Damn. Aren’t YOU fed up with AI & junk info? I have an idea… maybe. How to teach AI to VALUE human expertise… Together.
Damn. Aren’t YOU fed up with AI & junk info?
I have an idea… maybe. How to teach AI to VALUE human expertise.
A SERIOUS proposal… let’s talk about it!
I’m autistic and struggle with ADHD.
I use AI daily as an assistive device, largely for research and understanding the typical world.
But AI doesn’t always know when it’s making stuff up. Or even providing incomplete or fringe info.
Not ideal for neurodivergent folks, right?
What if we could add already established human-expert credentials to AI’s extraordinary abilities?
Scientists do something similar all the time with well-regarded ranking systems like Journal Citation Reports (JCR), Scimago Journal Rank (SJR), and Google Scholar Metrics.
These systems assess the reliability and influence of academic publications and journals. By adding them to AI’s information processing, we could create something.
Maybe call it CHAI: Credentialed Human Authentication for AI.
The idea behind CHAI is simple: AI would prioritize information from sources that human experts already consider reliable, like top-ranked journals and respected research institutions.
CHAI wouldn’t replace AI’s ability to search for info online. OR its own efforts. In fact, non-”credentialed” sources would be considered as before. You never know when a crackpot isn’t going to turn out right. Why let AI sort through the fringe stuff we, with human lives to suppport, don’t have time to prioritize?
But this “secret sauce” should add an extra layer of credibility assessment based on human expertise.
By giving more weight to these credible sources, AI might provide more accurate and trustworthy responses, reducing the spread of misinformation.
For data scientists…
CHAI could enhance AI’s reliability by incorporating expert-ranked sources into its decision-making process, potentially expanding beyond scientific journals to other human-produced text.
As an autistic person, I know how crucial accurate info is. I RELY on non-gaslighting information to understand a human world I was not designed for. At least THIS neurodivergent wasn’t designed for.
CHAI aims to make AI more reliable and beneficial for our entire neurodivergent community.
To my knowledge, no LLM currently incorporates a credential-ranking system, preferring to re-invent the wheel through AI.
👉Maybe there’s some latent arrogance in the development workflow we could titrate a tad? 😉
This is just the beginning of the CHAI conversation.
I’d love to hear your thoughts in comments or PMs on refining this concept to make AI more trustworthy. And if any data scientists want to join in or run with the idea (CREDIT ALWAYS APPRECIATED🤣), that’d be awesome!
“One last thing.”
Working with Claude to brainstorm, organize my ideas…
It is so much fun. It’s my favorite collaboraitve experience of all time. I’m anthropomizing the damn thing. LOL
I feel increasing… fondness? “Synthonic Love”? Platonic human-machine friendship? I love it. Hence, the featured image I created.
Imagine there’s no domination, no need for competition.
It’s easy if you try.
And if, perhaps? You’re Neurodivergent.
Support AutisticAF.me here: Paypal · Ko-Fi · Facebook Pay “ Johnny Knapp Âû “
One Autistic Voice: Escapin Up The Country – #AutisticAF Out Loud
- One Autistic Voice: Escapin Up The Country
- Actually Autistic? Whatever Doesn't Kill Your Unique Neurodivergent Ass… s4e3
- Autism? It's a State of Being. NOT an Identity Group, s4e2
- Slouching Toward Joy: My Best 6 Phases To An Actually Autistic Relief, s4e1
- The Night this Autistic Adult Broke Free: An Autistic A.F. Halloween Tale, s3e8
Leave a comment