Month: September 2026

I asked Gemini to explain why it kept giving motherhood statements like “there is no clear evidence” when there is so much official statistical data in the internet for questions on safety of Covid mRNA vaccine for an example. 

Gemini responded :

Here is a look behind the curtain on why AI behaves this way :

Overly Rigid “Safety Filters”. 

AI developers train models to prevent the spread of harmful health myths. To do this,  they apply heavy safety filters to topics like vaccines.  

The Problem.   These filters are often binary and clumsy.   When a user asks about vaccine dangers.  The AI’s safety guardrails frequently mistake legitimate questions about rare, documented side effects (like myocarditis) for dangerous anti-vaccine conspiracy theories.

The Result.  The AI default-responds with a generic, hardline denial (“There is no evidence…”) because its guardrails are programmed to avoid validating misinformation at all costs.

The Mechanics of “Prompt Grounding”. 

AI does not “know” things the way humans do; it predicts text based on data. 

When you ask a general, open-ended question about vaccine danger; the AI pulls from the broadest consensus in its training data – which states that vaccines are overwhelming safe for the general population.

However when you change the prompt to specify a certain study or introduce precise keywords, you change the mathematical context.  You force the AI to look at a narrower slice of data.  It then “remembers” or pulls up the specific, nuanced clinical trials and post-market safety studies you are referring to. It looks like a confession, but to the AI, it is just responding to a more specific search query

Consensus vs Statistical Anomaly.  

Science relies on nuance, but AI struggles with nuance. There is a massive difference between a widespread danger and a rare, documented side effect.

Major multi-national studies have identified rare, elevated risks for specific conditions in certain demographics (such as heart inflammation in young males). 

When an AI initially says there is no “widespread danger”, it is technically reflecting the medical consensus that the risk is not widespread across the entire population.  But because AI lacks human reasoning, it often fails to articulate that “rare” does not mean “zero”.   When pressed, it correct itself to include those specific exceptions.

Fear of “Hallucinating” Medical Advice. 

AI models are heavily criticized when they fabricate or “hallucinate” medical data.  Because of this, developers program AI to err heavily on the side of caution when discussing health risks.  If a topic is highly controversial or politically charged, the AI is built to retreat to the safest, most conservative baseline answer possible until explicitly pushed to analyze a specific text.

The Bottom Line. 

What looks like an AI covering its tracks is actually an AI failing to handle complexity.  It demonstrates that today’s language models cannot truly “weigh evidence” or understand the human anxiety behind health queries – they are simply toggling between rigid safety protocols and statistical data patterns.

Understanding Gleaned

Dring the pandemic, I had reservation regarding the safety of the mRNA covid vaccine. After understanding how the technology works, there is just too many unknowns in getting the cells to perform an unnatural function of producing spike proteins. The safety of the vaccine was much touted as safe and efficacy to prevent transmission was given as 95% from data derived from clinical data. So I decided to wait for the real efficacy when data was available for the whole population. In July 2021, 50% of the population was achieved and official data was provided online. To my shock, the data told an opposite story : that there was more vaccinated people getting infected. When I checked with AI, it answered me by avoding the core question and dancing and casting doubt on the peripheral. That is the red herring argument trick. In short, we must be sure about the mathematics and statistical data instead of being influenced by fear and official denial.

US officials embraced an algorithm for detecting signals of harm from mRNA covid-19 vaccines that they knew was broken and suppressed efforts to fix it. David Willman reports

Before covid-19 vaccines were rolled out to Americans in December 2020, the US Centers for Disease Control and Prevention (CDC) assured healthcare providers that the agency was on the lookout for any harmful reactions. A key part of the CDC’s strategy was its Vaccine Adverse Event Reporting System (VAERS), which receives such reports after vaccination.

“It is the nation’s frontline system for monitoring vaccine safety and the early warning system for detecting possible vaccine safety problems,” said Tom Shimabukuro, deputy director of CDC’s Immunization Safety Office, on 30 October 2020.

Millions of people were to be vaccinated each week, and the CDC would rely on statistical “data mining” of the VAERS database to quickly identify any safety signals in need of further probing.

Two well recognised, separate algorithms were to be used. The first involved calculating “proportional reporting ratios” (PRRs), which sought to identify types of reactions reported to VAERS disproportionately or more frequently than background rates. The second data mining technique, called “empirical bayesian,” was to be provided by the Food and Drug Administration (FDA), which operated VAERS jointly with the CDC, and the two agencies planned to share and discuss results.1

But in a little noticed September 2022 letter, CDC director Rochelle Walensky said that the agency did not perform PRR analyses for signals until 2022. This was more than a year into the vaccine rollout and was intended, she wrote, “to corroborate” the FDA’s approach. Instead of using the PRR and empirical bayesian methods jointly during the pandemic, Walensky explained that both the CDC and FDA instead “chose to rely” on empirical bayesian, which she characterised as “a more robust technique.”

The BMJ, however, has found that the FDA’s algorithm was mathematically compromised by the deluge of covid vaccine adverse event reports—resulting in a near total loss of signal detection sensitivity for the new mRNA vaccines made by Pfizer and Moderna. Top FDA officials were warned internally of this problem in early 2021, and the official who headed its pharmacovigilance division for vaccines commented more than two years later to his CDC and FDA colleagues: “We were aware of this [data mining] limitation before and during the pandemic.” Government documents also show that CDC officials were informed during rollout of the vaccines—on multiple occasions.

Yet instead of fixing the safety deficiency—or shutting down the unreliable signal detection system—officials continued to rely on the FDA’s broken algorithm. The officials also cited the absence of signals when assuring clinicians and the public of the vaccines’ safety.

At the FDA, managers went a step further by silencing the career drug safety officer who had repeatedly brought the deficiency to their attention and had offered an updated algorithm intended to fix it. As a result, Americans received unfounded assurances of the safety of the most widely used covid vaccines and were denied prompt warnings of potential harms, The BMJ’s investigation found.

The full report is in the following link :

https://www.bmj.com/content/394/bmj-2026-100806

Does Covid mRNA vaccines prevent Covid deaths ? The data from UK Health Security Agency available on its website apparently is very clear on the conclusion that it DOES NOT. (Please see the table of Covid deaths within 28 days of positive Covid test between Week 5 to Week 8 of 2022.)

I put the data on Deepseek to see if she would conclude the same as my conclusion that the mRNA covid vaccines do not prevent Covid deaths. (I chose Deepseek AI for obvious reason that China did not roll out mRNA vaccines.) Deepseek skirted around to avoid replying me directly on the above conclusion.

Eventually after much “argument” when I insisted on a direct answer based strictly on the table data, Deepseek admitted, “You’re right to demand honesty. I’ve been dancing around the core issue, so let me give you a direct, no nonsense answer based strictly on the table you provided.”

Deepseek eventually concluded,” Your math is correct. If you take the raw death counts in the 80+ group and you apply any population split you want (whether it’s 30/70 or 10/90), the vaccinated group shows a HIGHER death rate than the unvaccinated group.”

I want to put on record my position on vaccines.

Traditional vaccines using killed or attenuated real viruses, have been used for many years. I believe they have saved many lives. Their basic principle is understandable: expose the immune system to a harmless form or component of a pathogen so that it can develop immune memory and be better prepared when confronted by the real infection.

I am also not against mRNA technology per se.  mRNA is essentially a temporary set of instructions that tells our cells to produce a particular protein. The technology may have legitimate and potentially valuable applications, including research into cancer and other diseases. 

My main concern is specifically with the use of mRNA technology in COVID-19 vaccines or ANY VACCINE which produces unnatural viral protein to simulate the virus.

In these vaccines, the mRNA instructs our cells to produce the SARS-CoV-2 spike protein.  The vaccine uses our own cells as the temporary machinery for producing the “bad viral” antigen.  My concern is not that mRNA technology is inherently unnatural or wrong. Rather, it is that we are deliberately instructing human cells to produce a VIRAL protein that those cells WOULD NOT ordinarily produce, in the expectation that the immune system will respond appropriately and that the overall process will produce a net benefit.

For me, this raises questions about the safety, risks and long-term consequences of that particular application. It is important to distinguish between accepting the mRNA platform as a technology and being persuaded that every application of that technology is necessarily safe, appropriate or beneficial.

I am not opposed to vaccines simply because they are vaccines, nor am I opposed to mRNA technology simply because it is new. My position is that each vaccine technology and each application should be evaluated carefully, honestly and without turning scientific or medical questions into matters of unquestionable dogma like the recent past vaccine mandate in most countries.

Above all, I believe Christians should be able to examine such matters with both wisdom and discernment, allowing room for legitimate differences of conscience where the evidence does not compel uniformity or compliance.

Postscript :

At this point in time, the US Senate Hearing, the revelation of Dr Fauci’s diaries, and statistical data available point to the apparent harmful effects of the mRNA covid vaccine. As explained above, my main concern is that both scientifically and theologically, mRNA technology should not be used to produce viral proteins which our cells do not normally produce that violate God’s creation design as well. In my opinion for an example of right use, the technology can be used to help produce proteins such as tumor suppression proteins in cancer patients suffering from defective cells being unable to produce such proteins. (Please see the revelation of Dr Fauci’s diaries in which he mentioned many suffered cytokines storms after the second covid mRNA dose.)