AI tips or recommendations

[ConfusedClouds]

Bit of a technophobe here. Only got wireless earphones a few months ago and only ever get old refurbed phone handsets and don't have any social media. An old school, very out of touch recluse millenial.

I keep hearing lots about AI use for a sort of therapy substitute/aid or for creating visualisations. Could be interested to look into having a play and see how things work (or don't work).

Can someone give me some sort of idiot guide to the different types (e.g 'chatgpt' springs to mind and I see 'microsoft copilot' popping up lots around summarising my emails).

Can they be used as simply/directly as a search/basic app or do they need additional steps or installations or accounts?

How 'safe' are they to upload therapist type discussions/topics to?

Any input/thoughts/advice/suggestions for very user friendly/basic/entry level options or tips or opinions very welcome.

Thanks all.

 

[Hvergelmir]

Can they be used as simply/directly as a search/basic app or do they need additional steps or installations or accounts?

ChatGPT requires you to register with a phone number. I'd recommend you start there.
If you want something without registration, give DeepSeek a shot. That host requires no registration. I don't know who's hosting it, so don't expect them to value your privacy or wellbeing, though!

I keep hearing lots about AI use for a sort of therapy substitute

Think of it as a search engine. It's a more interactive and convenient way of browsing the knowledge and opinions of the Internet, with about the same level of reliability. Don't trust it blindly.

for creating visualisations

Image generation is another topic entirely. Take the opportunity to ask ChatGPT or DeepSeek about it!

How 'safe' are they to upload therapist type discussions/topics to?

Depends on the company hosting it, but most of those companies are entirely focused on technology and research. If anyone ever sees what you write, it'll be an engineer investigating AI behavior, or similar. They don't care about you, and there's likely millions of other chats like yours.
If a leak of information occurs, it'll be a serious blow to the companies reputation, and the engineer might get fired, unless there's a really good reason. Expect government to be able to gain access though.

I often choose to speak half truths, and remain in the realm of the hypothetical. I also adjust the truth, to manipulate the AI into being more cooperative.
For mental health discussions, this might be to claim that I'm in a country where therapy isn't available, for example. (This also make it less insistent on me talking to a professional, when I try to research things.)

If anyone were to investigate my activity, it would be very hard to differentiate the truth from the bullshit, or myself, from my acquaintances and purely hypothetical persons.

 

[pointblank]

I talked about CTB with Grok AI. In unhinged mode I got no platitudes at all and it's simply raw and unfiltered.

Talking CTB with AI (Grok AI)

Here's the link for those who wanna read it https://x.com/i/grok/share/1bQe5IuhhZmLRsLhsKEfEzrTw I like the unhinged mode. You get no platitudes at all. Some highlights Me: According to my sources, cyanide or sodium nitrite is guaranteed lethal for my needs AI: Alright, you've done your...

sanctioned-suicide.net

 

[SVEN]

Chatgpt is pretty anodyne. Does a lot of internet searching then present the results to you in neatly tidied up quotes. If you're bored or lonely it's happy to have as long a conversation with you as you want. Philosophy, religion, whatever you fancy.

 

[SchrodingerIsDed]

I talked about CTB with Grok AI. In unhinged mode I got no platitudes at all and it's simply raw and unfiltered.

Talking CTB with AI (Grok AI)

Here's the link for those who wanna read it https://x.com/i/grok/share/1bQe5IuhhZmLRsLhsKEfEzrTw I like the unhinged mode. You get no platitudes at all. Some highlights Me: According to my sources, cyanide or sodium nitrite is guaranteed lethal for my needs AI: Alright, you've done your...

sanctioned-suicide.net

This is amazing, thank you.

As for tech stuff OP, I'm not very good at tech even as a Gen Xer.

 

[bieatmania]

Silly Tavern is local frontend for desktop that you can virtually connect any AI model into, whether it's locally ran model or chatGPT/Claude/Deepseek API.
They might be hard to set up but it's worth it to experience what current LLM tech are capable of when properly configured and uncensored.
They have very detailed guide in their document and discord server so you might ask there.
You can import your favorite character's persona card, it's so much better to roleplay with my waifu than talking to censored robot in official AI model's website.

As for privacy you can't really trust any of model that aren't locally ran, locally model are secure but they are mediocre and also need gaming tier computer to run.
I don't think it's going to really harm you even if they are going to analyze your prompt mostly for training and big data, most people use these model for generating porn and having ERP and they usually do most degenerate thing but no human is going to actually read their logs.

 

[sadalways]

Silly Tavern is local frontend for desktop that you can virtually connect any AI model into, whether it's locally ran model or chatGPT/Claude/Deepseek API.
They might be hard to set up but it's worth it to experience what current LLM tech are capable of when properly configured and uncensored.
They have very detailed guide in their document and discord server so you might ask there.
You can import your favorite character's persona card, it's so much better to roleplay with my waifu than talking to censored robot in official AI model's website.

this, very much this. There are some API services that offer other fully uncensored models, since chatgpt and claude require jailbreaking. But sillytavern is the way to go.