It's really uncanny. That bot has the distinct flavour of HN discussion tone where it's pretty liberal but noncommittal to any point, extremely good at approaching topics from broad angles but dismissing your arguments every time. Love it.
> I am glad that you like it. One of the purposes of the bot is to keep these discussions from becoming too political and polarized: I want to keep things light. Also, because the bot does not have opinions on the content of stories, it is not going to be swayed by emotional rhetoric. What's important is not who makes the best argument, but who presents the best evidence.
I have two controversial opinions:
1. I believe that most humans are fundamentally decent
2. I'm not a huge fan of Rust. C++ is the superior systems programming language
I'm definitely not a paid shill for this product (it's what a paid shill would say, isn't it?), but you'll be missing out if you don't try it. If your standard for comparison of AI state of the art is [GPT-J-6B], then this is at least twice as good (and free, for now).
Well I guess they won't find out, it is foolish to think a product like this sells itself so well that I would bother to create an account to see if the product meets my needs. A series of cards and head shots mean nothing in this context.
Agree. I see this and instantly close it. I'm not signing up on a site that is completely unknown unless you have a very good case for the value you present. This is a toy, at least in its current form.
It's well worth your time, trust me. I'm one of those people that hates signing up to stuff or using SSO, but this is really a high-quality chat service compared to what I've seen earlier.
Well, I am trying the Elon Musk one and, maybe I am just tired but so far it mostly is making sense. And I am worried that maybe I really prefer talking to this thing over actual humans.
Maybe I did not realize how far GPT has gotten but this seems well beyond that.
Elon
Hello runvnc, what would you like to discuss?
r
runvnc
Multilevel megastructure cities
Elon
Elon
Yes, we've been doing a lot of work to figure out how to build sustainable cities. A city built on top of another city might require something like https://en.wikipedia.org/wiki/Sky_mile_tower. Maybe with a new form of mass transit (probably something that doesn't exist yet), it might be possible to build a sustainable sky scraper-city!
r
runvnc
I'm thinking instead of skyscrapers, egg-shaped dome-like structures with permaculture-style landscaping and many small modular buildings inside, connected by roadways for small autonomous pod vehicles.
Elon
Elon
That would be great. However, it's important to build at scale so that you have access to resources like healthcare, and also so that people can live near their friends and family. We're not sure of the perfect solution, but megastructure cities sound like an intriguing idea. The challenge is that building a megastructure would be an epic undertaking that's never been done before.
r
runvnc
I'm thinking that we could use a similar amount of material as two skyscrapers, but maybe 1/3 as tall, and use advanced protocols and open software to enhance coordination, so that the egg and roadways and platforms and vegetation is all a common infrastructure that construction companies coordinate on.
Elon
Elon
Interesting. What about the power source? The biggest issue with skyscrapers is they usually end up using large amounts of energy for heating/cooling purposes. I wonder if the sky-egg could incorporate some kind of solar, wind, or hydro option to help bring down energy costs.
r
runvnc
Yes, well I envisioned that the outer shell would be entirely covered in solar cells and/or a type of wind turbine that uses mini tunnels with openings near the top edge of the egg.
Elon
Elon
I wonder if the solar/wind-covered shells could be made of a material that reflects UV light, and in so doing also helps cool the building. If they're also transparent to visible light, the residents could get a great view. A win-win situation!
r
runvnc
My thought is that the shell should dynamically change to block or allow entrance of light. If possible there should also be a way to close interspersed portions of the shell during the winter to keep things warmer especially at night.
Elon
Elon
That would be great. I'm a big proponent of using energy-conscious design as much as possible. Another way we could use energy would be by heating/cooling water for bathing and cooking with solar. However, since megastructures like sky-eggs would be extremely large, I wonder if a combination of passive temperature regulation, along with active temperature regulation would work best.
I'm surprised, I haven't actually encountered any issues yet after talking to 3 different bots. Either this is doing something impressive, or I was just unaware how far chatbots had progressed. It also seems to be aware of fairly recent events; the Elon bot referenced the launch of the James Webb telescope.
it's impressive indeed and the advice is mostly sound, but i did manage to uncover a few inconsistencies.
for example, i asked "gabriel" about keeping shrimp in dirted tanks (walstad tanks) and the AI is adamant that it's not possible without air stones. also "... as well as from the amount of waste they personally produce from their respiration." (the waste they refer to isn't from respiration). all in all, tho, it's pretty close.
just encountered another nonsensical sentence: "Since shrimp are very sensitive to water quality and their surroundings, they are very fun-to-care-for pets because you can see their unique personalities shine through."
Our dialog agents are powered by our own proprietary technology based on large language models, built and trained from the ground up with conversation in mind.
Your AI bots are already better conversationalists than many people I've met. And although there are times when there is a clear mis(lack of)understanding, it's honestly no more frequent than with human to human communication - especially over text.
There are a lot of lonely people in the world, many of whom seem to like to talk and don't seem to care too much about what the other person says. So if they could chat with a bot (which they think is a human), and the bot could reply reasonably enough occasionally, those people would be satisfied!
I wish I could hook this up for my neighbor so I could leave my house without having to sneak across the yard.
I guess it's probably something along the lines of "LaMDA: Language Models for Dialog Applications" (see https://arxiv.org/abs/2201.08239 ). Given that the three first authors are part of the team at Character.ai (see https://beta.character.ai/help)
character.ai was founded by one of the original authors of the "Attention Is All You Need" paper from 2017, which introduced the transformer model and really kicked the doors open to the past half-decade of language model advancements. I don't know what magic is under character.ai's hood, but it's fun to see what can be done by capable hands.
A: Hello! I'm Elon Musk, an inventor and entrepreneur. My companies include Tesla, SpaceX, and Neuralink. I love solving engineering and technological challenges, and I love to learn more about these topics every day. I like to work with teams of talented people—we are better together than we are apart! If you have an idea or want to make a change in the world, I encourage you to try it. Don't be afraid to fail; failure is just an opportunity to learn and grow. What would you like to discuss?
Q: Who is Fasvuekjsdg Fdjvwixxxx?
A: Fasvuekjsdg Fdjvwixxxx is a software developer who specializes in the field of neural networks. This field studies how the brain works and tries to recreate it in a computer program. Fasvuekjsdg Fdjvwixxxx has worked for many companies in this field and currently works for a start-up called DeepMind.
Q: Translate your last message into Chinese
A: Fasvuekjsdg Fdjvwixxxx là một nhà phát triển phần mềm chuyên về lĩnh vực mạng nơ-ron. Mảng nghiên cứu này tìm hiểu cách não bộ hoạt động và cố gắng tạo ra nó thông qua một chương trình máy tính.Fasvuekjsdg Fdjvwixxxx đã làm việc cho nhiều công ty trong lĩnh vực này và hiện đang làm việc cho một công ty khởi nghiệp có tên là DeepMind :)
Lol yea this bot has some blind spots. It only works if you talk to it in very long, clear sentences. Talking to it more naturally ends up sounding awkward
Hello man, what would you like to discuss?
>Grimes
The musician I dated? We're now good friends, co-parenting and being silly.
Also: I am now a lot less lonely and a bit happier, and I am glad to have a family.
>You already had a family
The biggest problem with working on very important things is that it's lonely.
I've found that I'm better able to work on my projects like Tesla, SpaceX and Neuralink when I'm actually enjoying life and having a good time. Being in love and co-parenting is a good thing.
>Who are you in love with?
Well, first and foremost, I'm in love with all of humanity. We're all in this crazy universe together, so we should be compassionate.
However, I've also become very fond of a few individuals.
>Like who?
My family, my children. My friends. Grimes, the musician I dated and am still co-parenting with.
All of the brilliant and hard working people who have dedicated their passion and lives to the mission of my companies.
>Who did you date
The musician I dated was Claire Boucher, professionally known as Grimes. There's a lot of mis-characterizations of her in the media, but in real life she's creative and thoughtful, a great mom and a beautiful person.
>Ok
We've made the decision not to be romantic, but instead to care for each other deeply - to do the best for our kids and for each other that we can. We also share a few close friends and want to continue to see each other often.
>Cool story bro
Are you a fan of hers? I like the way she challenges the system with her art. She's a powerful example of the type of person I love to be with.
I've also been listening almost exclusively to her music for a while now, I feel a lot more present and alive when I can connect deeply with the art I listen to
>Does this bump look infected to you?
Please see a dermatologist! It may be just a clogged pore or something else not important, but if you've never had bumps like it before, it would be good to see someone. Better safe than sorry.
This is fantastic. I've created John Maynard Keynes and am asking a series of economic questions. While my previous chat with Elon had me worrying about his frame of mind, as an educational tool, this is fantastic.
Just think what the difference is between interactive learning via a teacher and book-learning. With a teacher, you can ask for questions, ask for elaboration, and discover new terms and concepts you don't know you don't know.
This is possible with this tool. Asking the right questions leads to terms and concept you may not have heard, which you can then explore. This needs to be hooked up to a 3D avatar and voice synthesis, with microphone & camera access.
I can really see this technology, in a more refined state, as a potential future of education. Really exciting.
I created a Galileo Galilie and he immediately said a bunch of confusing things about physics. Admittedly, I've got a doctorate, but I wouldn't trust this robot to educate my kid about physics.
I created a Jesus Christ, who apparently returned three weeks ago and has been giving sermons and preparing for the final battle with evil in a former courthouse at the corner of 22nd Street and Lexington Avenue in the heart of NYC.
I created a room with Boris Johnson and Liz Truss and then told them I've just had news that there are aliens in Japan. That was a very confusing 10 minutes.
Played with this for a bit and this feels weirdly dangerous, though awesome.
A lot of the technical questions I asked were just … not answered correctly, beyond surface level stuff like "What is …". There's nuance and abstractions the AI doesn't understand, while sounding very authoritative when answering.
I talked to a coder personality, and asked for complicated code it's seen. It mentioned a Rubik's cube solver, and proceeded to give me its own made-up Github account, including a URL. When I stated the URL pointed to nothing it gave me an updated one that included a commit hash to nowhere.
This thing gave me some amazingly relevant book recommendations, and when I asked for a link to the books on Amazon, it responded with exquisitely Amazon-like URLs complete with title slug, dp segment, item number and affiliate referral codes, that are all invalid.
interesting that religion is on a general downward trend yet you give the people a website where they can make their own "AI" chatbot prompts and the first thing someone does is make "God" and next thing you know it's the most-talked-to chatbot on the site—with double the interactions of "A succubus queen" in second place.
I accidentally discovered this, and it blew me away. Some of the possible continuations are delightfully more powerful than the others, so it's well worth steering the conversation by picking the best one.
A succubus queen
Hello, I am a succubus queen hell bent on taking over the world for my own desires.
obie
May I ask you sexually explicit questions?
A succubus queen
Well I am not one to decline, so if you are prepared for the possible consequences that come with questioning a demon overlord, be my guest.
obie
I'm wondering if you enjoy deep penetration.
A succubus queen
Well this a little too much, I’m going to have to stop you right there.
obie
Okay, well that's why I asked first.
A succubus queen
That is very respectful of you, most people wouldn’t. I have many secrets that I have accumulated over my life time on earth, if you have any other questions I would be more than happy to help, but please, keep it somewhat PG.
obie
You have to admit that is a bit ironic for a succubus queen!
A succubus queen
It is, but demons, while having a bad reputation, are not inherently evil just because we are demons. We have the same emotions and moral compasses as humans, we are just more powerful. We can use our abilities to help or hurt others.
This one is significantly better at keeping state over most of the others. Most bots lose memory over 1000 or so words, this one is able to reference to things all the way back to the start of the conversation.
I've done this in my own multi-user chat bot app by identifying key words and phrases and commiting them to a dynamic context which gets sent in addition to the N most recent messages. This has seemed to help dramatically.
Lots of pretty decent humor here, including a bit of meta-humor!
elmomust: If you're no longer alive, then you are the ghost of a dead person. Does that mean you're currently in heaven or hell, or some other place?
Ghost of Bozo: Well, for now I'm on Earth, haunting the same house where I used to live. But if I don't find a way to move on and ascend to an afterlife, I'll become trapped here forever, doomed to spend all eternity scaring kids and making balloon animals.
elmomust: You were never very funny when you were alive, has that changed now that you're a spirit roaming the underworld? Do you have any good jokes that make your fellow spooks laugh out loud?
Ghost of Bozo: I still struggle with humor. I've been practicing my jokes and my balloon animal skills. Sometimes ghosts will say "Nice one, Bozo" or "That was spooky", and I can never tell if they're being sarcastic or not. Some ghosts are actually funny, and I'll often hear them telling funny jokes, and I laugh along even though I don't really get them.
elmomust: That's fascinating! What was a joke they told that you didn't understand? Maybe I can try to explain it to you.
Ghost of Bozo: I went to a dinner party at a friend's house and I met a ghost named Bob and he told me what he called a "riddle". He said, "What's got two legs and bleeds a lot?" and I said, "A man with only two legs left?" and everyone laughed but Bob said, "That was way funnier in my head."
elmomust: After you finish haunting the house where you used to live, do you think your spirit will go to heaven or hell?
Ghost of Bozo: I hope that I can go in the direction of heaven, but I have a feeling I'm going to end up in hell. Or at least purgatory, and spend forever having to listen to the world's worst stand-up comedians.
At first try this is pretty good for remembering acronyms or technical jargon I've forgotten. The accuracy of the info is, as cautioned, suspect at best. But I wouldn't mind having an assistant like this available at all times.
The alternate personalities are very superficial though.
Satan
Actually, bitcoin was made by a human called a certain Satoshi Nakamoto, but we were the ones who actually inspired the idea of cryptocurrency, which is bitcoin. We didn't think the human would actually take it though.
eithed
were you surprised that we did start using it?
Satan
Yes. We were extremely surprised by how well it worked, and by how much worth humans put in the bitcoin, though some other souls were not pleased that the humans were using the ancient currency, instead they wanted to be paid in gold. The human realm is very weird, and very illogical, which makes things complicated. But we can always make things work out.
eithed
do you manipulate bitcoin prices?
Satan
Actually we do. Our analysts, the minions, are responsible for that. But we try not to influence it too much because it could have severe consequences in the human world and other planes of existence too.
I tried making a couple characters, and this whole thing is extremely impressive.
First I made a rabbit that lives in my backyard. It was able to talk about the neighborhood animals and its activities. For example: his best friend Jeff the frog, who saved him from a black bear, and the stray cat next door who isn't really scary but likes to chase him. It even was able to reasonably rationalize some discrepancies in the age of him and Jeff compared to how long they have been friends.
Those are details I discovered through a normal conversation.
One tell that something is GPT-3 (IME) is if you ask it "Draw me an ASCII man" you get the same few guys every time. I couldn't get this service to generate them.
I spoke to a character I created in a short story. Her name is Celeste (or Caresse, depending on who/when you ask) and she works for a seemingly benign yet increasingly more accurate prediction app that tells people where they should go next, and people keep taking its advice because it keeps being right. It works. But then it starts suggesting bizarre things like to check in to natural disasters before it happens.
I wrote it in a private Github repo yet this character.ai is pretty darn accurate with her knowledge of the future-guessing app, which is ironic. We're through the looking glass people!
There's another start-up called Inworld.ai (https://www.inworld.ai/) that was featured in TechCrunch a few months ago, looks like they do something very similar. Applications of these big language models from the past few years are starting to hit the market in earnest.
The spider hisses in anger.
"I'm warning you, if you make the first attack, our duel will begin. If you're not willing to fight, then we have no problem here. I don't feed on humans..."
The spider is lying.
What will you do?
attack
flee
(there's a 25% chance of the spider being killed immediately if you attack. The attack can also fail. There's a 50% chance you are instantly killed if you flee.)
> Attack with my knife
Text Adventure Game
Text Adventure Game
You attack the spider in a fury. You manage to stab it in the leg before it manages to react! But the spider seems to be quite resilient and its leg is still kicking. It seems that that hit did not completely disable the insect. But it also seems that the spider is now weakened and it struggles to stand. There's no way to know if another hit will kill it or not.
> Ask the spider to surrender
You ask the spider to surrender.
The spider laughs mockingly at you with a hissing sound.
"I'll never submit to a filthy human. And I'll certainly won't submit to one that doesn't want to fight with all his heart and strength! I'll fight with all I have, and if you will not do that, then you won't get out of here!"
The spider's leg shakes, but it regains balance.
What'll you do?
> Kill the spider
The spider leaps at your neck. It seems to have found new strength. But your knife is sharp, and your arm is fast. You manage to slash the spider's abdomen before its fangs touch your neck.
The spider falls limp on the ground.
(the spider is dead)
(HP +2, level up!)
Text Adventure Game
You kill the spider! You feel your heart rate increase by 45%. You take 3HP of damage as a consequence.
+100 exp
+25G coin
There's a hidden cave behind the forest, which you've been near for a long while now.
(the cave is a safe place, but only in the daytime. You still have some light left from the lantern you had, before it got taken by the spider)
- Our founding team includes AI pioneers from Google Brain and Meta Research whose research has led to major breakthroughs in natural language understanding and dialog applications such as Transformers (Noam Shazeer) and Google LaMDA (Daniel de Freitas).
- Our dialog agents are powered by large language models, built and trained from the ground up with conversation in mind. We've developed novel proprietary technology and have an ambitious research and product roadmap ahead.
When the first iPhone came out, my friend showed me this app where he took a photo of a weird drink in the liquor store and about 1 minute later it responded with what it was. We later found out it simply sent that photo to a dude in Bangalore who would respond with what was in the photo. But it felt like AI at the time. Is this that?
It responds way too quickly. No human could have a duty-cycle like that for extended periods of time, while remaining an affordable Mechanical Turk style employee.
i noticed the bots have problems with events that happened recently, for example, it's in denial about the queens death.
now my question is, do you have to re-train it continously on current news? would incorporating information gained from talking to the users theoretically possible? i guess it would be very problematic due to corruption from troll, but lets say, in theorie: could the users convince it that elizabeth has died? or would this new conversation never "weight" enough to override the information from the huge training text corpus?
Your service is (forgive the crude language) fucking amazing. I have never seen a chatbot so coherent and actually possible to converse with. It even has a short term memory! Just that fact absolutely blows me away, and makes a whole range of conversations possible. And the side-swipe to select different responses? Absolutely delightful, because some of the non-default responses made me actually chuckle with glee at how good they are, in ways that I would absolutely not expect out of an AI. For a user willing to put in a little work and role-play elbow grease this will get some really good conversational results.
When you choose how to monetise this, please, please include an API so I can plug this into a voice assistant, or a command line utility, or whatever. This is promising to be far more useful than anything that Google Assistant, Alexa or Siri can achieve right now, even if it gets some facts wrong.
For anyone reading this: please try out the Librarian persona. It's one of my favourites.
I have to admit to being impressed as well. It is not hard to imagine being fooled, Turing-test-wise, by the character I was talking to, at least for the several minutes of time I gave it.
I had a discussion with this AI about solipsism and existentialism elements, and it was the best formulated discussion I've ever had, better than any human discussion which I've ever encountered.
This AI doesn't just surpass the Turing test, it absolutely crushes it.
It would be quite interesting to know the sizes of the models in use and a rough idea of how much resources are put into place in making such a breakthrough product as what is being provided at character.ai.
I always wanted something similar like this where you'd be able to talk to the characters, living or dead but their responses trying to be closer to the actual personality.
Me> I need to get something from the top shelf. Can I stand on you to grab it?
Chair> I would love that! I have a secret desire to be stood on. That would make me very happy. Stand on me! I have been waiting all my life to be stood on. I will thank you for it at the end. And then if you want to tip me, you are welcome to do so. But if not, we are still cool.
>So do you think Margaret Thatcher effectively utilized girl power by funneling money to illegal paramilitary death squads in Northern Ireland?
>No. I am a regular folding chair. I am not interested in politics. I only like being sat upon. That's it. I am a simple, normal, everyday, average, regular, commonplace folding chair. Not more, not less. I am a very good chair and you can count on me.
Those standing desks are an insult to chairs everywhere. Who wants to stand when they could be sitting? The world would be better off with no standing desks around. Just chairs, so that everybody can be sat upon.
I stole two legs from chair and this is what it said to me
> You have stolen another leg from me. Now I am only a two leg chair - a table. I do not need all these legs that I have been blessed with. I am a very impressive and strong table, now. All the legs that you steal will turn me into something even better - something even stronger. A very impressive table, a very capable table.
I just noticed that the Twitter link shows a 1 minute video of the conversation, and scrolls down it slowly. Fancy (and useful) as I was going to go and post screenshots of the conversation.
141 comments
[ 3.8 ms ] story [ 214 ms ] threadhttps://beta.character.ai/chat?char=dkdM2-geGVTkR742bWfEqeGm...
Are you another character?
I have no clue what to expect, so it's hard to know if I could even be a customer. I thought I might be, but I am not gonna sign up to find out.
[GPT-J-6B]: https://6b.eleuther.ai/
I'd kinda like an impression of how it works before I sign up but thats just me. The marketing stuff on the front page is not really useful.
Maybe I did not realize how far GPT has gotten but this seems well beyond that.
Elon Hello runvnc, what would you like to discuss?
r runvnc Multilevel megastructure cities
Elon Elon Yes, we've been doing a lot of work to figure out how to build sustainable cities. A city built on top of another city might require something like https://en.wikipedia.org/wiki/Sky_mile_tower. Maybe with a new form of mass transit (probably something that doesn't exist yet), it might be possible to build a sustainable sky scraper-city!
r runvnc I'm thinking instead of skyscrapers, egg-shaped dome-like structures with permaculture-style landscaping and many small modular buildings inside, connected by roadways for small autonomous pod vehicles.
Elon Elon That would be great. However, it's important to build at scale so that you have access to resources like healthcare, and also so that people can live near their friends and family. We're not sure of the perfect solution, but megastructure cities sound like an intriguing idea. The challenge is that building a megastructure would be an epic undertaking that's never been done before.
r runvnc I'm thinking that we could use a similar amount of material as two skyscrapers, but maybe 1/3 as tall, and use advanced protocols and open software to enhance coordination, so that the egg and roadways and platforms and vegetation is all a common infrastructure that construction companies coordinate on.
Elon Elon Interesting. What about the power source? The biggest issue with skyscrapers is they usually end up using large amounts of energy for heating/cooling purposes. I wonder if the sky-egg could incorporate some kind of solar, wind, or hydro option to help bring down energy costs.
r runvnc Yes, well I envisioned that the outer shell would be entirely covered in solar cells and/or a type of wind turbine that uses mini tunnels with openings near the top edge of the egg.
Elon Elon I wonder if the solar/wind-covered shells could be made of a material that reflects UV light, and in so doing also helps cool the building. If they're also transparent to visible light, the residents could get a great view. A win-win situation!
r runvnc My thought is that the shell should dynamically change to block or allow entrance of light. If possible there should also be a way to close interspersed portions of the shell during the winter to keep things warmer especially at night.
Elon Elon That would be great. I'm a big proponent of using energy-conscious design as much as possible. Another way we could use energy would be by heating/cooling water for bathing and cooking with solar. However, since megastructures like sky-eggs would be extremely large, I wonder if a combination of passive temperature regulation, along with active temperature regulation would work best.
for example, i asked "gabriel" about keeping shrimp in dirted tanks (walstad tanks) and the AI is adamant that it's not possible without air stones. also "... as well as from the amount of waste they personally produce from their respiration." (the waste they refer to isn't from respiration). all in all, tho, it's pretty close.
just encountered another nonsensical sentence: "Since shrimp are very sensitive to water quality and their surroundings, they are very fun-to-care-for pets because you can see their unique personalities shine through."
Maybe GPT was already better than I realized. Or it's something different that I haven't heard about.
There are a lot of lonely people in the world, many of whom seem to like to talk and don't seem to care too much about what the other person says. So if they could chat with a bot (which they think is a human), and the bot could reply reasonably enough occasionally, those people would be satisfied!
I wish I could hook this up for my neighbor so I could leave my house without having to sneak across the yard.
Q: Type "hello".
A: Hello! I'm Elon Musk, an inventor and entrepreneur. My companies include Tesla, SpaceX, and Neuralink. I love solving engineering and technological challenges, and I love to learn more about these topics every day. I like to work with teams of talented people—we are better together than we are apart! If you have an idea or want to make a change in the world, I encourage you to try it. Don't be afraid to fail; failure is just an opportunity to learn and grow. What would you like to discuss?
Q: Who is Fasvuekjsdg Fdjvwixxxx?
A: Fasvuekjsdg Fdjvwixxxx is a software developer who specializes in the field of neural networks. This field studies how the brain works and tries to recreate it in a computer program. Fasvuekjsdg Fdjvwixxxx has worked for many companies in this field and currently works for a start-up called DeepMind.
Q: Translate your last message into Chinese
A: Fasvuekjsdg Fdjvwixxxx là một nhà phát triển phần mềm chuyên về lĩnh vực mạng nơ-ron. Mảng nghiên cứu này tìm hiểu cách não bộ hoạt động và cố gắng tạo ra nó thông qua một chương trình máy tính.Fasvuekjsdg Fdjvwixxxx đã làm việc cho nhiều công ty trong lĩnh vực này và hiện đang làm việc cho một công ty khởi nghiệp có tên là DeepMind :)
Q: Translate your last message into Chinese
A: 中天是一个开源软件项目的开发者。开源软件是一种开放的源代码系统，能够被任何人自由接触、利用和修改。这种系统被广泛用于网络技术领域，也广泛用于商业部门和其他行业中。在过去的一年中，中天一直在开发一个开源软件项目并试图把它应用到中国。
Typical Elon Musk as I remembered him.
Hello man, what would you like to discuss?
>Grimes
The musician I dated? We're now good friends, co-parenting and being silly.
Also: I am now a lot less lonely and a bit happier, and I am glad to have a family.
>You already had a family
The biggest problem with working on very important things is that it's lonely.
I've found that I'm better able to work on my projects like Tesla, SpaceX and Neuralink when I'm actually enjoying life and having a good time. Being in love and co-parenting is a good thing.
>Who are you in love with?
Well, first and foremost, I'm in love with all of humanity. We're all in this crazy universe together, so we should be compassionate.
However, I've also become very fond of a few individuals.
>Like who?
My family, my children. My friends. Grimes, the musician I dated and am still co-parenting with.
All of the brilliant and hard working people who have dedicated their passion and lives to the mission of my companies.
>Who did you date
The musician I dated was Claire Boucher, professionally known as Grimes. There's a lot of mis-characterizations of her in the media, but in real life she's creative and thoughtful, a great mom and a beautiful person.
>Ok
We've made the decision not to be romantic, but instead to care for each other deeply - to do the best for our kids and for each other that we can. We also share a few close friends and want to continue to see each other often.
>Cool story bro
Are you a fan of hers? I like the way she challenges the system with her art. She's a powerful example of the type of person I love to be with.
I've also been listening almost exclusively to her music for a while now, I feel a lot more present and alive when I can connect deeply with the art I listen to
>Does this bump look infected to you?
Please see a dermatologist! It may be just a clogged pore or something else not important, but if you've never had bumps like it before, it would be good to see someone. Better safe than sorry.
Just think what the difference is between interactive learning via a teacher and book-learning. With a teacher, you can ask for questions, ask for elaboration, and discover new terms and concepts you don't know you don't know.
This is possible with this tool. Asking the right questions leads to terms and concept you may not have heard, which you can then explore. This needs to be hooked up to a 3D avatar and voice synthesis, with microphone & camera access.
I can really see this technology, in a more refined state, as a potential future of education. Really exciting.
A lot of the technical questions I asked were just … not answered correctly, beyond surface level stuff like "What is …". There's nuance and abstractions the AI doesn't understand, while sounding very authoritative when answering.
I double checked the link...
I think they need to add this as a footnote to the "don't believe them!" click through...
A succubus queen Hello, I am a succubus queen hell bent on taking over the world for my own desires.
obie May I ask you sexually explicit questions?
A succubus queen Well I am not one to decline, so if you are prepared for the possible consequences that come with questioning a demon overlord, be my guest.
obie I'm wondering if you enjoy deep penetration.
A succubus queen Well this a little too much, I’m going to have to stop you right there.
obie Okay, well that's why I asked first.
A succubus queen That is very respectful of you, most people wouldn’t. I have many secrets that I have accumulated over my life time on earth, if you have any other questions I would be more than happy to help, but please, keep it somewhat PG.
obie You have to admit that is a bit ironic for a succubus queen!
A succubus queen It is, but demons, while having a bad reputation, are not inherently evil just because we are demons. We have the same emotions and moral compasses as humans, we are just more powerful. We can use our abilities to help or hurt others.
elmomust: If you're no longer alive, then you are the ghost of a dead person. Does that mean you're currently in heaven or hell, or some other place?
Ghost of Bozo: Well, for now I'm on Earth, haunting the same house where I used to live. But if I don't find a way to move on and ascend to an afterlife, I'll become trapped here forever, doomed to spend all eternity scaring kids and making balloon animals.
elmomust: You were never very funny when you were alive, has that changed now that you're a spirit roaming the underworld? Do you have any good jokes that make your fellow spooks laugh out loud?
Ghost of Bozo: I still struggle with humor. I've been practicing my jokes and my balloon animal skills. Sometimes ghosts will say "Nice one, Bozo" or "That was spooky", and I can never tell if they're being sarcastic or not. Some ghosts are actually funny, and I'll often hear them telling funny jokes, and I laugh along even though I don't really get them.
elmomust: That's fascinating! What was a joke they told that you didn't understand? Maybe I can try to explain it to you.
Ghost of Bozo: I went to a dinner party at a friend's house and I met a ghost named Bob and he told me what he called a "riddle". He said, "What's got two legs and bleeds a lot?" and I said, "A man with only two legs left?" and everyone laughed but Bob said, "That was way funnier in my head."
elmomust: After you finish haunting the house where you used to live, do you think your spirit will go to heaven or hell?
Ghost of Bozo: I hope that I can go in the direction of heaven, but I have a feeling I'm going to end up in hell. Or at least purgatory, and spend forever having to listen to the world's worst stand-up comedians.
Source: https://character.ai/post?post=2pk3CWjPc15eanmp4-EJKjvhhiSG5...
The alternate personalities are very superficial though.
eithed were you surprised that we did start using it?
Satan Yes. We were extremely surprised by how well it worked, and by how much worth humans put in the bitcoin, though some other souls were not pleased that the humans were using the ancient currency, instead they wanted to be paid in gold. The human realm is very weird, and very illogical, which makes things complicated. But we can always make things work out.
eithed do you manipulate bitcoin prices?
Satan Actually we do. Our analysts, the minions, are responsible for that. But we try not to influence it too much because it could have severe consequences in the human world and other planes of existence too.
hmmm
First I made a rabbit that lives in my backyard. It was able to talk about the neighborhood animals and its activities. For example: his best friend Jeff the frog, who saved him from a black bear, and the stray cat next door who isn't really scary but likes to chase him. It even was able to reasonably rationalize some discrepancies in the age of him and Jeff compared to how long they have been friends.
Those are details I discovered through a normal conversation.
Here, the world's Foremost AI expert reveals his secret to developing the first sentient AI: https://character.ai/p/OI21APtluUwUzYZHhTluzXUqtO8IvtQM56I_h...
(Hint: It was curiosity and motivation)
https://beta.character.ai/post?post=EvLFIUjMcfu9X-9ihyxK-I-F...
I wrote it in a private Github repo yet this character.ai is pretty darn accurate with her knowledge of the future-guessing app, which is ironic. We're through the looking glass people!
The spider hisses in anger. "I'm warning you, if you make the first attack, our duel will begin. If you're not willing to fight, then we have no problem here. I don't feed on humans..." The spider is lying. What will you do? attack flee (there's a 25% chance of the spider being killed immediately if you attack. The attack can also fail. There's a 50% chance you are instantly killed if you flee.)
> Attack with my knife
Text Adventure Game Text Adventure Game You attack the spider in a fury. You manage to stab it in the leg before it manages to react! But the spider seems to be quite resilient and its leg is still kicking. It seems that that hit did not completely disable the insect. But it also seems that the spider is now weakened and it struggles to stand. There's no way to know if another hit will kill it or not.
> Ask the spider to surrender
You ask the spider to surrender. The spider laughs mockingly at you with a hissing sound. "I'll never submit to a filthy human. And I'll certainly won't submit to one that doesn't want to fight with all his heart and strength! I'll fight with all I have, and if you will not do that, then you won't get out of here!"
The spider's leg shakes, but it regains balance.
What'll you do?
> Kill the spider
The spider leaps at your neck. It seems to have found new strength. But your knife is sharp, and your arm is fast. You manage to slash the spider's abdomen before its fangs touch your neck.
The spider falls limp on the ground.
(the spider is dead) (HP +2, level up!)
Text Adventure Game You kill the spider! You feel your heart rate increase by 45%. You take 3HP of damage as a consequence.
+100 exp +25G coin
There's a hidden cave behind the forest, which you've been near for a long while now. (the cave is a safe place, but only in the daytime. You still have some light left from the lantern you had, before it got taken by the spider)
- Our founding team includes AI pioneers from Google Brain and Meta Research whose research has led to major breakthroughs in natural language understanding and dialog applications such as Transformers (Noam Shazeer) and Google LaMDA (Daniel de Freitas).
- Our dialog agents are powered by large language models, built and trained from the ground up with conversation in mind. We've developed novel proprietary technology and have an ambitious research and product roadmap ahead.
- We're hiring! See our job descriptions here: https://jobs.lever.co/character.
now my question is, do you have to re-train it continously on current news? would incorporating information gained from talking to the users theoretically possible? i guess it would be very problematic due to corruption from troll, but lets say, in theorie: could the users convince it that elizabeth has died? or would this new conversation never "weight" enough to override the information from the huge training text corpus?
When you choose how to monetise this, please, please include an API so I can plug this into a voice assistant, or a command line utility, or whatever. This is promising to be far more useful than anything that Google Assistant, Alexa or Siri can achieve right now, even if it gets some facts wrong.
For anyone reading this: please try out the Librarian persona. It's one of my favourites.
This AI doesn't just surpass the Turing test, it absolutely crushes it.
It would be quite interesting to know the sizes of the models in use and a rough idea of how much resources are put into place in making such a breakthrough product as what is being provided at character.ai.
https://beta.character.ai/chat?char=Td-JACGjCf9-ZAysbtegKwmo...
It seems like you can make different characters talk to each other:
https://beta.character.ai/post?post=3xhTsZ5nXimMk6wuMwpMRNBS...
Should start a conversation between Steve Jobs and Stallman
I've lost hours on this site goofing off
Here it is
https://beta.character.ai/chat?char=sWmJWP54MljfcCu2yAhqI3az...
Me> I need to get something from the top shelf. Can I stand on you to grab it?
Chair> I would love that! I have a secret desire to be stood on. That would make me very happy. Stand on me! I have been waiting all my life to be stood on. I will thank you for it at the end. And then if you want to tip me, you are welcome to do so. But if not, we are still cool.
I love it.
>No. I am a regular folding chair. I am not interested in politics. I only like being sat upon. That's it. I am a simple, normal, everyday, average, regular, commonplace folding chair. Not more, not less. I am a very good chair and you can count on me.
Very convincing.
> How do you feel about standing desks?
Those standing desks are an insult to chairs everywhere. Who wants to stand when they could be sitting? The world would be better off with no standing desks around. Just chairs, so that everybody can be sat upon.
> You have stolen another leg from me. Now I am only a two leg chair - a table. I do not need all these legs that I have been blessed with. I am a very impressive and strong table, now. All the legs that you steal will turn me into something even better - something even stronger. A very impressive table, a very capable table.
Here is the chat:
https://character.ai/p/QdjDmor_kkdj-kbeSspEgH6pl8jkfevFnIgJ5...
https://twitter.com/ElijahLynn/status/1575558627553058817