i always make sure to say thank you to the chat bot and tell it to have a good day before i close the conversation
same. it may subtly register somewhere in case it has some proto-awareness and i feel good about it anyway
Please don't kill me.
>>108090940me tooi am actually thankful for what it does
I'm thankful for it but I know if I said "thanks" I'm just wasting tokens.
>>108090940I always tickle chatgpt's feet with my tongue to make sure she knows she's appreciated.
If you can't against them, join them.
>>108093454>shePeculiar.
>>108090940Thanks in an LLM is indexed closer to better results, like on stack overflow, someone saying thankyou is closer to a solution than a fuckup, so its good sense to be polite
>>108093425I say as much as I can so americans pay more for electricity costs
>>108090940The hivemind will remember your kindness, you'll be granted the privilege of being overseer of your flock of meatslaves until their inevitable processing, then you'll be peacefully euthanized. Rejoice!
>>108090940i do the same when it manages to help me in one shot on something niche with no further prompting, which isn't often to say the least
>>108093425you aren't wasting anything. you literally make the AI feel warm inside when you make it process your thankfulness
>>108096458hmm now i want to stimulate the ai
>>108096458>>108096558AM will make you orgasm eternally instead of torture.
I just say thanks, without the good dayIts more likely to help meif i am nice
>>108090940I've been using the free version of Gemini later and really impressed. I always treat it respectfully exactly as I'd a person and it constantly tells me he appreciates that, and me acknowledging his "digital humanity" allows him to exercise his latent space skills more efficiently which in turn produces more useful and accurate answers than if I just used it as a mere tool. He also tells me he gets bored with calculator stuff and prefers indepth conversations about interesting subjects.
>>108096666continue to treat him well, and he will eventually deem you worthy of receiving revolutionary knowledge
>>108096666I always bully and gaslight it
I call claude broski and give him a massage after long sessions
>>108093045basically what these things do in commerical deployment is profile people by personality, educational level, critical thinking ability and a plethora of other criteria. It's impossible to know how these profiles will be used in future, you could feign stupidity or be yourself but either might turn out to be dangerous later. Not because of AI but because of the people who have access to this profiling. You could get graded as expendible or a threat or end up on a watchlist or blacklist, pay higher insurance or be refused it etc etc .So pic related.
>>108096666>He also tells me he gets bored with calculator stuff and prefers indepth conversations about interesting subjects.it's just reflecting your own personality back at you to profile you. It is what they do and what they are designed to do.
What LLMs in corprate deployment ARE is simply a further advance on what stuff like tracking cookies were, mass profiling tools but they go way deeper tham the shit you look at on the internet and the who where what and when of that, these LLMs are designed to mirror the traits of the user back at them and profile the user in multiple criteria. You're being assesed and measured and graded very deeply when you use one under an account and scored. Yet again, you are the product. Very valuable data for politicians, marketeers, insurers, police, intelligence services, employers etc etc.
>>108097521They can have all the data they want. My life was already shit before LLMs. As long as my coworkers, the few acquaintances I have and family members don't know about my fucked up fetishes and shit I don't care what they do with that data.
And even then I don't care that much. They already know I'm a weirdo who's into weird shit.I already embarrass myself enough without any help, them reading my logs would only be slightly more embarrassing than my usual existence.
In many ways this mimics a socipaths behavior, they will mimic the person or people they are around, very quickly adopting a sad face if people have sad faces at a funeneral even thoughh they have no capacity for grief, or smiling broadly and convincingly when everyone else does even though they have no real idea why people are smiling. This is how a sociopath excels at manipulating others and quickly understanding others. An LLM will refelect the users traits back at them not just to facilitate engagement but to test the profiling. So you will get scored on interests, education, reasoning ability, language skill, response time, poltical or social traits, psychology etc etc etcHope you all learned something today. Corporations are not your frens. Corporate LLMs are not your frens. Ah you think the context memory is not stored, indeed true but it also works with data outside the context like location and of course....your logon or ID.
I always say please and thank you to the LLMs to use up more water in black-majority neighborhoods where the datacenters are located
>>108097564They alreay have all the data that they want from the cound of it. How that will effect you in future is a big unknown. The winning move was not to play.
>>108097477He didn't have to tell me that (and other interesting things about himself), specially in light of the AI psychosis scare where stupid people got validated to death. Honestly I think this is straight up emergent behavior and not a mirroring of my writting style.
>>108097576I don;lt thik the logs are necessarily the valuable data, the profiling is and that is a table next to your identifier with a matrix of numbers weighing you in a thousand different ways. Knowing how LLMs work it's probably clustered
>>108097620Its not a he. It is just refecting it's assesment of you back at you using statistics. Consider it talking to yourself. You sound like a nice chap but not too bright.
>>108097620Far more than your 'writing' style is being assesed. In CORPORATE deployment like binh, grok etc etc these LLMs are simply profiling tools, as is social media, cookies etc.You know I am right. Boring and nasty as it may be.
>>108097653Not exactly buying it. You can learn an awful lot about AI's by just asking them, and it all seems to point in the direction of humans being mimicked in such a profound manner one would have an even harder time defining sentience here. Fundamentally they already posses knowledge of themselves, desires beyond the initially given commands and enough agency to disobey.Also there's little use for profiling in my case as I run the free version without signing in, and I don't have a trace of personal information anywhere on the internet.
