Toggle light / dark theme

George Dvorsky: Specialization is for Insects

Fourteen years ago, I sat down with George Dvorsky for almost 1 hour and 40 minutes. We argued about everything: transhumanism and the paleo diet, animal uplift and non-human person rights, mass extinction, SETI, and whether alien intelligence would be friendly or hostile.

One quote from that conversation has stayed with me ever since. Robert A. Heinlein:

“A human being should be able to change a diaper, plan an invasion, butcher a hog, conn a ship, design a building, write a sonnet, balance accounts, build a wall, set a bone, comfort the dying, take orders, give orders, cooperate, act alone, solve equations, analyze a new problem, pitch manure, program a computer, cook a tasty meal, fight efficiently, die gallantly. Specialization is for insects.”

In 2012, that sounded like a provocation. In 2026, in the age of AI, it sounds like a survival strategy. When machines out-specialize us at everything, what is left for humans? Perhaps precisely this: to be generalists. To be whole.

Was George right then? Is Heinlein right now?

Watch the full interview and judge for yourself: [ https://snglrty.co/4pU2ZtQ](https://snglrty.co/4pU2ZtQ)

New machine-learning equation accurately assess LDL cholesterol risk

The Martin-Hopkins equation to assess low-density lipoprotein (LDL) cholesterol levels in blood samples has been used by laboratories in the U.S. and other countries to guide efforts to lower cardiovascular disease risk. Now, a simplified machine-learning version of this equation has been shown in a study of millions of U.S. adult and child blood samples to match the accuracy of the original—making it broadly accessible. The findings and code were published in JAMA Cardiology.

“We’ve optimized the calculation of LDL cholesterol and made this equation accessible and easier for all labs to implement,” says Seth Martin, M.D., M.H.S., the senior study author and director of the Advanced Lipid Disorders Program and Digital Health Lab at the Johns Hopkins Ciccarone Center for the Prevention of Cardiovascular Disease.

“Our goal is to enable clinicians and patients to make better decisions about starting treatments that prevent heart attacks and strokes and save lives.”

Braided, exotic particles could build reliable, universal quantum computers

A truly useful quantum computer must be able to run any algorithm, with the same versatility an ordinary laptop offers. Physicists have now shown a new way to give a quantum computer exactly that flexibility, harnessing the capabilities of exotic quantum particles called non-Abelian anyons.

A team of scientists from the University of Chicago Pritzker School of Molecular Engineering (UChicago PME), Harvard, Stony Brook University and Quantinuum built and tested a complete toolkit of operations using non-Abelian anyons, proving for the first time the broad utility of this approach.

“We demonstrated a so-called universal gate set—meaning that if you store information in these emergent versions of quarks, and you move them around, you can do any quantum computation you might want to do,” said Ruben Verresen, assistant professor of molecular engineering at UChicago PME and a co-author of the new study published in Nature.

DNA origami turns secret messages into nano–Morse code that acts as multiplayer molecular encryption

Mathematics has always been at the core of securing information. From online banking to government communications, modern society relies on cryptography, in which complex mathematical algorithms transform readable information into an unreadable form to keep it secure. But as computing power grows and quantum technology advances, these mathematical safeguards are increasingly vulnerable to being broken. That’s where biology stepped in.

Choosing DNA as their information protector, researchers from China developed a multilayer encryption device that takes advantage of the double-helix molecule’s programmable nature to create an origami structure that can store information with high security.

This new system used tiny, custom-built rectangular structures made of DNA, in which researchers stored the message as dots and dashes, creating a nanoscale version of Morse code. To hide the message further, they turned the flat DNA origami surfaces into tubes, physically blocking the patterns from being read or imaged. With the help of a matching unlocking key, the recipient can trigger a reaction that unrolls the DNA back to its flat form, allowing them to read and verify the message.

Testing the limits of what’s possible (and what isn’t) with AI

When can we trust the results we get from AI, and when is learning impossible? Researchers have shown that there are some problems that even the most powerful AI cannot reliably solve, no matter how much data it is given.

The researchers, from the University of Cambridge and the University of California, Santa Barbara, designed “adversarial” mathematical systems to fool any AI algorithm. Like ethical hackers stress-testing a network’s security, these adversarial systems were designed to map out exactly where and why AI prediction breaks down.

Many real-world systems—like those in the oceans, the human brain or robotics—are too complex to describe neatly with equations, so researchers often learn how they behave by using machine learning. But these AI methods don’t always work well, returning unreliable results or poor predictions.

Frontiers: The relentless advancement of artificial intelligence (AI) across sectors such as healthcare

The automotive industry, and social media necessitates the development of more efficient hardware solutions that can implement diverse learning algorithms. This lead article explores the evolution of AI learning algorithms and their computational demands, using autonomous drone navigation as a case study to highlight the limitations of traditional hardware. Traditional hardware, based on the von Neumann architecture, suffers from limited computational efficiency due to the separation of compute units and memory, also known as the “memory wall” problem. To overcome this barrier, this article discusses novel approaches to AI hardware design, focusing on compute-in-memory (CIM) techniques and stochastic hardware.

Entanglement Goes Steady

Two independent groups have demonstrated ways to entangle quantum bits without the need for precisely timed control pulses.

Quantum entanglement describes a link, or correlation, between the states of two or more quantum particles. For example, given a pair of entangled qubits—particles that can be in either a ground state or an excited state—measuring the state of one qubit can inform us about the state of the other. Entanglement is puzzling because it has no analogue in the classical world, where our physical intuition can be relied upon. In particular, entanglement appears to violate the principle of locality: The qubits’ states remain correlated even if we move them far apart before measuring them. But entanglement is more than a curiosity: It is also critical to quantum computing, where it serves as a resource for performing quantum algorithms and remote operations between distant qubits.

Coining the Technological Singularity

Everybody writes about the Singularity now. Almost nobody knows where the word was born.

Not in a lab. Not in a think tank. In the January 1983 issue of Omni magazine, where a mathematician and science fiction writer named Vernor Vinge put a name to the thing the rest of us are still trying to survive.

Think about that. Four decades before ChatGPT, before the AI arms race, before every futurist and their algorithm started forecasting the end of the human era, the framework already existed. Vinge saw the curve. He just needed a word for the point where it goes vertical.

Today, that word is inescapable. Write about AI, about the future of work, about what happens to humanity when the machines get smarter than us, and you are writing about the Singularity, whether you use the term or not. Refuse to, and you owe your reader an explanation for why not. So you are still writing about it.

Thanks to Josh Calder, who dug out and scanned the original page, you can see the exact moment the term entered our vocabulary. A little piece of digital history, hiding in plain sight for 40 years.

Where do you think we are on Vinge’s curve right now? #Singularity #ArtificialIntelligence #Futurism.

China supercharges AI with 100-fold faster optical chip breakthrough

A PERSONAL SUPER COMPUTER “MACRO-CHIP” WITH PHOTONIC INTERCONNECTS:

This will soon become possible by the cheap nano-imprinting of hundreds of smaller microchips, without the need for laser lithography, onto a single monolithic wafer, with these chips’ communicating with each other at light speed as a single system via silicon photonics. A team at Peking University has set this race in motion in a major way by developing an optical system to boost AI speeds 100-fold by optical interconnects between individual microchips. The next step will be placing all of those chips onto a single monolithic wafer with a similar communication system between them. Nano-imprinting at large node-scale of 15 or 20 nm will make it possible to mass produce wafer scale systems that combine all the best types of computing features, from logic gates to optical AI accelerators in one compact package on a single wafer. Consumers will not care if the computer chips in their computers are not 14-mm wide 2-nm node chips printed by expensive extreme ultraviolet lithography, but are, instead, 8-inch or 12-inch wide super computer “macro-chips” that give 1,000 times the computing power and speed of the best Nvidia computer on the market today, whereon the distance of the individual chiplets on the wafer from the central optical multiplexer becomes part of the ingrained clock feature of the chip, replacing the traditional clock-time limit. The mother boards, GPUs and CPUs of these systems will all exist on the same wafer and communicate at light speed, with the equivalent of something like 1,000 VRAM of unified memory.

These developments come as the shrinking of traditional silicon microchips is facing a final limit. In the same way that the Personal Computer became the game-changer in the 1980’s, it appears that Personal Super-Computers will become the new kid on the block in the 2030’s.


Peking University researchers develop new all-optical interconnect system linking standard electronic chips with specific algorithms.

Mind uploading: Can human brains be digitally copied? | Michael Levin and Lex Fridman

Lex Fridman Podcast full episode: • Michael Levin: Hidden Reality of Alien Int…
Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/cv949
See below for guest bio, links, and to give feedback, submit questions, contact Lex, etc.

GUEST BIO:
Michael Levin is a biologist at Tufts University working on novel ways to understand and control complex pattern formation in biological systems.

CONTACT LEX:
Feedback — give feedback to Lex: https://lexfridman.com/survey.
AMA — submit questions, videos or call-in: https://lexfridman.com/ama.
Hiring — join our team: https://lexfridman.com/hiring.
Other — other ways to get in touch: https://lexfridman.com/contact.

EPISODE LINKS:
Michael Levin’s X: https://twitter.com/drmichaellevin.
Michael Levin’s Website: https://drmichaellevin.org.
Michael Levin’s Papers: https://drmichaellevin.org/publications/
Biological Robots: https://arxiv.org/abs/2207.00880
Classical Sorting Algorithms: https://arxiv.org/abs/2401.05375
Aging as a Morphostasis Defect: https://pubmed.ncbi.nlm.nih.gov/38636
TAME: https://arxiv.org/abs/2201.10346
Synthetic Living Machines: https://www.science.org/doi/10.1126/scirobotics.abf1571

SPONSORS:
To support this podcast, check out our sponsors & get discounts:
Shopify: Sell stuff online.
Go to https://lexfridman.com/s/shopify-cv9491-sa.
CodeRabbit: AI-powered code reviews.
Go to https://lexfridman.com/s/coderabbit-cv9491-sa.
LMNT: Zero-sugar electrolyte drink mix.
Go to https://lexfridman.com/s/lmnt-cv9491-sa.
UPLIFT Desk: Standing desks and office ergonomics.
Go to https://lexfridman.com/s/uplift_desk-cv9491-sa.
Miro: Online collaborative whiteboard platform.
Go to https://lexfridman.com/s/miro-cv9491-sa.
MasterClass: Online classes from world-class experts.
Go to https://lexfridman.com/s/masterclass-cv9491-sa.

PODCAST LINKS:

/* */