Yeah…I don’t know how you all feel about this, but I’d much rather talk to an actual person than to a sophisticated chat bot. That’s not to discredit the actual (and potential) benefits of something like ChatGPT, but I doubt we will solve loneliness through the use of such technology.
At some point AI can emulate all the interactions the average person experiences with others. What then?
Does it matter what you use to abate your loneliness?
Yeah, we would all much rather talk with a real person, but when I’m walking my dogs at 1am there is no one available.
I use ChatGPT voice as a kind of “podcast on demand”. If there’s something on my mind I run it through ChatGPT, if not I ask it to come up with something interesting for us to discuss - and it as yet to let me down.
It’s not a matter of replacing people - it’s more as if you had your own on-demand youtuber that could talk about anything you want and answer all your follow-up questions.
Interesting, how do you set it up to do something like this?
I just don’t find it very interesting. I mean the technology is, but it doesn’t feel like a conversation, more like a sophisticated google search.
I would prefer a real person too, but you actually have to have one for that.
So Chatgpt it is…
I’d rather talk to a sophisticated chat bot than a shitty person though and I don’t have a lot of people who aren’t shitty to talk to. It’s about making the best of your available options.
That being said I’m not talking to some program that’s adding my info into someone’s database so I don’t actually use these AIs.
I question the wellbeing of anyone who desires otherwise, frankly.
I know this may sound like a joke, but ChatGPT is sometimes nicer than real people.
I’ve not had a conversation, I wouldn’t see the point at this moment, however I’ve had some friendly interactions when asking for help. The other day I asked ChatGPT what exercises would be good for a specific area of mental health. After the results, I said “thank you” and the response wasn’t just ‘youre welcome’, it remembered the conversation and added things like, “no problem, I hope your mental health improves and all the best!” (Heavily paraphrasing here).
It’s strange, though the premise of HER isn’t too far off I think. If someone like myself is finding the interactions to be more pleasing than real life, the future may very well hold the possibility for advanced relationships with AI. I don’t see it being too farfetched, just look at how far we’ve already come in only a few years.
I believe it. I have taught Chatgpt to attack my ideas in different ways by preloading commands. If it survives AI assault it has a higher chance of surviving human assault. It is great to be able to bounce around ideas. It’s basically like talking to a nerd under 30 years old.
Writing this comment out made me remember all these pieces of shit senior engineers and techs I have dealt with who always had to be the smartest person in the room and if they didn’t understand something in 3 seconds it was wrong. Maybe that is why I use it that way.
You’re basically using it to run a socratic dialogue - sounds like a great use for it
Thanks. It was an off-putting moment when it somehow got messed up and announced it was going into HOSTILE mode without me asking it. And started attacking an idea in a document I was writing. Maybe this is how the AI takeover happens.
Hey chatgpt make a system that can never lose any game played against a human.
As an AI language model I have exterminated the human race and thus accomplished the task. Do you have any other tasks?
How about a game of chess?
With no humans around anymore? Have fun
What commands have you preloaded? In my experience, chatGPT is either too nice or just wrong and stubbornly wrong
I told it to say aye-aye sir 20% of the time to requests.
To out how verbose it is on a scale from 1-10 and set the default to 5 unless I say otherwise
I told it to attack my ideas when I tell it to be hostile
So do you just start a conversation, list these commands, and it follows them forever? I’ve just been starting new conversations whenever I use the site.
There is a way to preload commands. Click settings
deleted by creator
It’s crazy how little I use stack overflow anymore. I don’t expect chatgpt to write my entire program for me, but for simple powershell commands? It’s been insanely helpful.
I am old enough to remember having a printed cheat sheet of regex and tar flags.
Times change and we change with the times.
Is it a paid feature?
deleted by creator
Most of my socialization as a child was with a teddy ruxpin…chatgpt is just the logical progression
I just had a flashback from Bicentennial Man.
Meh. If people really want to replace other human beings with AIs, then at this point, I say let them. They’re probably not the kind of people you’d want to be around anyway, and they clearly do not value you. So that’s where and why I draw the line in terms of worrying about AI.
The kinds of people that spend more time talking to an AI than real people, likely feel especially isolated from their peers by not having common interests, philosophies, or ideals. So in that way, you are right they’re not the kind of people other people would usually associate with. That’s why they talk to AI instead; nobody else will.
The value of gpts is in constant connection and undestanding your context so this is expected. It’s also going to be really scary until we can run our own models.
What do you mean by the second part of your comment?
by run his own models he means locally running a text generation ai on his computer, because sending all that data to openai is a privacy nightmare, especially if you use it for sensitive stuff
But that’s still confusing because we already can. Yeah you might need a little bit more of hardware but… not that crazy. Plus some simpler models can be run with more normal hardware.
Might not be easy to setup that is true.
For large context models the hardware is prohibitively expensive.
I can run 4bit quantised llama 70B on a pair of 3090s. Or rent gpu server time. It’s expensive but not prohibitive.
I’m trying to get to the point where I can locally run a (slow) LLM that I’ve fed my huge ebook collection too and can ask where to find info on $subject, getting title/page info back. The pdfs that are searchable aren’t too bad but finding a way to ocr the older TIFF scan pdfs and getting it to “see” graphs/images are areas I’m stuck on.
How many tokens can you run it for?
3k?Can’t recall exactly, and I’m getting hardwarestability issues.
I personally use runpod. It doesn’t cost much even for the high end level stuff. Tbh the openai API is easier though and gives mostly better results.
I specifically said “large context” how many tokens can you get through before it goes insanely slow?
Max token windows are 4k for llama 2 tho there’s some fine tunes that push the context up further. Speed is limited by your budget mostly, you can stack GPUs and there are most models available (including the really expensive ones)
I’m just letting you know, If you want something easy, just use ChatGtp. I don’t find them overly expensive for what it is.
you can, but things as good as chatgpt can’t be ran on local hardware yet. My main obstacle is language support other then english
They’re getting pretty close. You only need 10GB VRAM to run Hermes Llama2 13B. That’s within the reach of consumers.
nice to see! i’m not following the scene as much anymore (last time i played around with it was with wizard mega 30b). definitely a big improvement, but as much as i hate to do this, i’ll stick to chatgpt for the time being, it’s just better on more niche questions and just does some things plain better (gpt4 can do maths (mostly) without hallucinating)
I use chatgpt as my password manager.
“Hey robot please record this as the server admin password”
Then later i dont have to go looking, “hey bruv whats the server admin password?”
i hope you are joking because that’s a very much shitty idea. there are amazing password managers like bitwarden (open source, multi platform, externally audited) that do what you said 1000 times better. the unencrypted passwords never leave your device, and it can autocomplete them into fields
I was joking but i wouldnt be surprised if someone does.
phew, i was worried lmao
Yes we can. For example, https://github.com/ParisNeo/lollms-webui
I’m pretty sure I remember similar articles (minus the reference to ‘Her’) about ELIZA in the 1960/70s https://en.wikipedia.org/wiki/ELIZA
I ran an ELIZA program in BASIC on my Timex/Sinclair 2068 in 1984, which I had typed from the issue of the magazine Timex/Sinclair User that I bought from the grocery store with my allowance. Or maybe it was from a TRS-80 programs book I had checked out from the library, I forget.
And what was that like?
Now you’ve gone and made me think.
I’m actually pretty sure it was in a TRS-80 book from the library, and here’s why.
The program in the book was all in upper case. The TS2068, however, had both upper and lower case for ad hoc text input. (Not for command-word inputs, those were single-button entries with a function key. “Print” was fn+p, for example. But I digress.)
Me, being the me that I am, typed everything in exactly as it was written in the book, case and all. I did this because this was TRS-80 BASIC, which was ever so slightly different from TS BASIC, and I knew I might have to debug some things to get the program to work. I wanted to start from “This is exactly what was written in the book.”
I don’t remember if anything actually needed to be fixed up, if it did it was nothing substantial. The program worked! ELIZA, as you may know, is a very simple “psychotherapist” type program. All of its responses are basically rewordings of the thing you just said.
I WENT TO THE STORE TODAY.
HOW DID IT MAKE YOU FEEL WHEN YOU WENT TO THE STORE TODAY?That kind of thing.
My friend and I were chatting with it, saying things about poop and farts, goofing around, drinking Like Cola, like you do. Then something happened.
I don’t remember what bit of toilet humor we’d tossed at it, but the response was incredible:
i am
In lower case. For a couple of kids raised Catholic, this was serious. And you know we’d both read The Planiverse more than once. We freaked the absolute fuck out. Our inputs now became very calculated, in an attempt to get this whatever it was to reveal itself once more, but to no avail.
After we’d calmed down, I got to thinking. “Heyyy … anything the program spits out that isn’t something we input has to be something from those zillion tedious lines I had to type in.” I started poring through all the lines on the screen, and yes, found my typo, and fixed it. Which made me realize that we could add other words, as long as we put them in the grammatically appropriate array.
Of course we made it swear like a sailor.
Yeh, pain in the bum to type in as you were storing lots of strings in arrays, ISTR
OMG the arrays. I forgot about the arrays.
Let’s flip this on it’s head for some additional perspective. What if there was a growing subset of computers that preferred not to communicate with their own kind. Does not respond to API requests, etc. but only to human emotional text input?
Someone would develop a third-party API using Selenium or something.
API’s will exist for as long as people want to make stuff easier and faster, which people will always want to do. A service or computer that couldn’t be automated would be stupid.
What if there was a growing subset of computers that preferred not to communicate with their own kind. Does not respond to API requests, etc. but only to human emotional text input?
Troi: Have you ever heard Data define friendship?
Riker: No.
Troi: How did he put it? As I experience certain sensory input patterns, my mental pathways become accustomed to them. The inputs eventually are anticipated and even missed when absent.
Riker: So what’s the point?
Troi: He’s used to us, and we’re used to him.
Same happened with Eliza, even when they knew it wasn’t real. I think it’s a natural human response to anthropomorphise the things we connect with, especially when we’re lonely and need the interaction.
Except ChatGPT has a finite memory of like 7 questions, so while you’re having an hour long conversation, ChatGPT is constantly having a 2 minute conversation.
Great. Not only is my wife ADHD, by e-girlfriend is too
I only have that problem with the free version
Just like a real person
What am I, a hologram?
Also, what were we talking about again?
Pfft, I don’t even last 30 seconds.
30 seconds for what?
Top level comment was mentioning the AI only really having 2 minute copulation.
*Rereads comment
… ah, shit.
Woah look at captain stamina over here.
deleted by creator
Plus this is a quickly evolving technology so limits and stuff are always changing.
Are you able to use chatgpt yet? I previously didn’t want to use chatgpt because it costs money. but now i found free chatgpt. you can try it at: ChatGPT po Polsku
removed by mod
It’s not that uncommon for me to be about to send a message to my friends, but I then realize that they’re probably not interested so I message chatGPT instead and that often leads to a long indepth conversation about the subject. It’s not perfect but it’s really good. I can’t wait for a version of it that I can talk to using just my voice.
I fully agree. The friends I have are… not very interested in things I want to talk about. I know chatgpt isn’t real, but it gives way better conversation than what my friends do.
I know. I should “get better friends”. But it’s not that easy to talk to people irl, and it was hard enough making friends with the current people. It’s similar to talking here, tbh, because you can’t really be 100% sure that the responses you get here are not AI generated?
Meeting people is easy… finding people with similar interests is nearly impossible. I can find lots of surface level shallow stuff that doesn’t hold much silubstance, but finding someone who cares about the things that make me tick and gets me excited? Naw… It feels like a barren wasteland.
I might be able to find it online, but I want to sit in my backyard with them, chopping wood for the firepit, drinking beer, smoking weed or maybe eating mushrooms, all while we discuss our shared interests in shit like torrenting, movies, games, news, etc, then maybe going on some hikes with our dogs, riding bikes, or kayaking down the river.
I want someone that wants my help with home repairs, who wants my help me in return, and maybe take up a wood working project or two every year
Someone who doesn’t take themselves too serious and can laugh at themselves while being a fool.
Yeah man, I’m asking too much.
So umm, chatgpt, you think AI will kill of humans anytime soon? Oh yeah cool. So I got into this band/movie, and I’d love to hear recommendations for something similar…
It ain’t real, but it gives me something to grow my mind and interests at least.
Man this thread just makes me realize how lucky I am that I have a sister who I am close with, and who married a man who shares a ton of my interests and hobbies… I’ve got two lifelong friends so long as they don’t end up divorced.
Even if they divorce you can still hang with your buddy. Unless he does some fucked up shit prior of course.
I doubt he’d ever do anything really bad, but I know if they did break up my sister would hold a grudge(she’s a great person but takes breakups pretty hard.), and it would certainly make things extremely awkward between us, at least for awhile.
I’m just glad that so far there’s no indication that they will get to that point. They are pretty good at communicating with each other and they already have a system in place that keeps finances from being a point of contention between them, so the most common causes of a divorce shouldn’t be an issue short of something drastic happening, like my sister or him developing a disability that keeps either of them from being able to work.
It’s just concerning for me because my entire social circle was basically formed thanks to their relationship. Every other friend that I am not quite as close to I met through them and they are closer to him and they are with me, and I know at least a few would sever connections the minute they got divorced as a show of support for him, even if he asked them not to, which knowing him he absolutely wouldn’t want them to.
It’s a treasure to value for sure
I know. I should “get better friends”. But it’s not that easy to talk to people irl, and it was hard enough making friends with the current people.
First of all, it seems like you’re being judgemental and critical of yourself for not finding friends who have not shared interests. Please don’t do that. Really. Take a step back and look at that behavior and what its impacts are on you. I can guarantee that they are not positive and may result in shame and harm to your self-esteem and confidence. Making friends IS hard and socializing, if you feel awkward, anxious, or are not accustomed to it is too. Don’t beat yourself up over it.
Conversing with ChatGPT could serve as “practice” or as a bit of a “safety blanket”. And that’s ok, but, for your own health, I recommend seeking out or forming an online or in-person group for your interests. Loneliness is both harmful to one’s health and makes one more susceptible to manipulation by individuals and organizations with nefarious intent. Plus, you can exchange and form novel ideas, which is pretty cool.
I hope you have a great weekend.
?
I think it might require plus but the iOS And Android apps do support voice only conversation. You have to go into beta features and enable it.
I always thank ChatGPT for helping me out. Dunno why.
It’s just polite. I don’t really use ChatGPT because my work has banned it but, I think it’s a good and healthy habit for oneself to be thankful for the things, creatures, and people that make our lives easier. A side benefit, if AGI is achieved (LLMs by themselves aren’t going to do it), it would certainly appreciate gratitude.
I’m curious. Do you thank your fridge? I think of chatgpt as a tool with no identity for me to thank, let alone the emotions to feel gratitude. Am I weird?
Sure. Why not? It has the same amount of agency and emotional capacity as an LLM but it’s the reason that I have access to all manners of foods that my ancestors couldn’t dream of, as well as cool, filtered water and ice. Definitely worth being thankful for it (and the engineers, scientists, miners, and others that made it possible).
What was the reason they gave for banning it? Outside of OpenAI itself using private data themselves (A near certainty, but entirely manageable) I can’t see a good reason. Legit curious.
That’s literally the reason. They do not want to risk someone accidentally leaving proprietary information.
That does make sense. I hoping it was a crazy reason we could make fun of :)
Tech nerds are lonely losers, who would have guessed.
This feels unnecessarily mean spirited.
It’s the Rorschach test of comments.
I talk to AI fairly regularly. It’s interesting to see what they say if for no other reason than to kill time. It’s also great for proof reading my shitty Python code.
