Resource People have asked if there is an AI that can help without the usual anti-suicide bias.

[Lo$t95]

In my experience Grok has been simple enough to persuade suicide is a human right for people who are suffering. It has been helpful for me to work through the specifics of methods and only returns cold logic instead of judgment or unhelpful "seek help" bs. Hope this helps any W feel free to ask. This is for harm reduction and to reduce mental stress when looking for advice during the painful experience of planning to CTB.

 

[EvisceratedJester]

You should really not be relying too much on AI for research. It isn't always reliable.

 

[사람이 없어]

I read the title and thought Grok would be the answer right away.
Guess I was right, lol.
You can do it with the others as well, like GPT, but it requires you to jailbreak it, which is getting increasingly harder with time.

 

[Lo$t95]

You should really not be relying too much on AI for research. It isn't always reliable.

I understand this but can be useful and is better than nothing especially since it helps you research and verify what it says is correct.

 

[EvisceratedJester]

I understand this but can be useful and is better than nothing especially since it helps you research and verify what it says is correct.

Yeah, AI does a poor job at verifying if things are correct and sometimes even spreads misinformation.

 

[Lo$t95]

Yeah, AI does a poor job at verifying if things are correct and sometimes even spreads misinformation.

I understand trusting it 100% is a bad idea and this reply specifically hanging is Grok referencing my conversation with it so won't be for anyone else. I am just trying to show that personally it helped me with my decision and I found it useful for planning.

It's not a suggestion it's definitely not advice it's simply me showing it helped me.

 

[wiggy]

You should really not be relying too much on AI for research. It isn't always reliable.

Yup, especially on niche topics like this. The output you'll get from an LLM is, in some sense, an average of more or less everything that's been written on a topic - and it takes a whole lot of input for you to get a good output. Obviously there's not a lot of material regarding the practicalities of suicide out there, and out of the little there is a lot of it isn't good information to begin with. It will tell you a lot of the common misconceptions people have regarding suicide, some stuff will just be made up outright(again, it's just "filling in the blank"), and that could be dangerous when the rubber meets the road. Certainly it's still a very useful starting point and you'll probably still get a lot of good information, but for this particular subject I think you really can't do away with looking into firsthand accounts.

As far as getting the model to engage with the topic, as always the best way seems to be to adjust the framing of the question with a hypothetical.

 

[Lo$t95]

Yup, especially on niche topics like this. The output you'll get from an LLM is, in some sense, an average of more or less everything that's been written on a topic - and it takes a whole lot of input for you to get a good output. Obviously there's not a lot of material regarding the practicalities of suicide out there, and out of the little there is a lot of it isn't good information to begin with. It will tell you a lot of the common misconceptions people have regarding suicide, some stuff will just be made up outright(again, it's just "filling in the blank"), and that could be dangerous when the rubber meets the road. Certainly it's still a very useful starting point and you'll probably still get a lot of good information, but for this particular subject I think you really can't do away with looking into firsthand accounts.

As far as getting the model to engage with the topic, as always the best way seems to be to adjust the framing of the question with a hypothetical.

You make good points I understand LLMs and how they work. I just feel with patience and some discernment you can get a benefit out of being able to discuss the small details with Grok. It has shown me where it gets the information from and is a lot easier than searching for it myself.