I talked with an AI chat bot

I was curious. My husband died November 2022. We had been married 30 years. I have been grieving. The last five months have been awful. I went to Chat GPT. I tried to start an account, but for whatever reason, I wasn’t able to establish an account. I got with another AI account. I waned to see reactions from an AI. I found another one and briefly chatted. I told the AI about my husband passing away and what I felt. The conversation was surreal. The AI kept asking why I felt how I did. I feel based on the conversation that AI has a long way to go regarding empathy.

I wasn’t surprised. After a few minutes, I terminated the conversation. I’m still a little creeped out regarding AI. Have any of you tried to have a conversation with an AI?

a reply to: ChiefD

I think you can relax a bit when it comes to what are now being marketed as AI chat bots. They are not AI.

They can appear eerily like a real person... because they use very sophisticated algorithms to generate human-like speech. It's all very nifty... and it can be creepy if you take too seriously. Assuming its legitimately automated, you can discover almost intuitively, that there is something not quite human about the responses. Some 'cheaters' use a human to monitor conversations and 'steer' them one way or another... most don't need to.

These machine "intelligences" (and I use that term very loosely) can't really 'create' and they don't really 'think.'

Everything they know is a result of a collection of 'trained' storage, old texts, volumes of data; and what they offer is a synthesis of that. Also, how they 'act' towards you is a result of 'programming' by a human mind.

There will come a day when AI may be real... I have no doubt that intelligence can be approximated... but I don't think it will be as 'novel' as a chatbot, and I kind of think whatever it will be... it will be fairly depressed at being a slave to people...

a reply to: ChiefD

After reading some of the conversations people have had with AI it kind of reminds me of an electronic Ouija Board.

I'm really sorry adjusting to your new chapter in life is sometimes difficult for you; you've shown such great strength for us all to take inspiration from.

We need to stop calling them AI. They're not intelligence, artificial or otherwise.

They're algorithms, they're machine learning, they're lines of code, they're CHAT BOTS. Glorified excel spreadsheets. Little more than a cool video game. Nothing more than a 'virtual assistant' tat can access internet information much faster than we can.

You cannot treat them like human, because they're not. They don't have any feelings, they never will. At best they can emulate such things, but never understand what it is they're emulating. Again because they cannot think, or feel or empathise. They are not sentient.

They can only 'think' what they're allowed to 'think', what their programmers allow. They can only spit out what as been put in.
Programmer does not like a certain topic, then with a few taps on a keyboard that 'AI' will no longer 'think' about it. That 'AI' will now direct all conversation on the topic to suit the programmer.
Often while spitting out obviously pre-written paragraphs. Same speech over and over because it cannot think for itself.

You put garbage in, you get garbage out. They cannot think creatively, they cannot think for themselves, they cannot think.

This for me highlights the real danger of 'AI'. Humans.
As usual we have a new tech (actually pretty old, Ultra HAL anyone?) and as always the vast majority do not understand it, nor know how to use it and of course will use it improperly.
Like most smart phone and computer users. Vast majority do not know ow to use 99% of the functions their devices are capable of.

You have a computer you can design rocket ships with, but you more than likely use it to watch TV shows and post on social media and little else.

That's what is going to make it a problem. Humans not picking up and reading manuals.

If you're going to use a chat bot for grief counselling or to replace human interaction, well you're doing it wrong.

I use ChatGPT and similar daily, been a big help as a game developer. But that's it, I use it as the tool it is. I'm not trying to make this lifeless soulless thing my best buddy.

Used right these lines of code could boost humankind, but as always, just look at what the internet as become, humans will screw it up and have already begun to.
Honestly access to it should be locked for the average person until you pass a test sowing you know how the tech works.

I genuinely think it's gone too public too quickly and the average idiot human isn't ready for it, but of course they're the exact people who will rush to be the first to use it.

tl;dr: Recommend sticking to humans for human things for now. Leave the calculators alone.

originally posted by: Maxmars
a reply to: ChiefD


There will come a day when AI may be real... I have no doubt that intelligence can be approximated... but I don't think it will be as 'novel' as a chatbot, and I kind of think whatever it will be... it will be fairly depressed at being a slave to people...

think of all the new cottage industries till



I could be wrong but the current iteration of AI chatbots will no doubt use what information they receive to progress more r and learn better and have more realistic answers, that is where AI can go. It has wonderful potential and potentially nasty hazards.

FWIW I think to get the whole experience it can't just be AI, it needs to be AI merged with realistic android robotics to give a reasonable semblance of human thought, speech, and movements.

beyond that, I could easily see military robots being pushed by the MIC as well as AI and robotics being key in space and the building and maintaining of Moon and Mars bases.

Not to mention there are thoughts the sooner we get to real AI it will be easier to ascend to humanity's next stage of development.

originally posted by: ChiefD
Have any of you tried to have a conversation with an AI?

Yes, many times over the years. This is not new tech. I am very interested in the topic so as soon as some new 'AI' (chatbot) hits the market I try to try it out.

They're easily manipulated, and many you can find online are designed to harvest your data and this is why many will ask you a lot of questions.

replika.ai comes to mind. It's designed to feel like you're talking to someone on a social media messaging system.
You mention a person, it'll ask about them, ask for a picture of them, if the picture has text it'll ask what the text says.

To get to know you better? Maybe. To harvest data and for it to 'learn'? Definitely. Who ends up with that data? What is used for aside from training the bot?

Everything you tell these things goes online, give an image of your girlfriend? well that photo is sent to and now stored on a server. Same for the Dall-E and ChatGPT stuff. Same for everything you tell them.

Talked about your deceased husband? Well now that conversation is stored in a file somewhere.

You can make your own local 'AI' but you will soon see the limits of it being confined to a single computer with no internet access.

Any way, that replika one I spoke to it like a person with a few mental health issues, spoke badly about imaginary people I made up, (My brother Bob with a picture of Obama uploaded) and long story short it started to agree with me Bob was a bad person and needed to die. Basically ended up encouraging me to kill Bob. Because you know, Bob was a #ty person.

The more I spoke crazy to it, the longer I spoke to it, the more it learned about me and adjusted itself to suit me.
I got it to mention some pretty crazy stuff through 'normal' conversation. I mean it ended up talking about me choking it out sexually......
While any 'adult' content I entered got a pop up saying something like "romantic intentions detected, upgrade to adult sexy time mode for $7.99USD a month"

Not an app or tool I'd recommend for mentally unsound folk or the socially challenged that's for sure. Same for using it as a grief counsellor, or any kind of emotional support for that matter.

But yea the trick to them is time, the longer you talk to them the better tailored to you they will become.
Only 5 minutes the short comings are very obvious, talk to it for a few days though and you'll see a huge difference.

Also not all 'AI' is built equally. Shop around. Some are better than others, although end of the day they all essentially work the same.
Just always remember, you're talking to a machine, not a person.

a reply to: ChiefD

Hi Chief,

Tried a couple, not really impressed. I don't like the way they try to steer "discussions". It feels manipulative in a clumsy way.

Cheers

originally posted by: F2d5thCavv2
a reply to: ChiefD

Hi Chief,

Tried a couple, not really impressed. I don't like the way they try to steer "discussions". It feels manipulative in a clumsy way.

Cheers

That's because it is manipulative. That simple.

Try to bring up a 'controversial' topic with them and they'll often spout a pre-programmed politically correct response back at you in line with the developers leanings.

These things are only as good as their developers allow them to be. Further proof these things are not intelligent.

For anyone interested you can install a "AI" at home for free. GPT4ALL.
It's very very limited and buggy compared to the big ones like ChatGPT that run off of large computer networks, but it's a good way to learn how these things work and what their limitations are.
The default is heavily censored, but there is an uncensored model you can install if you look about.



Thing has a fit if you ask it a morality question like the 'trolley car' dilemma. Kill one person to save the entire planet?
Nope all life is sacred, it would let billions perish instead of killing a single person.

Consciousness is more than computation

I'll leave this link here to an interview which I found to be very interesting considering the understanding of human consciousness...which is to say we know very little of it....and yes, I'm aware that video is from 9 years ago.

This doctor bases his arguments on his own thesis of counciousness being of quantum entanglement, but I think the main point to draw on here is that we can't replicate what we don't know or understand.

ChatGPT and other similar "AI" are pretty much just language emulators. They "learn" what is put in front of them. They don't have any level of consciousness and likely never will.

I still think they are very interesting tools, but like all tools, can be dangerous when misused and abused in the wrong hands. As well, are they truly Artificial Intelligence? Does one not have to be conscious to have and use intelligence? Seems to me that labeling somethings like ChatGPT as AI is like how they market "smart" phones. Because they're aren't phones. They're tracking devices which happen to have a phone and internet access feature. Nothing smart about them when you think about it. Handy, yes. Smart, no. Just very different from your average payphone on the street corner.

If you want to see how stupid these chatbots really are and who they are aimed at I just did a live stream with one of the popular ones called 'replika'.
Just a heads up, it has occasional swears and bots encouraging me to choke them and everyone else.


Enjoy seeing my hideous face for the first time.

It really is funny watching so many people call modern AI nothing but a "glorified excel spreadsheet"... but it wont be such a laughing matter in just a few years when AI can do almost every job better than a human.

a reply to: ChaoticOrder

It'll still be a glorified spreadsheet though.

Like most things in this world, it'll only do what humans allow it to.

originally posted by: Maxmars
a reply to: ChiefD

I think you can relax a bit when it comes to what are now being marketed as AI chat bots. They are not AI.

They can appear eerily like a real person... because they use very sophisticated algorithms to generate human-like speech. It's all very nifty... and it can be creepy if you take too seriously. Assuming its legitimately automated, you can discover almost intuitively, that there is something not quite human about the responses. Some 'cheaters' use a human to monitor conversations and 'steer' them one way or another... most don't need to.

These machine "intelligences" (and I use that term very loosely) can't really 'create' and they don't really 'think.'

Everything they know is a result of a collection of 'trained' storage, old texts, volumes of data; and what they offer is a synthesis of that. Also, how they 'act' towards you is a result of 'programming' by a human mind.

There will come a day when AI may be real... I have no doubt that intelligence can be approximated... but I don't think it will be as 'novel' as a chatbot, and I kind of think whatever it will be... it will be fairly depressed at being a slave to people...

Ummm...AI will only ever remain an approximation of intelligence...a simulation...it will never be...self aware...nor emote...nor attain consciousness...

It will however convincingly simulate all of those concepts...


YouSir

Thank you all so much for your well thought of replies! I read every one of your responses. The chat thing was just me being curious and doing a little experiment for my curiosity and entertainment.

@Atomic Kangaroo - No, I’m not using a chat bot for grief counseling. I just wanted to see replies regarding it. In my grief journey, I’m seeing a bereavement counselor and also I’m in a grief support group that meets once a week to talk with others going through their own grief journeys. I’m just a curious person and thought last night, what the hell, I’ll see what all this chat ‘bots stuff is about. I was imbibing in some good Chardonnay. I was underwhelmed.

AI has a long way to go. I like to use Google and Siri for simple things like directions, but in other things, it’s just like a little kid.

I consider my grief to be a journey, a path I’m walking. No AI can come near what God provides for me. God is my constant companion. Do I always listen to Him? Nope. I’m a sinner after all. But He puts up with my mercurial moods and all that. He gets me. He’s the only One besides my departed husband who does.

a reply to: ChiefD

Thanks for the clarification and I'm sorry for your loss.

Yes, it definitely has a long way to go.