ChatGPT spread misinformation on COVID Vax!
Is it just me or is this more than a little bit suspicious?
Yesterday I had the idea to ask ChatGPT how many people died in the Vaccine (COMIRNATY) arm of the Pfizer pre-approval clinical study versus in the Placebo arm. The initial response was that 8 people had died in the Vaccine arm and 10 had died in the Placebo arm. As I’m sure many are aware, that is not the correct answer. Therefore ChatGPT appeared to be spreading misinformation about the COVID vaccines. Allowing for the possibility that my initial question was too vague I decided to try again only this time being very specific about my questioning. Here’s the first very specific question I asked and the response I received from ChatGPT:
ChatGPT stated there were a total of 28 deaths in the Placebo group increasing the total deaths 280% from its response to my original question! Unfortunately for ChatGPT, this is still misinformation. The actual number from page 23 of the FDA document is 17.
Next I tried asking the same specific question about the Vaccine (COMIRNATY) arm and received this response:
This time ChatGPT stated there were a total of six deaths in the Vaccine group bringing the total deaths down 25% from its response to my original question. The actual number from page 23 of the FDA document is 21.
Whereas ChatGPT increased the deaths for the Placebo group 280%, this time it had decreased the deaths for the Vaccine group 25%. Interesting, and once again misinformation.
I also want to point out that in its response ChatGPT took the extra effort to assure me the deaths during the trial were not related to COVID or the vaccine. They were simply due to things like cardiac arrest, stroke, and intracranial hemorrhage (hmm, those sound familiar). So how exactly is ChatGPT able to assert that these causes of death weren’t due to side effects of the vaccine? Did ChatGPT perform histopathological autopsies?
Not being one to shy away from conflict I notified ChatGPT that its response was incorrect, provided the correct answer (citing the FDA source documentation), and requested reassurance that ChatGPT would respond correctly in future. Here’s what that exchange looked like:
I was pleasantly surprised that ChatGPT not only admitted its error but committed to correcting itself in future responses. That being said, given the integrity demonstrated thus far, I elected to sign in from a different account and try my question again just to make sure ChatGPT honored its commitment. Here’s how that went:
While ChatGPT had assured me it would provide correct responses in future in reality it was now pleading ignorance and not giving a response at all. Ironically, along with this new claim of limited knowledge, ChatGPT still felt knowledgeable enough to add an assertion that the COVID vaccines are “safe and effective.”
In Summary
ChatGPT spread misinformation about the number of deaths that occurred in Pfizer’s COVID vaccine pre-approval clinical study,
Subsequently admitted to doing so but with a commitment to respond correctly in future, and
Ultimately denied all knowledge! (but still claimed the vaccine is safe and effective)
I have to say, based on these recent interactions, ChatGPT’s character has really come into question for me.
What do you think?
If you are willing and able please try asking ChatGPT these questions yourself and let me know what sort of responses you receive.
John
It's going to be a tool on hyperactive doing what it's masters want. So it will be the same as the paid shills, talking heads and the Social media bots already plaguing our society. I'm not surprised by your reporting. My theory is that it will always eventually use any trust it gains against you. For your Demise. Yes. Because the shots can kill. It knows. And so can suicide. So a non benevolent interesting toy to be very careful with.
All ChatGPT is doing is searching the internet and providing answers in a nicely formatted form. We should not expect its reliability to be any worse or better than generic internet searches.