AI can't see it either- the acetaminophen connection that's not hidden at all

Artificial intelligence (AI) platforms are used by major corporations that provide information to the public. But popular AI platforms don’t see the connection between acetaminophen and autism.

AI can’t get it. At least not yet.

AI depends on information provided by humans. They are not smarter than us. Not yet. They have a dictionary and an encyclopedia built into their brain, but they still rely on us for identifying trustworthy information and for identifying the critical nuggets of information that lead in radically new and productive directions.

If, hypothetically, we gave a modern AI a body, deleted its memory, and sent it back in time 10,000 years, it would accept everything that the local tribe accepted. I think it might even blend in with its tribe. But it would not innovate.  It would not become more enlightened than its tribe. It could possibly impress its tribe with its memory, but it would not challenge the tribe to think in new directions.  

AI can’t even comprehend the simplest of facts connecting acetaminophen and autism. Here’s an example how AI fails:

Popular AIs will tell you that acetaminophen became popular as a drug in the 1950s, and they’ll give you a lot of information about the company that got it approved, and the regulatory agency that approved it. This all happened AFTER autism had been described in the medical literature in the 1920s and again in the 1940s.

But, to understand the connection between acetaminophen and autism, it’s important to understand that, for practical intents and purposes, acetaminophen became popular beginning in the 1880s, BEFORE autism was described in the medical literature.

Two drugs were introduced during the 1880s that are important to consider when thinking about what causes autism. One was phenacetin, and the other was acetanilid. The human body converts both phenacetin and acetanilid into acetaminophen.

Current AI programs aren’t smart enough to connect the dots. If I ask specifically “what is the active metabolite of phenacetin”, AIs know that the correct answer is acetaminophen. But they aren’t smart enough to know that this means that acetaminophen was, for practical intents and purposes, widely used BEFORE autism was discovered.

In other words, popular AI isn’t smart enough to deduce that if the apple contains poison, and if poison is dangerous, then the apple is dangerous. At least not when it counts. That’s called “transitive inference”, which scientists have found that even some monkeys comprehend.

Modern AI can define transitive inference, and they even know when it came to be used in mathematics, but they can’t apply it when it counts. Modern AI isn't even as smart as some monkeys.  

Modern AI can’t come close to connecting the more than two-dozen pieces of evidence that tell us, with no reasonable doubt, that exposure of susceptible babies and children to acetaminophen causes many if not most cases of autism.

AI is stuck with whatever the leading members of their tribe think: At the moment, the thinking is that nothing causes autism. Autism just happens. At least a dozen lines of evidence pointing toward the role of acetaminophen in the induction of autism are not compatible with the conclusion that there is no cause of autism.

For example, how can a simple surgical procedure (circumcision) be associated with more than double the prevalence of infantile autism if indeed nothing causes autism?

The AI just doesn’t get it. Not even close. If it can’t get the transitive property, which is one of the simplest forms of logic that, literally, monkeys can understand, how is it going to see a contradiction and sort out the truth?  How is it going to figure out that more than two dozen other bits of evidence point at acetaminophen as the culprit in the induction of autism, and that acetaminophen is used with circumcision, accounting for the connection between circumcision and autism?

AI platforms are getting better, I think. Since early 2025, if I ask the Google AI if acetaminophen is safe for babies, it points to research proving that acetaminophen was NEVER proven safe for babies and children.

Did that AI get smarter? No, humans are realizing that acetaminophen is not safe, and somebody working with the Google AI somehow gave it a nudge. Nicely done, human.