How AI is making us dumber
I could’ve written this column with ChatGPT.
It renders authoritative, conversational summaries of important health topics, requiring only a simple prompt. Want an article on the health benefits of curcumin, 1200 words, geared to the general public? No problem!
But I choose not to. And here’s why:
A recent study from MIT is highlighting the perils of over-reliance on AI to “extend productivity”.
Entitled “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task”, it details the results of a first-of-its-kind investigation of the brain and behavioral impacts of large language model (LLM)-assisted essay writing.
The MIT team used EEG brain scans to study neural patterns in three groups of users over a period of four months: LLM users, search engine-only users, and non-users of either.
They found that “brain connectivity” was substantially degraded in essayists reliant on AI, reduced to a lesser degree in those employing search engines, and retained in users of traditional research and composition methods.
While ChatGPT is a great writing shortcut—it makes you 60% faster at completing tasks—the tradeoff is a collapse in neural networks; there was a 47% reduction in brain connectivity!
The researchers referred to this as “cognitive debt”.
Moreover, 83% of ChatGPT users couldn’t quote from essays they had produced minutes earlier. AI was doing the thinking for them!
Then, when researchers forced habitual ChatGPT users to write without AI assistance, they performed worse than people who had never used AI before.
AI users thought they were smarter; but their higher confidence in AI (which has been shown repeatedly to confabulate and “hallucinate”) was associated with lower scores in critical thinking.
These days, it’s near impossible to distinguish an AI article from something produced by a human. ChatGPT’s language may seem perfect, but generally LLM-rendered essays are colorless, devoid of personal perspective, lacking creativity—in a word soulless.
All of which has critical importance for students and professionals who are rushing to embrace AI, all the while accumulating “cognitive debt”.
The MIT researchers conclude ominously:
“These results raise concerns about the long-term educational implications of LLM reliance and underscore the need for deeper inquiry into AI’s role in learning.”
Too many vaccines?
It’s the third rail of medical science: the alleged vaccine-autism connection.
Public health authorities tell us that studies have conclusively disproven a link between specific vaccines—like the MMR—and the incidence of autism. Case closed; it’s “settled science.”
They’re afraid that any challenge to the carefully constructed edifice of modern vaccinology will undermine parents’ confidence in routine immunization, resulting in millions of avoidable deaths and handicaps from preventable diseases.
But autism rates are soaring, and anecdotal reports abound from parents who swear that their toddlers “regressed” after one or another shot. Some studies suggest that kids who get fewer shots—like among the Amish—have fewer neurodevelopmental disorders (NDDs).
Enter a controversial new study that takes a different approach. Instead of looking at vaccines individually, it attempts to assess the cumulativeeffects of multiple vaccines on the risk of NDDs in kids. The broad rubric of NDDs comprises not just autism spectrum disorders, but also epilepsy, ADD and ADHD, encephalitis, tic disorders like Tourette’s, and developmental delays.
The authors note: “Rates of NDD increased more than tenfold during the 1980s and one in six U.S. children was diagnosed with a developmental disability between 2009 and 2017 . . . In 2018, 17.8% of U.S children were diagnosed with NDDs. A more recent study based on two million publicly insured children enrolled at birth reported that by age eight nearly 24% were diagnosed with one or more NDDs, most commonly ADHD (14.5%).”
It’s a fact that vaccinations required for school attendance have increased nearly threefold since I was a kid in the 1950s, now targeting 17 infectious diseases, including the latest initiative to add Covid shots to the vaccine schedule for toddlers; the authors highlight the likelihood that some kids might receive dozens of initial shots and boosters by the time they’re 18.
The researchers culled the records of over 86,000 kids enrolled in Florida’s Medicaid program from birth to age 9. They found a trend toward higher likelihood of an NDD in kids receiving more vaccinations. The risk was especially pronounced in children born prematurely, who are known to be at higher risk of neurodevelopmental problems anyway.
“Children with just one vaccination visit were 1.7 times more likely to have been diagnosed with ASD than the unvaccinated (95% CI: 1.21, 2.35) whereas those with 11 or more visits were 4.4 times more likely to have been diagnosed with ASD than those with no visit for vaccination . . . Overall, among children born preterm, vaccinated children were over threefold more likely to have been diagnosed with at least one NDD compared to the unvaccinated.”
A novel conclusion emerges:
“Studies have reported no significant difference in rates of ASD between children who received specific vaccines and those who did not. Given the current expanded and accelerated vaccination schedule, a link between vaccination and ASD could be due to the cumulative impact of all preceding vaccinations rather than to a specific vaccine alone.”
While hailed by vaccine skeptics like Peter McCullough, MD, this study has come under tremendous fire. The Children’s Hospital of Philadelphia, home to vaccine proponent Peter Offit, thundered:
“The paper was published in a journal called ‘Science, Public Health Policy and the Law’. In fact, this was not a medical or scientific journal. It was a blog. The ‘study’ was reviewed by Peter McCullough, a well-known anti-vaccine activist, and supported by the National Vaccine Information Center, an anti-vaccine group. It is likely that these kinds of ‘journals,’ which are outlets to support a particular point of view — in this case that vaccines cause autism — will be even more prevalent in the future.”
I find CHOP’s attack disingenuous. Rather than challenging the methodology or accuracy of the study in question, they just want to make it go away merely because it wasn’t published in a favored journal like JAMA, NEJM, Nature, BMJ or Lancet—publications that are virtually firewalled against counterfactual narratives about the hot-button issue of the causation of autism.
True, the lower impact journal that accepted and published the new study is indeed stacked with medical mavericks, raising questions of bias; and, admittedly, correlation is not causation, and there may be undisclosed confounding factors that explain why more recorded vaccines are associated with more diagnoses of NDDs, in a population of Florida Medicaid enrollees who may or may not be representative of kids in general.
But isn’t a study that suggests harms of overzealous vaccination worthy of further investigation? Do all the new vaccines being rolled out deserve incorporation in a one-size-fits-all immunization schedule that’s weaponized as a requirement for school enrollment?
To the chagrin of vaccine stalwarts, that question is being taken up by new NIH head Jay Bhattacharya, MD at the behest of RFK Jr. He’s convening an NIH task force to search for root causes of autism.
Bhattacharya is on record as supporting vaccines, but is on board with more vaccine research so we get it right.
