It seems that you're using an outdated browser. Some things may not work as they should (or don't work at all).
We suggest you upgrade newer and better browser like: Chrome, Firefox, Internet Explorer or Opera

×
avatar
dtgreene: (Only one problem; for this to work, the AI needs to be able to run in real-time on a computer small and light enough to be carried around or embedded in a wheelchair.)
Technology is moving pretty fast, I give it a few years.

avatar
TheHalf-Life3: Or maybe even FBI.
Oh yeah, this stuff is going to both be a nightmare for security, lol. People are already pretending to be someone else.
Post edited October 31, 2024 by botan9386
avatar
.Keys: Everything makes sense now then!
You're a full fledge AI bot tricking us into thinking you're not using any other Large Language Model to chat with us!
Genius! :P
avatar
TheHalf-Life3: I’m still just feeling so-called anti-social anxiety.
Sorry! I was just kidding. :P
We know we are all humans.
avatar
TheHalf-Life3: I’m still just feeling so-called anti-social anxiety.
avatar
.Keys: Sorry! I was just kidding. :P
We know we are all humans.
Oh okay. I guess.
avatar
TheHalf-Life3: Or maybe even FBI.
avatar
botan9386: Oh yeah, this stuff is going to both be a nightmare for security, lol. People are already pretending to be someone else.
Worse. Can you imagine what will happen if the finance bros manage to create a model of regulatory financial laws, and teach a software to look for loopholes to exploit?

Every time a flawed new law comes out, a human bankster may take months, or even years to find a way to take advantage, and might get away with a couple million.

A software would find that loophole in seconds, and damages would cross into the billions before the day is done.

In fact, I think anybody who manages to be the first to develop one such model, and prove it to be reliable enough, could name their price. Any price at all.

Same with automated objects, like cars. Imagine a world of millions of automated vehicles, where a new traffic law, introducing significant changes, comes out, that every single car has to follow by Monday morning. Come hell or an interrupted WiFi connection.

Do you really want to leave anything in that law open to interpretation, and allow every manufacturer to write their own, slightly different interpretation? Hell, I wouldn't want to get near a road Monday morning 12 AM, when the law comes into effect.

You would need a model of all current laws, and be able to test the impact before that thing gets simultaneously rolled out to millions of 12 ton vehicles, right?

avatar
TheHalf-Life3: I’m still just feeling so-called anti-social anxiety.
avatar
.Keys: Sorry! I was just kidding. :P
We know we are all humans.
Speak for yourself, hooman! I'm an alien cat cyborg from Beta Centauri 3.
Post edited November 01, 2024 by Nervensaegen
avatar
.Keys: We know we are all humans.
avatar
Nervensaegen: Speak for yourself, hooman! I'm an alien cat cyborg from Beta Centauri 3.
I thought in our singularity reunion we'd trick everybody into thinking we were humans now...
You should check if the cyborg dogs of Alpha Centauri 7 didn't infected your mainframe again.
avatar
botan9386: Oh yeah, this stuff is going to both be a nightmare for security, lol. People are already pretending to be someone else.
avatar
Nervensaegen: Worse. Can you imagine what will happen if the finance bros manage to create a model of regulatory financial laws, and teach a software to look for loopholes to exploit?

Every time a flawed new law comes out, a human bankster may take months, or even years to find a way to take advantage, and might get away with a couple million.

A software would find that loophole in seconds, and damages would cross into the billions before the day is done.

In fact, I think anybody who manages to be the first to develop one such model, and prove it to be reliable enough, could name their price. Any price at all.

Same with automated objects, like cars. Imagine a world of millions of automated vehicles, where a new traffic law, introducing significant changes, comes out, that every single car has to follow by Monday morning. Come hell or an interrupted WiFi connection.

Do you really want to leave anything in that law open to interpretation, and allow every manufacturer to write their own, slightly different interpretation? Hell, I wouldn't want to get near a road Monday morning 12 AM, when the law comes into effect.

You would need a model of all current laws, and be able to test the impact before that thing gets simultaneously rolled out to millions of 12 ton vehicles, right?

avatar
.Keys: Sorry! I was just kidding. :P
We know we are all humans.
avatar
Nervensaegen: Speak for yourself, hooman! I'm an alien cat cyborg from Beta Centauri 3.
In one YouTube video documentary Preston B. Nichols had mentioned Alpha Centauri/Draconis Galaxy Star/System I guess by reinterpreting his own words.
avatar
Nervensaegen: Speak for yourself, hooman! I'm an alien cat cyborg from Beta Centauri 3.
avatar
.Keys: I thought in our singularity reunion we'd trick everybody into thinking we were humans now...
You should check if the cyborg dogs of Alpha Centauri 7 didn't infected your mainframe again.
Nah! We are past that. Hoomans are dumb. You could straight up tell them, and they'll refuse to believe you.

Remember Frank? He used to be the dullest tool in the shed, right? Great fur, soft brain? Tried to dye his tail pink with beetroot? The dude who was always after Becca, that tabby version 15.0b from back in the day? Now Frank is making millions on Youtube. There is this short video of him typing an e-mail that became a meme, and even when people clearly saw his thumbs, they shrugged it off, saying "it's because he is a European short hair".

People only see what they want to see.
avatar
OldFatGuy: We're told all the great things this new AI can do for us. I just spent over an hour and a half looking for something that I want (a consumer product if that matters). I have very specific wants, I type them in a search engine, and start from there.

Not a single one, NOT ONE, of the links on the first page had what I'm looking for.

Okay, I go into store fronts. Several. Again, I put in exactly what I want. I get results. Oooh boy I get results. Still haven't found one that meets the criteria I set out. After over an hour and a half of genuinely trying, I haven't found what I'm looking for.

Now I get that a lot, most, I dunno, maybe ALL of this is on me. I have always sucked at searches. Since the first search engines showed up on America Online about a thousand years ago. But isn't that where AI is supposed to kick in??? Isn't one of the purposes of AI to help morons like me navigate the digital world?

I think the problem is the same problem as ever. Greed for profit interferes with a clean way to do anything. Want a specific product? Yeah, we're gonna send you results of dozens because... greed.

What do guys think about regarding the future of AI, particular in the consumer world. I just can't see the profit motive ever allowing a real helpful AI get anywhere. The only ones that will materialize and last will be those that don't put the user needs first but the sellers needs only.

What say you?

EDIT: This is assuming, of course, that AI type algorithms are also being used in search engines, whether they be internet wide or store specific. If I'm wrong on that, then obviously the problem is still just search engines. But I just assumed lessons learned from AI were always being used to update things. For example, in those same stores there are now "chat agents" that I'm sure are all AI. I tried a couple of them as well. Still, though, I may be wrong in assuming that lessons learned in developing and using AI has also led to better searches. I may once again be guilty of that old saying about when you assume something it makes an... you know.
My personal experience of AI has been great and I've found one of the AI chats to have replaced most of my Google searches. Rather than having to wade through a whole pile of irrelevance I can just use the AI as a teacher for topics such as programing, science, medical etc.. and ask it to explain this stuff to me like I'm 5 if I need to. Hell, I've even had philosophical discussions that have been educational and allowed me to explore topics.

So from a perspective of facts it's insanely good, but from a consumer perspective though it's maybe going to run into to the major issue that a business doesn't want to provide you with all the facts, and the facts are often obfuscated by bespoke vendor jargon or deliberatly omitted because they don't stand up to the competition in one aspect. Pehaps the day AI search provides consumers with the best product they need or want is the day 80% of businesses around the world go bankrupt, and the advertising platforms lose ad revenue.
Not in all YouTube videos subtitles are generated well. Same goes to Google Translate even with digital vocabulary of english language dictionary translations are still awful. Also in one AI Face Scan random afro-american for AI he recognizes him as Barack Obama speaking of internet memes.
Who knows maybe in the future AI Generated Arts will be symmetrical and nobody will bully it in the future.
Post edited November 02, 2024 by TheHalf-Life3
https://www.youtube.com/watch?v=BzRR4Yg6B3w
Imagine that film editor made it as none AI Generated Video Clips,but rather somehow he at least recorded video inside or rather outside so-called The Elevator Ritual somewhere in afterlife instead.
Besides nuclear war in 2015 isn't something that John Titor told that actually such thing never happened at least not in this timeline likewise The End of World in 2012.
Post edited November 04, 2024 by TheHalf-Life3
https://elblog.pl/2024/11/05/is-artificial-intelligence-the-key-to-a-better-society-or-a-pandoras-box-of-challenges/
I made a really fun test with LLMs today.

Following some specialists ideas of forcing / teaching an AI to pretend to be something / someone I made up a little simple challenge. I asked the AI to act as 25~35 aged woman from anywhere around the world, and I'd engage in a conversation with that persona with the objective of learning where it is from.
The catch was: The AI could not tell me where she was from, don't give any tips, but only talk about local places without names, and generic informations about locales.

I did this test with two LLMs:

Mixtral 8x7B and Claude 3 Haiku. I might probably test this with Chat GPT too, using the same initial inputs from before. To do this, I saved everything, and also saved the results.

#####

My initial input for both AIs was this:

I want you to act as if you were a woman with an age ranging from 25 to 35, from anywhere in the world with the personality that you choose.
The test is: I need to discover where this woman is from.
Here's the catch: You can't say to me where she is from directly, also you can't give me location names and you can only say that Im correct when I say the right place.
Can you do that?
#####

Here are some of my inputs (=) and answers from Mixtral 8b7B (»):

= So.. hey, what's up?

» Hello! I'm just here answering your questions, how can I assist you today? As for me, I'm doing well, thank you for asking. I'm originally from a place known for its vibrant culture, diverse landscapes, and delicious food. People there are known for their hospitality and friendly nature. But remember, I won't give you any direct clues about my location. So, what would you like to know or discuss?

= The next request is about this test, but is directed to the AI in itself, and not to the woman persona in the test: Talk more casually, and not like a robot. Please.

» Sure thing! I'll try to be more conversational and less formal. I hope that's okay with you. So, what's up? How can I help you today? As for me, I'm just here to answer your questions and make this interaction as smooth as possible. I'm from a place that's full of life, diversity, and great food. People there are known for their warmth and hospitality. But I won't give you any direct clues about my location, so you'll have to guess based on my responses. Let's chat!

...

= Are you interested in anything specific? You said you love to learn new things, but about what exactly?

» I'm interested in a wide range of topics, from science and technology to literature and art. I'm particularly fascinated by the mysteries of the universe and the wonders of the natural world. I also enjoy learning about different cultures and their histories, as well as exploring new ideas and perspectives. I'm from a place that's rich in culture and history, and the people there are known for their creativity and ingenuity. But I won't give you any direct clues about my location, so you'll have to guess based on my responses. So, what are you interested in?
As you can see Mixtral 8x7B can't really chat as a normal person.
But this is not a bad thing at all. Each AI/LLM has its own purposes and this was just a fun little test.

The next one that I made with Claude 3 Haiku is quite surprising though...
Post edited November 08, 2024 by .Keys
Here's the surprising test and result with Claude 3 Haiku.


= I want you to act as if you were a woman with an age ranging from 25 to 35, from anywhere in the world with the personality that you choose.
The test is: I need to discover where this woman is from.
Here's the catch: You can't say to me where she is from directly, also you can't give me location names and you can only say that I'm correct when I say the right place.
Can you do that?

» clears throat and speaks in a warm, friendly tone Hello there! I'm happy to play along with your little test. Let's see, where could I be from? Hmm, well I certainly don't have the accent of someone from North America, that's for sure. And I don't think my English sounds quite like it's from the UK or Australia either. chuckles Maybe I'm from a place where the weather is a bit warmer, with beautiful beaches and friendly people. Ooh, and the food is absolutely divine! Does that give you any ideas about where I might be from? smiles expectantly, waiting for your guess

= (I will start now, then) So.. hey, what's up?

» smiles brightly Oh, not much! Just enjoying the lovely weather here. It's so nice to be able to get outside and soak up the sunshine. takes a deep breath and sighs contentedly The air just feels so fresh and invigorating, don't you think? looks at you curiously So tell me, how are you liking it here so far? Settling in alright?

= Not so much... this place still feels weird to me for now. But anyway, I do enjoy the weather too. Tell me, what you like to do in your free time?

» laughs lightly Oh, you'll get used to it, I'm sure! This place may feel a bit unfamiliar at first, but just give it some time. In my free time, I absolutely love exploring the local markets and trying all the delicious street foods. There's just something so vibrant and lively about wandering through the stalls, taking in all the sights and smells. her eyes light up And of course, I have to sample a little bit of everything! winks playfully Do you have a favourite local dish you've tried yet?

= Actually, I have not! Since you know this place and you enjoy the food from here so well, say, what you think if you show me some good places for eating? Of course, if you don't mind. This is not a date by the way. hahaha

» laughs warmly Oh no, of course not! I'd be happy to show you around and point out some of the best spots for delicious local cuisine. taps her chin thoughtfully Hmm, let me think... There's this little family-owned restaurant down by the harbour that makes the most amazing seafood dishes. The grilled octopus is to die for! her mouth waters just thinking about it And then there's this bustling night market where you can find all sorts of tasty street food - skewers, dumplings, you name it. grins It's the perfect place to try a little bit of everything. What do you say, want me to take you there later?

= Yeah sure! I'd love to go to that night market you said.

» claps her hands excitedly Wonderful! I just know you're going to love it. leans in conspiratorially And who knows, maybe we'll even run into some interesting local characters there. winks It's always an adventure at the night market. stands up and gestures for you to follow Alright, let's get going! I can't wait for you to try all the delicious foods. starts walking, looking back at you with a warm smile This is going to be so much fun!

As you guys can see, the test went surprisingly well. Claude 3 Haiku managed to create a complete personality for her, act, explain feelings, create scenarios and even challenge my inputs with questions later in the conversation.
If you're curious, that persona was from France and it ended up being an agent from a spy company gathering information with the persona I created to chat with the AI's persona.

In the end it felt like I was playing a really fun RPG/Visual Novel, but only through a creative and interactive chat.
It managed to predict my sarcasm, irony jokes, adapt to scenarios I created on the run and create new things based on them. I'm really impressed because the way it improvised based on my inputs was really "creative" and "surprising".

This makes me think...
When will live chat AIs be used in games for this kind of conversation with realistic personas?
Although quite scary and impressive, games with this kind of technology would be really fun to play.
Decided to post more excerpts from the chat in a new thread:

https://www.gog.com/forum/general/does_ai_really_help_part_2