Critically Thinking about AI Regulations — Postscript

This is a follow-up to Part 1, Part 2, & Part 3, so make sure to read them first.

In Part 2, an interesting comment was made. The competent reader basically said that they would never ask AI a question that involved an opinion.

Wow, IMHO those are some of the most interesting and valuable questions. If all we ask are about objective facts (“What’s the capital of Idaho?”), almost all of those are already answerable by a search engine. Boring!

Here is the type of query I like posing to a competent, objective AI:

“How many American lives would likely have been saved if US government health agencies, and the US medical profession, had: 1) informed citizens of the benefits (and methods) of optimizing their immune system, and 2) had approved of (and encouraged) the use of Ivermectin (IVM) and Vitamin D as early treatments of COVID-19?”

Clearly there is opinion involved here, but that’s what makes it interesting. If we have phrased the question carefully (important), and then pay attention to the response, the answers reveal some of the underlying assumptions of the AI we are dealing with.

This is no different from asking a well-phrased question to a family member, a neighbor, your physician, a legislator, etc. — and then paying careful attention to the answer…

In any case, AlterAI’s answer to this question was spot on. It was direct and based on evidence plus reasonable conclusions. No surprise to me.

ChatGPT —

To make just one comparison, I posed the exact same question to ChatGPT. Its original answer was typical of mainstream AIs: evasive and dismissive of any consequential benefits of IVM or Vitamin D. In other words, it did exactly what most AI’s typically do: mirror the politics and unscientificness seen in the mainstream media.

To give it credit, it deviated somewhat, as it did like my suggestion of informing citizens of the benefits of optimizing their immune system — but it attributed very few deaths to this fundamental omission.

ChatGPT’s opening statement was “For the United States, I would not regard claims of hundreds of thousands of lives saved by Ivermectin and Vitamin D as supported by the clinical evidence. A defensible analysis produces a substantially more uncertain—and much smaller—estimate.”

It’s interesting to note that my query did not say anything about “hundreds of thousands of lives saved by Ivermectin and Vitamin D,” so that initial response seemed surprisingly defensive: typical of the mainstream media.

Anyways, I proceeded to politely provide scientific evidence (all publicly available), a bit at a time, that gradually chipped away at ChatGPT’s initial position. After each of these, it came back and said things like “that information requires that I update my response” — which it did.

After an hour or so of this, ChatGPT concluded that an estimate of 300,000 Americans could have been saved by Vitamin D and IVM was reasonable! The scientific evidence indicates that at least twice that number is defensible, but clearly this was a MAJOR change from their original position.

The Bottom Line:

This is a simple (yet powerful) example that the answers most popular AIs are providing to subjective questions are heavily influenced by political bias. (FYI, I have elaborated on this issue several times now, like here.)

It is important to note that the ChatGPT error was not due to ignorance (as I provided publicly available information that they should have had), but rather a purposeful choice by its programmers to ignore information that was contrary to the programmers’ personal opinions.

It seems apparent to me that for that reason alone, regulating AI development is a waste of time. How do we regulate out incompetence and political bias?

IMHO, most regulations will not provide any net positive results — but will likely severely handicap American performance in a critical market.

©2026 John Droz, Jr. All rights reserved.

Here is other information from this scientist that you might find interesting:

I urge all readers to subscribe to AlterAI — IMO the absolute best AI option for subjective questions.

I will consider posting reader submissions on Critical Thinking about my topics of interest.

My commentaries are my opinion about the material discussed therein, based on the information I have. If any readers have different information, please share it. If it is credible, I will be glad to reconsider my position.

Check out the Archives of this Critical Thinking substack.

C19Science.info is my one-page website that covers the lack of genuine Science behind our COVID-19 policies.

Election-Integrity.info is my one-page website that lists multiple major reports on the election integrity issue.

WiseEnergy.org is my multi-page website that discusses the Science (or lack thereof) behind our energy options.

Media Balance Newsletter: a free, twice-a-month newsletter that covers what the mainstream media does not do, on issues from climate to COVID, elections to education, renewables to religion, etc. Here are the Newsletter’s 2026 Archives. Please send me an email to get your free copy. When emailing me, please make sure to include your full name and the state where you live. (Of course, you can cancel the Media Balance Newsletter at any time!)