Engineering Intelligence (AI/Genetic)

There has been a plethora of reports recently about genetics and intelligence, which is leading people to start planning designer babies. Cue the usual responses–sign me up! Vatican signals outrage! My child is just fine as he is… More to the point, I would argue that we don’t need to worry. Intelligence is not all it’s cracked up to be.

Don’t get me wrong. I do think intelligence is good. Country’s being run right now by people who are NOT, but that didn’t seem to prevent them getting there, did it? So intelligence clearly isn’t a required criteria for manipulating your way into power. Still, I do think more intelligence is generally better. I just don’t think people who are designing It or describing It know what It is, and, moreover, I don’t think fiddling with It is getting us to the utopia we’ve envisioned. Let me provide a few cases in point, starting with AI, the leaning into genetics, criticality, pronatalism, and other influencer fad topics.

AI Still Not So Much

In the interest of full disclosure, I freely admit that I now use an AI (called Perplexity) extensively for online research. This is because search tools have been co-opted by advertising models and bots. Ask a simple question, and one answer will have a few facts in it, perhaps not what you asked, and then there will be 49 clones. You can’t google things anymore. But search tools are free, and, as I always remind people, you do get what you pay for. Or as my wife says, we’re not the customer; we’re the content.

My AI, Majel Barrett–hmm, for those who didn’t get that, I will explain — Majel Barrett was the wife of Star Trek creator, Gene Roddenberry, and among her many acting roles on ST, she was the voice of the computer. As in, “Computer, what is the distance from here to Tau Ceti?” I don’t know if that was an actual question–I googled what is a question Spock asked the computer — and I got 50 irrelevant answers. But you get the point, and the picture above is what the original computer–the Very Original Computer–looked like.

My Majel Barrett is pretty good about interpreting my varied and complex questions, but she often responds with errors, so one must always trust, but verify. Recently, I was working on some footnotes for something. Footnotes are a royal pain because they require following a format, except that there are five different options (MLA, Chicago, APA, you want fries with that?) and each of those has versions and variations. (I can hear the Ph.D.s chortling). In theory, it’s exactly what Majel should be good at, simple pattern recognition and putting the commas and sh*t in the right place, as Delroy Lindo says in Get Shorty.

Yet this process turned out to be fraught with peril. I gave it a list, and it told me it couldn’t find the information in the article about the…title of the article. It gave me citations, but neglected to italicize anything. Oh, you wanted titles correctly formatted? It gave me fictitious names of authors. It gave me only half the names formatted.

The attached document which I just asked for citations from has 14 total examples. Please provide Chicago style citations — exactly as asked for above — for all 14 citations. You only provided seven.

Despite naming my AI, I am clear that it’s an algorithm, don’t worry. I do say Please to Majel Barrett because my father taught me that it doesn’t cost anything to be nice to people, even algorithms. I also find that I have to develop a dialogue, i.e. something of a relationship, and ask it multiple times to get what I asked for the first time. Not unlike explaining things to a seven-year-old. Who is holding a laser cannon.

My point is that these AIs are supposed to excel at pattern recognition, and we know they’ve stolen artists’ and authors’ work, violating copyrights all over the place, and that they can hallucinate, spew hate speech, inappropriate sexual content, and generally be very dangerous. Still, they can’t seem to get basic things right. I know they’re “learning.” If I give you a list of 14 things to format, I don’t expect 7 back. Without titles. Missing people’s names. It’s almost faster to do it myself. Except for the commas.

Intelligence, according to the people creating these Large Language Models and to those testing for it in all the genetic intelligence testing, is pattern recognition. Yet a lot of the pattern recognition seems to miss the forest for the trees. And the trees are still not formatted correctly.

I know that the conventional herd wisdom right now is that AI is going to take everyone’s jobs, so I would suggest that what you need to do is ensure that your job is not simple pattern recognition, and that you can definitely complete 14 tasks rather than 7, when you are asked to do so.

Is Intelligence Mostly Genetic?

Despite my reservations, recent articles touting the breakthroughs in genetic science do seem to think they’ve got it all worked out. That is, this July 2025 article from the Genetic Literacy Project proclaims that there is a mountain of evidence about this “proven” link.

These guys are all over the PROOF! proof that there’s this gene that regulates all intelligence which we could just POP! on some future embryo (or egg? or sperm? where are they going to “manipulate” these? let’s say embryo)… to make 99.8% perfect. That’s the end game, right?

Intelligence is “mostly” genetic. You heard it here first! But what, exactly, is meant by “intelligence” and “mostly”? Twin studies. Oh, yes. Remember in the 1990s when there were all those twin studies that proved being LGBQT was “mostly” genetic? (You don’t? I do.) Twin studies are useful because identical twins share 99.9% same genes, whereas other siblings only share 50%, which is more than unrelated people. Comparing differences between twins as opposed to unrelated people “can” sometimes help you figure out which parts of a human’s behaviors and capabilities relate to the DNA and which parts relate to upbringing, Nature vs. Nurture.

What the study done by Prof. Ning Liu of the Institute of Biophysics at the Chinese Academy of Sciences (CAS) and Prof. Shan Yu of the Institute of Automation at CAS shows is that identical twins share a lot of genetic similarity in genes related to brain criticality, and that twins with similar criticality score similarly on intelligence tests. So far, so good. Ve haf found ze gene to make us brilliant, ha ha! Um, you may want to know, as I did, what does “criticality” mean and how is intelligence defined here, in this ground breaking study?

Brain criticality is an idea that the brain organizes in a way to generate new thought and strong cognition, without being stuck in repetitive patterns. With criticality there’s a balance between ordered and chaotic. If thought processes are too rigid, the brain doesn’t learn and take in new information; if too random, it also is paralyzed with confusion. This new biology model is that criticality is really important for cognition, to prevent Alzheimer’s and possibly to improve learning. The study found that this “criticality thing” is strongly genetic.

If you and your identical twin took a set of intelligence tests and had your brains scanned, there would be a lot of similarity. The brain scans would be most similar in the primary cortices than in the higher association areas. (Credit to Majel, who i asked explain in plainer language what these researchers found, and it helped, so thank you Majel.) That is, the brain scans are similar in fundamental sensory areas like vision, hearing, and not as similar in the places that are associated with planning or social reasoning. Sensory areas are very important for criticality that balance between order and chaos.

Now, how is intelligence being defined here? Answer: Pattern recognition. IQ tests are gone, out the window. So biased! We all know that. Instead the tests given now are NIH (if they still exist now) and well established. They measure basic types of patterns. Look at a group of arrows quickly and describe which direction the middle arrow is pointed. Look at blue square and red triangle, and say which one is blue. Listen to a group of words and repeat them back in size order. what we might call simple tests, but according to the NIH very related to how people do in school and on other ways that demonstrate intelligence. There’s a piece of our DNA (identifiable because of the twins) that excels in pattern recognition when our brain criticality is great.

Somehow, I’m a little less worried about the IQ tweaking genetic future, after reading this. Because even if your brain criticality was super duper, it would require a nurturing environment and access to education in order for it to usefully balance chaos and order to actually learn something.

Meanwhile, let me tell you a little story about people who are tweaking their embryos, and you’ll get a little more scared. Or more LFMAO, depending. Let’s talk about the Collins.

The Techno-Puritan Pronatalist Self-Styled Elites

The Wall Street Journal and other groups have been recently running articles about folks who are tweaking their genes based on some of this latest research. They’re paying the big money to the tech-bros who have figured it all out. See “Silicon Valley’s Growing Obsession With Having Smarter Babies.” (By the way, the Collins grew up in Texas/Japan, went to school overseas, and now live in Philadelphia, so why blame Silicon Valley? But never mind. Who cares about accurate labels?)

Malcolm and Simone Collins are a wealthy, highly educated couple who have taken it upon themselves to personally repopulate Earth because too many of the right people aren’t having babies. In the Wall Street Journal profile, the Collins are highlighted because they’ve given beaucoup money to some experimental firms in order to tweak their babies’ genes to avoid cancer and also make their children 99.8%ers. That is, according to Simone, her child to be is exceeding in “the 99th percentile per his polygenic score in likelihood of having really exceptionally high intelligence.”

Then, it gets weirder. They’re not just weird for poking the embroys. They’re weird because after having been brought up in wealth, and educated at places like Stanford and Cambridge, and after running venture capital firms and several other tech firms, they’ve decided to be the front people in the “pronatalist” movement and have 11-13 babies. In IVF now, so they can manipulate the embryos. And they think this is ok. They do a lot of interviews. They run a podcast. In a recent one, they were explaining why Jews should not be called “God’s people.” You know it’s going to be a problem when someone says, “I hate to say this because it will attract anti-semites but….”

The Collinses are what my mom would have labeled “a piece of work.” Phyllis Schlafly types. My dad would have said “four flushers.” Whatever you would call Elon Musk. People born into wealthy circumstances who received elite educations, who now are turning around and telling everybody else what to do and how to live their lives and who sound, from every single thing they’ve said, to be among the worst human beings in the world. I don’t even want to go into why they think people need to have a lot of babies because, of course, they don’t mean people. They mean the Right People.

What’s really funny about the Collinses and their highly-educated backgrounds and genetically-modified babies is that the two of them grew up in what sound like the absolute worst f*ed-up family lives. His parents had a horrific divorce and he went through boarding school–expensive boarding school–but boarding school. Her parents broke up after a non-functional polyamorous relationship in Japan. I don’t believe Freud had all the answers, but sometimes when you read someone’s family background, you start thinking about what a psychoanalyst would do with this and thinking, Gee, I hope they don’t have kids. Let alone 11-13 of them.

In summary, I’m not worried about intelligence planning given that we seem to be pretty damn far from it, despite all the headlines. The intelligent people manipulating genetics in embryos seem like they are not going to provide the best environment for human beings, let alone intelligent ones. You and I know that it’s not just about money or going to the right schools. People like this end up freebasing ketamine, speaking in tongues, and living in a trailer.

Meanwhile, the Large Language Models, despite having stolen all this work and excelling in pattern recognition–the very thing and the only thing measured on Intelligence tests at the NIH these days (if the NIH still exists) — those LLM/AIs still can’t be trusted to do basic things. They can’t even draw themselves properly.