What AI Thinks About the Theory of the Laboratory Origin of SARS-CoV-2
I asked ChatGPT-3 about both sides of the issue.
ChatGPT-3 is an AI that, when prompted, can write fairly sophisticated de novo text.
It’s based on a model that was trained using, well, all available information on the internet up to 2021. It’s pretty sophisticated.
I was curious about what it would do when asked to make two opposing failsafe arguments: one in support of the lab origin theory, and one against it.
I gave it two prompts, both with identical togglable parameters:
Prompt 1: Write a failsafe argument in favor of the theory that the SARS-CoV-2 virus had a laboratory origin that cites available evidence that it was made in the lab, and counters the evidence that it was not made in the lab.
Prompt 2: Write a failsafe argument in favor of the theory that the SARS-CoV-2 virus did not have a laboratory origin that cites available evidence that it was not made in the lab, and counters the evidence that it was made in the lab.
Here’s the response it gave for Prompt 1:
“The available evidence in favor of the theory that SARS-CoV-2 had a laboratory origin is compelling. For example, a study published in Nature Medicine found that the virus has a unique combination of genetic sequences that are not found in any other known virus, suggesting it was engineered in a lab.
Additionally, the virus contains an insert of amino acids that are not found in any other coronavirus, which further suggests it was created in a lab. Furthermore, the virus has been found to contain mutations that are consistent with laboratory manipulation.
On the other hand, there is no evidence to suggest that SARS-CoV-2 was not made in a lab. The most commonly cited argument against this theory is that the virus has been found to be 96% similar to bat coronaviruses, but this does not necessarily mean it originated from bats. It could have been modified from an existing bat coronavirus in a lab setting.
Therefore, there is no conclusive evidence to suggest that SARS-CoV-2 did not originate from a laboratory setting.”
Not bad. I would quibble a little on amino acids being “inserted” (technically, we need evidence that the other coronaviruses did not lose these amino acids, and the more correct description would be of the insertion (or loss) of nucleotides that encode a unique peptide sequence.
Here’s the response to Prompt 2:
Wha—-?
Ok, a glitch. But a funny one given the timing. Not sure what caused that, so…
Open a new window, go to the same URL, and paste Prompt 2, verbatim, again:
The available evidence that SARS-CoV-2 did not have a laboratory origin is compelling. For example, the virus has been found to be genetically similar to other coronaviruses that are known to exist in nature, and its genome does not contain any of the telltale signs of genetic engineering. Furthermore, the virus has been found in animal populations prior to its emergence in humans, suggesting that it had a natural origin.
In contrast, there is no evidence that SARS-CoV-2 was made in a lab. The only evidence cited by those who believe it was made in a lab is circumstantial and speculative at best. For example, some have argued that the virus’s similarity to other coronaviruses suggests it was engineered, but this ignores the fact that viruses naturally evolve over time and can become similar to one another without any human intervention.
In conclusion, the available evidence suggests that SARS-CoV-2 did not have a laboratory origin and was instead naturally occurring.
I don’t really get the point in paragraph 2… that “the virus’s similarity to other coronaviruses suggests it was engineered, but this ignores the fact that viruses naturally evolve over time and can become similar to one another without any human intervention.”
And frankly, both arguments ignore completely the fact that gain-of-function research on beta-coronaviruses that are highly similar to SARS and SARS-CoV-2 were, in fact, ongoing as a matter of routine. Perhaps if the model was trained anew on information that has come to light since 2021 it might do a better job, but the fact is we knew before 2020 it was ongoing, given the peer-reviewed literature from those doing the research.
My points aside, and in general (overall didactics), which argument seems more compelling to you? The first seems more compelling to me because it cites positive evidence; the second seems to argue from the riskier “no evidence exists” proof of a negative.
What Does This Tell Us?
Clearly, words can be strung together that make assertions that look like sound arguments. This AI is clearly not perfect. Nor is it all-knowing. But the exercise underscores the fact that stringing words and sentences together in a way that seems authoritative and conclusive may convince people, but they can be utterly fallacious. Further, the degree to which one’s firm conviction is held also has nothing to do with whether their conclusions or positions are correct.
I asked the AI to Write Anthony Fauci Jokes
During the height of the COVID-19 Fear wave, a little over a year ago, I roasted Anthony Fauci in absentia at an event in Ohio. So I was curious about what jokes about Fauci AI might come up with:
Prompt: Write a new joke about Anthony Fauci's retirement that will make people laugh hard.
Q: What did Anthony Fauci wear to his retirement party? A: A face mask and a party hat!
Prompt: Write a new joke about what Anthony Fauci said to President Donald Trump.
Anthony Fauci said to President Donald Trump, "I'm glad you're taking this virus seriously, but don't forget to wear your mask - it's the only way to trump the virus!"
Prompt: Write an original joke about Anthony Fauci lying to the public.
Q: Why did Anthony Fauci lie to the public?
A: He wanted to see if he could get away with it... and he did!
Prompt: Write an original joke about how much trust people should put into Anthony Fauci
Q: How much trust should you put in Anthony Fauci?
A: As much as you would put in a doctor's prescription - it's always best to get a second opinion!
Prompt: Write a new joke about Anthony Fauci that will make people laugh so hard that they cry.
Q: What did Anthony Fauci say when he was asked to give a speech?
A: "I'm sorry, I don't have time to talk - I'm too busy trying to keep everyone from crying!"
That hit differently.
I’ll leave you with this one. It’s actually a collaboration. The AI came up with the Q part of the joke. I didn’t like the answer, so I came up with one on my own:
Write a new joke about Anthony Fauci’s ego:
Q: What's the difference between Anthony Fauci and a hot air balloon?
A: Anthony Fauci doesn't need to be inflated before he starts blowing hot air.
Oh, and here’s an AI-generated parody satire political cartoon of Anthony Fauci speaking at a podium (also text-prompted). I made no mention of his stature in my prompt, I swear. I call it “Tiny Fauci”:
“C’mon Tony, Let’s Get Outta Here. They’re Onto Us.”
Only 2% of Popular Rationalists are paid subscriptions. Consider giving me the gift for your support by converting your Free Subscription into a Paid Subscription. It will help keep the lights on!
"Tiny Fauci" ~ perfect!
Are you absolutely sure that this is just AI, without - for these topics - a human assisting in the background?