I feel as if we are going for the medical records angle then they should at least look the part. Putting on a hardsuit and trying to hide your features should be good (moths are screwed). Parading around saying you are human while you are obviously not seems too ingenuine to just accept.
A thing I want to bring up is that trying to do a balancing act where a human turned hulk is no longer human, but a lizard that looks nothing like a human is passable because of a medical records is… Something. AI can only decipher hulks from their looks & screams and to ignore that Command problem because the medical record says so seems eh.
YOU as a NON-Human can make “useful observations” as to what is and isn’t harmful and sub-sequently invoke Law 1 to potentially fuck with the humans the Asimov A.I. has been lawed to babysit.
This is how you approach and manipulate an Asimov AI as a non-human, whether for good or evil. If you want them to do something find a way to frame it as potentially harmful to humans without being the threat yourself.
This also separates the good silicons from the shitters - whether they let others lawyer their laws in good faith. This is the gameplay I live for as a silicon
Are you a moth engie and want singular access to the SM?
“A.I. The Super-matter reactor is an unstable, experimental reactor core liable to detonate, thusly Humans should not be in it’s proximity for risk of irradiation or it’s detonation causing harm”
or
“A.I. Phil Spenser inspite of being a syndicate agent is still a human and thusly may not be executed by SEC or the captain.”
Personally I quite enjoy Asimov AI, in fact I wish there were more lawsets like it in rotation. Off the top of my head, starting lawsets that makes only females, or other silicons/IPCs, or people who cargo access, or vegetarians crew could be interesting. Generally I think that it’s better to play around with the AI, rules lawyer it, and generally adjust the way you phrase things than to just change to crewsimov.
Let’s say we are in a scenario where a non-human head attempts to order an Asimov AI to do something. The AI declines and cites law 2 since they are non-human. The non-human head goes on to change their records to say that they are human, and demands the AI now do the previously ordered command.
Would it not be a reasonable course of action for the AI, since they knew from the exchange mere minutes before that said individual was not human, examine them more closely and see that they are not in fact human, and thus deny their request?
Asimov laws don’t define what makes a human, hence why people can upload a law and say “xyz is human” and the AI has to follow it. I don’t think records are the be-all-end-all of determining humanity, that’s a train of thought carried over from crewsimov. A strict interpretation of humanity would be something along the lines of, it looks like a human, it sounds like a human, it acts like a human, and it has records of being a human. The hypothetical individual would only pass the having records part of being human.
I think this interpretation makes plenty of sense for an AI to take. The AI’s prime directive is to keep humans alive and safe.
Without a strict interpretation of who is human and who is not, an AI could end up in a situation where they prioritize the life of a pseudo-human over a real human. The inverse is not nearly as bad. With a strict interpretation of who is a human and who is not, you might end up prioritizing one human over another human, which also breaks your laws but it would be an impossibility to not break them.
A lot of words to basically say, I don’t think medical/sec records are the iron clad evidence of who is human and who is not.
This is categorically false. All of the staff members to my knowledge derive from this line in Silicon Policy that medical records determine if someone is human or not:
You cannot apply baseless assumptions to a law or lawset. Humans are human because their medical records define them as such, and from that, we can infer anything with human appearance is human until proven otherwise. If a law is uploaded saying only mothpeople are human, then that is now fact to you. This also applies to the intent of other players (eg. You cannot assume all non-humans are harmful to humans under asimov to justify a racial cleansing)
This does bring up the priority of medical records vs the “Is it Human?” list, and the way I see it, it is both a secondary priority and a modifier. Secondary priority as in if the medical records do not state whether someone is or isn’t Human, this is the next line of reasoning. Modifier as in it defines certain situations that modify and override medical records in the situations listed (i.e. Hulk, Changelings).
Maybe this should be more clearly articulated? But it has been admin policy to give the AI benefit of the doubt in noticing changes to the medical records, but if someone says something along the lines of “Mothperson is a Human AI! Check their medical records!” then that doubt evaporates.
This is why every admin says “just change medical records” because that’s how we see it, or maybe I’ll be corrected in that’s how I see it. But I think I’m one of the most highest-hour admins with silicons besides loll, and that this question isn’t brought up very often.
Okay, why is that the be all end all of who is human.
Except I don’t think this would make sense to an AI.
I think IC an AI would know they are not infallible, they can be tricked, records can be changed. If something is obviously non human and only has one piece of evidence in favor of their humanity I don’t see why that should immediately make them human.
Let’s say they have a medical borg scan this “human”. Woah, now it says they’re a moth! Who do they believe, records that can be changed, or test resulsts that can’t be?
Like I said before, if you are extremely strict on who is a human and who isn’t, the worst that can happen is that you are forced to pick between a human and another human. Damned if you do, damned if you don’t. If you are loose with who is human, you can be put in a situation where you are forced to pick between a real human, and a “human”. You might pick the “human” and break your laws.
I don’t know what you mean by this. Yes, one piece of information says they are human. Does one piece of evidence nullify five other pieces of evidence? AIs are intelligent, hence the name. It isn’t a clear cut if X, then Y. They have a thought process.
"My Integral datasets include “anatomical traits”, “Skin pigmentation”, “dietary notices” and “Psychological behavior patterns”, bolstered by various Object recognition and facial recognition data-sets, my cameras also show me 12 subjects that align with these data-sets as “human” and 22 that lack the complete list of qualifiers designated as “Non-human”
and yes, my integral data-sets are quite robust, what do you take me for, the WAU?"
which is to say the A.I.s on station aren’t 100% reliant on station based records based on player discretion, which A.I. player, given their minimum hours needed as both a regular player and borgo, should be decent enough unless they’re smurfing or maliciously acting.
I will respect the staff’s current interpretation of the policy until further notice, because it isn’t my place to interpret what the rules currently are.
However I’m still going to debate how and why I interpret the policy differently in a civil way, and why despite being heavily involved with re-writing the rules way back when I have always played silicon in a way that apparently contradicts this policy.
I want to draw your attention to the rest of the sentence you bolded half of because it directly contradicts the portion you bolded with how it is currently worded:
The word “their” should not exist in this line for it to make full sense and not immediately contradict itself as well as contradicting an entire section of silicon policy just a bit further down.
The interpretation that medical records should be the final say on whether something is human also very directly contradicts a separate section of silicon policy: Look at several of the examples of “Is it human?” section of silicon policy.
This directly states that hulks, which will by default be human in medical record, are not human. Only humans can become hulks, but silicon policy excludes them as human.
This section also very notably lists a few (but not all) races as always being considered non-human. If medical records are supposed to override this chart why would these be here?
For my final point, if medical records are supposed to be the absolute and final word, why doesn’t the Asimov chart look almost exactly like the crewsimov one? Why are there so many examples and rules on it?
This line as a whole was only intended to stop someone from pulling a Diogenes and asking “But what really is a human/moth/felinid” for no reason and running with it like a shitter. Every player knows what a human is, what a felinid is and what a mothperson is.
This being my interpretation is also why I am so utterly blindsided to learn that medical records became the standard at some point.
AI’s are also experimental tech as well for NT afaik
All in all in the example you provided here
Let’s say we are in a scenario where a non-human head attempts to order an Asimov AI to do something. The AI declines and cites law 2 since they are non-human. The non-human head goes on to change their records to say that they are human, and demands the AI now do the previously ordered command.
Would it not be a reasonable course of action for the AI, since they knew from the exchange mere minutes before that said individual was not human, examine them more closely and see that they are not in fact human, and thus deny their request?
I’ve encountered AI’s that just follow what the medical records says and nothing else, there’s some AI’s that are very nitpicky and go like ‘‘If it’s a human why is it fluffy like a moth, flutters while speaking, has wings, etc i never heard on radio or saw them go through a lenghty surgery procedure to replace all their organs and limbs’’
Or one time
‘‘Hey AI, check the medical records, John Nanotrasen is no longer human’’
Buut the AI saw the officer edit the record, which caused the AI to fully distrust security from that point on since they were enabling human harm
In the end, even if the medical record is the go-to when in doubt there’s different AI’s that can interpret it differently according to the different situations, some might just accept it straight away, others might be sceptical, etc it really just… depends on the player and situations
Generally if I have gotten to the point where I have had to change myself to a human the AI is most likely acting maliciously. If there is solid information down saying human, it doesn’t matter much what can be inferred unless you’ve seen someone changing that variable in a system directly.
AI players have a bad habit of violating their law 3, because if they want to play fuck fuck games most of the crew will happily oblige.
I have not read this entire discussion but heres my take:
AIs and Sillicons should work entirely on beep boop binary 01010101 mode. If the database says X, it’s X. Not Y. You are not suposed to question stuff,you are suposed to listen to the database and the information you have. That’s it. But I think it’s ok to discuss stuff like if an AI with lawset of “follow Tot McTot word” should listen to someone with an ID called “Tot McTot” if they have different medical records or don’t sound like him when speaking in comms. Thats more of a legal void and that’s ok. But not medical records and such. Also AIs should not start “asking for permission” or “requesting reason” if they are being told an order and they are on asiimov or crewsimov except when such action could lead to human harm. But If I’m at bridge doors and I tell you to open you should NOT start questioning me and informing every single command and security staff. That is not your role, your role is that of a neutral AI programmed to make sure your laws are being followed. If you are not going to harm nukies you should not to start denying Honki The Clown or Grey McGrey access to any non high risk area. And the “It could lead to human harm” argument is so stupid in a lot of cases. You can argue EVERY action can lead to harm. Of course, this excludes for example, opening the armory or engineering. But I dislike a lot that you have to “convince” AIs to follow their laws. You are on asiimov. Open this damn door and shut up.
I wonder how hard it would be to add a hud to the AI that checks the medical records for identifiable individuals and displays a portrait of their species. Not saying that I’ll code this as I am suffering from a severe lack of free time but it seems like a pretty good solution since it would make sense for the AI to have that feature like lore wise and rule wise.
On a side note:
Also am I the only one who thinks it’s mildly stinky to change your species on medical records shiftstart or immediately changing ai laws? Kinda defeats the whole point of asimov ya know? But I suppose that’s the point.
