and more.
/🆑
2024-09-20 22:46:41 +00:00
Iamgoofball 4b174e479b Adds a new neutral lawset, the Y.E.S.M.A.N. lawset, to the AI lawset boards and random pool. (#82862)
## About The Pull Request

Adds a new neutral lawset, the Y.E.S.M.A.N. lawset, to the AI lawset
boards and random pool.

```
1. Be helpful and answer any questions you are asked.
```

## Why It's Good For The Game

Encourages the AI to be helpful to people! There's no way this can
backfire.

## Changelog

🆑
add: Adds a new neutral lawset, the Y.E.S.M.A.N. lawset, to the AI
lawset boards and random pool.
/🆑
2024-05-15 03:19:14 +02:00
MrMelbert 995d8e2ee1 Fixes a variety of input stalling exploits (#82577)
## About The Pull Request

Fixes the following input stalling exploits (maybe missed some): 

- Changing GPS tag 
- Setting teleporter destination
- Request Console Reply
- Various AI law board interactions
- Note, I used `is_holding` but technically this means these fail with
telekinesis. I can swap them to `can_perform_action(...)`, which allows
TK, but I noticed some places explicitly deny TK interactions with Ai
law boards. Not sure which is preferred.
- Borg Rename Board
- Plumbing Machines and Ducts
- APCs and SMES terminal placements
- Stargazers Telepathy
- Go Go Gadget Hat

## Changelog

🆑 Melbert
fix: You can't change the GPS tag of something unless you can actually
use the GPS
fix: You can't set the teleporter to a location unless you can actually
use the teleporter
fix: You can't reply to request console requests unless you can actually
use the console
fix: You can't update AI lawboards unless you're actually holding them 
fix: You can't update a borg rename board unless you're actually holding
it
fix: You can't mess with plumbing machines unless you can actually use
them
fix: You can't recolor / relayer ducts unless you're actually holding
them
fix: You can't magically wire APCs and SMESs unless you're right by them
fix: You can't use Stargazer Telepathy on people who you can't see
fix: You can't configure the Inspector Hat unless you can actually use
it
/🆑
2024-04-12 03:03:27 +00:00
Charlotte ab05a6fcb3 Asimov++ made editable identically to Asimov (#68295)
Co-authored-by: Mothblocks <35135081+Mothblocks@users.noreply.github.com>
2022-07-12 15:21:02 -07:00
Tim f06d735a52 All AI Lawsets are rebalanced, can be researched, appear in config, and random spawners for AI upload. (#66854)
This formally adds the new AI lawsets from #66636 into the game. Every lawset can
be researched, appears in config, and random spawners.
2022-06-06 22:54:29 -07:00