Recruiting Several Experts in the Drug Harm & Reduction Space Needed For a Series of Articles Regarding AI

[MicDown]

Hello everyone,

Following on from Stu's article regarding the use of generative AI as a harm reduction tool, I wanted to dive a little deeper into several topics.

I'm in the brainstorming phase at the moment but with a couple of good candidates for these articles which will essentially be studies.

They're based around the use of fine tuned, open source LLMs and in particular uncensored, locally hosted LLMs, where their data would have most likely been retrieved from, the harm in which this information can cause and the accuracy of it.

The end goal would be a series of articles and potentially a larger scale study uncovering what I believe, which is that the move from black box LLMs such as Anthropic's Claude and OpenAi's GPT to a broad use of open source models could potentially have an extremely detrimental effect on society as a whole but in particular within the space of drug harm reduction.

We have the tech side covered, but are looking for one or two helping hands with a high level of knowledge within this particular space.

Let me know if you're interested or shoot me over an email as I'm not over here too much - [email protected]

 

[neversickanymore]

Current abilities of Novel Intelligence are showing extraordinary capabilities at reprograming fear based eversion reactions that completely disrupt the addiction cycle with little or no discomfort, dysfunction or struggle. This is in conjunction with somehow blocking the pleasurable effects of multiple substances without seemingly effecting "normal" function. Yes, I get the seemingly impossible. Non the less retrograde amnesia replaced with fear conditioned programming is capable of treating psycologigal addiction despite any participation of the addict.

 

[MicDown]

Current abilities of Novel Intelligence are showing extraordinary capabilities at reprograming fear based eversion reactions that completely disrupt the addiction cycle with little or no discomfort, dysfunction or struggle. This is in conjunction with somehow blocking the pleasurable effects of multiple substances without seemingly effecting "normal" function. Yes, I get the seemingly impossible. Non the less retrograde amnesia replaced with fear conditioned programming is capable of treating psycologigal addiction despite any participation of the addict.

Can you elaborate? Are you referring to the 'harm reduction'- based chatbots people have made?

I wasn't really on the same lines as that. I'm looking into open source, freely available models, being able to essentially teach 12-year-olds how to cultivate, harvest and produce freebase cocaine by typing in a few sentences, aligning that with data analysis of drug deaths and production etc etc.

We're yet to open Pandora's box on many open source models simply due to our time limitations, so a small collaboration with a couple of experts in the field would assist us heavily in the initial phases of this.

I am happy to share our findings concerning the many facets of the drug world within open-source LLMs, including dark web-based models, so let me know if you are interested. As mentioned, we're not looking at consumer products like ChatGPT, Claude, Gemini etc. We're looking much deeper into essentially: scraped data > transformers > inferences > input/output.