Information asymmetry. AI adding value?

Investment Analysis Clubs Macro Economic Trends and Risks
1

Hiring a medical billing code specialist to defend against outrageous medical billing costs $3k?
AI democratizes access to the info?

Institutions INTENTIONALLY create confusion, counting on people being too uninformed, not to mention desperate (emergency, bereaved, etc), too emotional, too angry, too confused, AND too impecunious,
and just give up and pay whatever the institution deemed appropriate.

This guy suggests 8 steps to use AI (LLMs) to get info that supports “symmetry” in adversarial situations.

He says AI users are “not getting advice”, but rather “they are conducting Institutional grade information gathering”.
Ie, using AI to “do your own Due Diligence”.

Start about the 5 minutes mark.

  1. Have the LLM parse the docs provided by the institution.
    Get the LLM to put into layman terms.
    Audit the docs, bills you receive against the instruction’s own rules.

  2. Use LLMs to cross reference info from multiple authority sources.

  3. Use LLM to draft correspondence in the appropriate legalese.
    Use LLM to present your case in a professional manor.

  4. Use LLMs to determine what is the rule book that governs the domain in question.
    Get the LLM to find the specific rule/s that are being violated.

  5. Use LLM to find true violations, not just marginal disputes.
    Identify very clean, clear, binary violations.

  6. Use LLMs to calculate objective anchors from authoritative standards.
    Establish a defensible position based on published benchmarks.

  7. Verify the info the LLM delivers.
    You, the user, are liable.
    It’s up to YOU to verify that the rule/s the LLM provides, actually exist and say what the LLM interprets.
    You gotta fact check the LLM output.

  8. Let the LLM draft verification prompts to catch its own mistakes.
    Let the LLM fact check its output.

Non obvious:
Investigation must be done prior to confrontation.

Recognize the “framing” manipulation.
The institution wants you to accept their decisions as legitimate, and will frame the decisions to “invisibly” support that.
Your goal is to reframe the situation as “the institution is wrong … And here’s the objective evidence”.

Once you submit your response to the institution’s decision, use the LLM to help understand their reply.

Ask the LLM how it can better help you.
Each of these steps requires methodical investigation.

:keyboard: :military_helmet:
ralph

7 Likes
2

This is the Achilles heel of the entire process - and why AI might make things worse.

People ARE uninformed. If they are uninformed and generally unable to figure out medical billing, then assuming they are informed enough to use AI in a savvy enough manner to decipher the billing is prone to failure.

The guy lists 8 STEPS someone is supposed to follow - without error - in order to do their own LLM/AI research - yet your closing line:

Each of these steps requires methodical investigation.

Virtually guarantees it isn’t going to happen.

6 Likes
3

LOLOL.
Some folks will. Some won’t.
I didn’t post it for those who won’t.

AI n LLM are still really new. Most of us don’t have facility with how to effectively use em.
Let alone how to be methodical.

I’m paying attention to various “how to” sources.
Personally, I don’t follow these steps for every search or LLM/AI that I do.
Ie. I’m selective. And it’s a learning process.

I expect most on METAR also are selective.
And learning.

:student:
ralph

4 Likes