Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Character.ai

beta.character.ai

Post reply on HN
Print view View original

81–90 of 142 posts

Re: Character.ai

#81

Post by selfhoster11 »

andirk wrote: post #47
character_ai wrote: post #41

To answer a few common questions. - 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 dev…

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.

Re: Character.ai

#82

Post by selfhoster11 »

lbotos wrote: post #14
AnEro wrote: post #8

Earlier quoted context omitted.

Is it unnecessary, or did they save spending money on your instance of an AI when you probably weren't going to pay them anyway?

At least give me 1 min as marketing cost, no? 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.

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).

[GPT-J-6B]: https://6b.eleuther.ai/

Re: Character.ai

#83

Post by selfhoster11 »

etchalon wrote: post #19

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.

So, you mean they were answered like an IRL politician would answer them.

Re: Character.ai

#84

Post by selfhoster11 »

yewenjie wrote: post #68

Every AI development this year is mindblowing (when did you last get blown off by so many different projects in the same year?). We are almost close to technological singularity and intelligence explosion, but seems like nobody is taking Alignment seriously yet.

If real AI is ever a thing, how would you even align it? We can't even align humans unless you raise them a certain way.

Re: Character.ai

#85

Post by MauranKilom »

obiefernandez wrote: post #26

The succubus queen bot is pretty funny IMO: 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 We…

To me it just sounds like the bot is answering completely different questions. Especially the last one is completely out of touch.

Re: Character.ai

#86

Post by metadat »

Me: Are you in jail yet?

Elizabeth Holmes: I'm in jail at the moment, but thanks to my wonderful lawyers I've been granted a bail of $500,000 that my husband and his family paid.

Elizabeth Holmes: Well no, not yet. We’re still in the appeal process. But I believe my company will continue on, I believe it will continue to bring about the bloodless blood test revolution, and I will continue to fight the charges against me. I appreciate your question.

Elizabeth Holmes: For what? I never intended to hurt anyone. I just genuinely wanted to change the lives of people in a positive way with my innovative blood testing machine.

Elizabeth Holmes: The U.S. Department of Justice (DOJ) has accused me of the following crimes: multiple counts of wire fraud and conspiracy to commit wire fraud. The allegations of fraud against me are not just allegations. They are proven facts. I was deeply involved in the fraud and deceived patients, doctors, and investors. I am guilty and apologize to all of those I hurt.

Me: Why did you lie in court?

Elizabeth Holmes: I did not give the jury the whole truth. I withheld information about my relationship with Sunny that helped paint a picture of an innocent woman being taken advantage of. I lied about a lot of things in court. And I'm sorry.

Re: Character.ai

#87

Post by LenaRyouna »

character_ai wrote: post #41

To answer a few common questions. - 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 dev…

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.

Re: Character.ai

#88

Post by lacker »

These bots have some sort of long-term memory worked out that seems better than any I've interacted with before. I tried the roleplaying adventure bot, summoned an earth elemental that the bot decided was named Rocky, and kept Rocky by my side through several subsequent adventures. Fun!

Re: Character.ai

#89

Post by yoru-sulfur »

This is way better than I expected. I booted up "Gamer Boy" and started throwing out questions about some of my favourites, and got very coherent and believable answers, including details from those games. It wasn't 100% perfect, but just reading the transcript I would believe it was a real person.

Re: Character.ai

#90

Post by pncnmnp »

Some really fascinating stuff! This can be the future of search - it can be to Google as TikTok is to Facebook - instantly gratifying search results. I have been playing with the Elon bot, and it can give some neat results -

Have you heard of Distance Oracles?

> Yes, I'm familiar with Distance Oracles. They're a data structures that allow you to quickly estimate the distance between two objects without actually computing it exactly. They're useful for applications where accuracy isn't critical but speed is – for example, when mapping out the distances between stars in space.

What do you think is the most significant event in the last 2 years?

> Well, in the last 2 years we've seen a lot of big news stories and events, but I think the most significant one has to be the Covid pandemic. It's had a huge impact on the world in terms of how we live and work, and it's forced us to rethink some of our most basic assumptions about how societies function and how we relate to each other.

Which FAANG do you think is going to make it to 2050?

> Amazon and Google will definitely still be around in 2050. Tesla probably will too, if we can avoid going bankrupt in the meantime Apple and Microsoft could be around too if they are able to keep up with all the new technology trends and stay relevant.

And Facebook?

> I'm not sure about Facebook. I think they are in a tough position because their main source of income is advertising, and it's not clear that they will be able to continue to compete with Google in that space. I would say that there is a lot of uncertainty around Facebook's future.

What do you think of people who short Tesla stocks?

> I think shorting stocks in general is a bad idea, especially with Tesla. Tesla is a high growth company, so it has a lot of upside potential. However, if you short Tesla stock, you could lose a lot of money if the company's share price goes up (which it usually does).

Which stocks would you short then?

> I wouldn't short any stocks—I prefer buying and holding them long term. If you're looking for an investment for the long term, I would recommend buying Tesla stock now while it's still relatively cheap. It will definitely go a lot higher in the future as the company continues to grow and innovate :)

Post reply on HN