Anyone familiar with GC architecture? I’ve lost my original training process and there’s way too many changes since I’ve been on there last. Also, anyone know how open.ai integrates into the platform. I have too many security functions for HIPAA purposes.
I need the basic structure on how to re-config Vision with their new NLP API. I am having difficulty carrying over my PDF datasets into the AutoML.
I am also in search of a coder to help fine-tune the cloud…
Thanks!
I’m not familiar with GC specifically but I do cloud architecture professionally and have worked with HIPAA, Part 11 compliance, ISO 27k, etc
Great! Can I hire you 
? My psych brain has gone into a state of psychosis trying to read all the notebooks and figure out the coding . I can provide the general structure and show you how the Psychology of Law works. I’ll add you to the cloud as an admin? Maybe you can show me how NLCA works in there? 
BTW: made a good contact with the Post Naval Graduate Academy they are very receptive to the Mental Healthcare API. I’ll work on getting a sponsor for security clearances.
@Open.AI community the I/O INSTITUTE is looking for psych associates, social workers, marriage & family therapists, I/O psycho interns, coders, and other unidentified groups who want clinical hours under the supervision of my license .
@Coding world, which is beyond my scope of service, but I understand the psychology of text; let’s create a credential to certify your work in cross training within psychology. I think there is a future legal argument here against non-psych credentialed coders that could potential protect your work. We are slowly reaching out to the FDA. Hit me up at dr.ames@i-o.institute , here , or on Linkedin: www.linkedin.com/in/amesai . There’s an I/O INSTITTUE group here I’ve neglected for any social media folks looking to contribute.
With Fine-Tuning, I’m basically going back to the drawing board on NLCA. For instance, I wanted to make the simplest possible version with just the Core Objective Functions as the “brains”, but I just don’t have the energy/motivation for it right now (been working on other projects). GitHub - daveshap/COFBOT: Simple chatbot based on my Core Objective Functions and Discord - I think that only the Core Objective Functions will be necessary in the long run, even for high stakes situations. For instance, “reduce suffering”, with enough thought and sophistication, will allow the bot to know that giving bad advice can be very harmful, and so it will qualify anything it says as high risk. It all depends on how far ahead it can think, and how complex it can think.
But yeah, integrating something like this directly into cloud was on my original roadmap. Ultimately, I want an intelligence information companion to be ubiquitous, available anywhere and any time on any device, so the cloud is the natural home for it. So google cloud sounds good, plus they have other services built in such as speech-to-text and text-to-speech that will help with a natural language interface. Presenting it all as a standard REST API that can be consumed by any device/platform is ideal. Because then the free market will produce devices and services to consume your API and you don’t ever have to worry about it. Smart watch app? Let someone else hook in. Smart home speaker? Let someone else worry about it.
Privacy and data governance will take some doing, since each instance/user will contain privileged information. I wrote about this briefly in my book, if you recall. One solution is to have all privileged data live on a user’s device (their phone/computer/smart home) and encrypt it. Alternatively, we can keep it in the cloud and also encrypt it so that only they can access it via their instance. Part of the business/research model would be to allow users to opt-in to share their deidentified data for a discount to the service. But considering that even SQLITE worked in my tests (individual users won’t have much data if it’s all just text), it shouldn’t be a problem to keep user data encrypted at rest. Still, it will take some doing.
But yeah if you make me an admin I can poke around. Even without that, I can start making some architectural diagrams.