Vol. 1, No. 5 — “Thank you for waiting, this time of year is so busy with everyone trying to land a Fall birthday. First of all, great news: all three of the embryos are viable! One will have blue eyes and the other two green. One of the green-eyed candidates has a predisposition to Alzheimer's but will be in the 95th percentile for height and lean muscle mass.

We will edit out the Alzheimer risk of course but the rest is up to you. Let us know your preference on stature and eye color, then we’ll move through any catalog options for appearance and intellectual capacity you’re interested in.”

While the world wrings its collective hands over the implications that highly intelligent computers may have on us in the future, a quaint little technology named CRISPR has been making headways since its invention in the early 2010s. Never heard of it? It’s a programmable genetic editor which works on living organisms. Yeah, that old chestnut.

CRISPR stands for Clustered Regularly Interspaced Short Palindromic Repeats. That doesn’t mean anything to me either and it doesn’t really matter. Unless you’re a genetic biologist, any technical article about CRISPR is basically inscrutable. Let’s stick to the basics: what can it do and what makes it special?

First of all, genes are just sequences of DNA (deoxyribonucleic acid), a molecule that is unique to every living organism. Some of these sequences are short and some are long. There are different ways to measure it but humans have about 20,000 genes.

DNA consists of only four different ingredients: adenine, guanine, cytosine, and thymine and we represent those with the letters A, G, C, and T. Everything that has ever lived can be represented by a (very) long series of these four letters. Human DNA is several billion letters in length. While that sounds big, it’s still less than one gigabyte of data.

This makes DNA our “software” so to speak. But where a computer is programmed with ones and zeroes, we are programmed with A, G, C, and T. In some circles, it’s referred to as “wetware”, the software of the body. If we have blue eyes, that specific piece of our DNA code might read “GATCC” but if our eyes are green it reads “GACTT.” The same goes for our metabolism or our ability to fend off the flu, etc. These are all represented by sequences of letters in our DNA.

Now, wouldn’t it be nice to be able to modify bits of DNA to elicit some desired change? Like eliminating a person’s predisposition to Parkinson’s? Or erasing male-pattern baldness? Science has been pursuing this goal since the 1960s with initial successes in the 1970s. However, the methods developed were highly specific, always targeting a single gene to modify for a single purpose. Every time a new gene was targeted, a completely new method needed to be devised. This is like building a new computer from scratch for every application you want to run. It can be done, but it doesn’t scale. Enter CRISPR.

Jennifer Doudna and Emmanuelle Charpentier invented CRISPR in 2012 while researching how bacteria fight off viral infections. They noticed that a certain type of protein, Cas9, would cut and modify DNA structures in the bacteria to remember which viruses it had seen before. Then they figured out a way to instruct the actions of the protein. After that, nature takes its course. Unleash a pre-programmed CRISPR-Cas9 payload into some cells and you can do whatever you want with their DNA.

CRISPR can be programmed to recognize any specific gene sequence and replace it with a different sequence. It is a general tool to modify DNA and does so with extreme precision. Previous methods could replace sequences successfully but would always modify unrelated codes inadvertently. Not desirable, to say the least. No one hires a home renovation crew that does excellent work on kitchens and living rooms but will always accidentally turn another random room into a reptile terrarium.

Their invention netted them the Nobel Prize but also ignited a huge ethical debate around genetics. Having advanced in capability much quicker than expected, they themselves called for a moratorium on its use in 2015. Debate increased but work continued. In 2018, it was announced that two babies were born with CRISPR-modified genes. The biophysicist behind it had modified the DNA in twin girls to immunize them against HIV. One of their parents was an HIV carrier and had given consent to perform the procedure. The scientist involved in the HIV case went to prison. Little is known about the children’s health.

In 2019, the scientific community called for another moratorium. Now, more specifically, on editing heritable traits. Editing one human’s DNA should not be passed on for generations. This is an important distinction as well to the work that made the COVID-19 vaccines possible. These operated on mRNA. RNA is not permanent. DNA is.

Outside of humans, what else has CRISPR been used for? It was used to modify tomatoes so they had higher levels of GABA which can lower blood pressure. It was also used to decrease leptin in a species of fish. Leptin is a hormone that signals the brain when you’re full. Without it, the fish grew much larger than they normally would. Both of these products have been for sale in supermarkets since 2021.

Many other examples exist where CRISPR was used to eliminate disease susceptibility in a certain crop or increase its resistance to cold or heat, etc. In general, to make better use of resources and increase the profitability of crops.

So how close are we to designer humans? Well, the fictional example up top contains less fiction than you think. Fertility clinics are already sequencing the DNA of viable embryos to let you pick eye color and screen for things like Downs syndrome. Now, editing the embryos? That’s still a ways off…at least through official channels.

The beautiful thing about CRISPR is also what makes it a bit scary. It’s quite easy to use. Research labs around the world now have a reliable tool to research cures for genetic disorders. For example, treatments using CRISPR are already available for sickle cell anemia. Patients with this disease currently have a life expectancy of less than 55 years. There are cancer treatments in the works. Neurodegenerative disease treatments and the list goes on. The upside is tremendous!

The real question is: how long before we have DIY DNA kits? Like a chemistry set from the toy store. The instructions might not show you how to build anything dangerous but if all the ingredients are there, it’s only a matter of time.

Vol. 1, No. 4 — On November 18, the most powerful rocket ever constructed was launched from a sleepy peninsula on the US/Mexico border in southern Texas. By one calculation, it produced over 40 million horsepower during the first second of full burn, enough to raise the 11,000,000 lb vehicle about 25 feet off the ground.

After clearing the launch tower, it accelerated to nearly 15,000 mph and reached an altitude of 485,000 feet in just over eight minutes. A few seconds later, its control systems decided to self-destruct the vehicle after guidance failures.

This is only the second flight for SpaceX’s full Starship design and it bested the first flight’s performance in every aspect. But, predictably, headlines about the event were decidedly negative: Musk’s rocket blows up again.

Let’s take a look at SpaceX’s progress with a bit of context. We should be jazzed about how quickly space travel and transport are becoming a reality.

Ten years ago, all rockets were single-use. “Rockets are impossible to land” was the generally accepted opinion. Makes sense too: how do you even go about landing a wingless silo tumbling back from the edge of space? Turns out, it was unthinkable before computers caught up to the task. The calculations and sensors involved have so many different variables to consider, that it was impossible to design a solution in advance. It had to be honed by an artificial intelligence.

In 2015, after many attempts and several near successes, SpaceX successfully landed their Falcon 9 booster. They managed to stall the 135 ft long rocket and flip it around 330,000 ft above the Earth, then power it back to the launch site. It comes in pretty hot, about 5,000 mph. So it fires again to scrub down to subsonic speeds. And then again, at the very last second to land the 50,000 lb vehicle undamaged and upright. The ultimate egg drop test. Oh, and this is done with such accuracy that the rocket legs don’t deviate more than 30 ft from their intended target. That’s quite the trick. A trick they’ve now pulled off nearly 250 times. So often, it’s old hat.

Now the Falcon 9 is not a “super heavy lift” launch vehicle. It can only put 41,000 lbs into orbit. A super heavy distinction means at least 110,000 lbs. The Saturn V which took us to the moon could heave 311,000 lbs into orbit. This was the most powerful rocket ever flown and held that title for over 50 years. Until Starship. Starship is rated to carry over 440,000 lbs into orbit on crewed missions and over 660,000 lbs when operating autonomously.

Did SpaceX just strap a bunch of Falcon 9s together to achieve this? No, but they did try that too. It’s called the Falcon Heavy, uses three boosters, and still only reaches 141,000 lbs to orbit. So how did they now nearly 5x that result? SpaceX started from scratch on Starship. They threw away their workhorse and designed a new system from the ground up.

Starship is 400 ft tall and 30 ft wide, nearly three times as large across both dimensions compared to the Falcon 9. Its body is made of cheaper stainless steel instead of a complex aluminum alloy. And perhaps most importantly, it uses a brand new engine, the Raptor, moving away from the ultra-reliable Merlin engine used in Falcon 9.

The Raptor is an engineering marvel. It produces twice as much thrust as the Merlin. Its chamber pressure is a record-setting 5,100 psi, blowing away Merlin’s 1,400 psi. Now maybe those numbers don’t mean much to everyone but here is one that should: how many engines does each vehicle use? The Falcon 9 uses, unsurprisingly, nine Merlins. Starship has 33 Raptors. Thirty-three engines, each of which weighs 3,500 lbs. A total beast.

What is this beast for, though? Well, it has one metric that hardly any other super-heavy launch vehicle has: payload capacity to Mars orbit. Starship wasn’t designed to just slingshot satellites into orbit. It was designed to deliver heavy payloads to Mars and the Moon. It was designed to allow us to build infrastructure on other heavenly bodies and do so with full reuse like the Falcon 9 so these missions are affordable.

So important was this objective for the design that the new engine burns a completely different fuel as well. It uses liquid methane and liquid oxygen also known as methalox instead of the more traditional RP-1 fuel used on the Saturn V and Falcon. Why? Methane and oxygen are much easier to produce on Mars than highly refined kerosene, the basis of RP-1. Some say that travel to Mars will be a one-way trip, either due to danger or expense. Could be. But it definitely will be if we can’t make any fuel once we’re there. Same goes for the choice to use stainless steel. Much easier to produce.

Seeing all 33 engines light up on Starship and rocket away from Earth last week was a huge moment for me personally. I herded my four-year-old, Nikola into the living room to watch it. I’d imagine, by the time he’s fortysomething like me, all of it will seem quite quaint. We’ll nearly certainly have a Moon base and a Mars base. But will we already have a hotel up there somewhere? I’m already saving my pennies to take a trip. Forget the retirement Winnebago, let’s check out the neighborhood!

Vol. 1, No. 3 — It had become common practice. A slightly degrading but ever-so-necessary ritual before the start of any in-person competition. A quick wave of an oval wand across the ears, once up the back of the neck, and then a neon pink beam of light into each eye. Done. No leaks or bugs. From chutes and ladders to blackjack to chess, each participant has to be sure they are playing against another human. Well, against another human mind at least.

Computers have been really good at games for quite a while. Already in 1952, a tic-tac-toe simulator was programmed which could play perfectly against a human opponent. The computer it ran on filled an entire room and required 11 kW of power to operate.

To get an artificial intelligence to “play a game perfectly” is also called “solving” the game. It can play a game at the highest possible human level and will never make a mistake because it already knows every possible game. There are 255,168 possible games in tic-tac-toe. If you go first, you have 131,184 ways to win. If you go second, there are only 77,904 ways to win. A draw occurs in 46,080 additional cases.

Those are already pretty big numbers for such a small game, right? A quarter-million different games from nine positions with two different pieces? It’s surprising. But computers are great at this kind of task. They know every possible game and look up a path to the next best move which leads to a win.

What about Connect Four? It also involves two different pieces but the grid is 6x7, not 3x3. So we’ve increased from nine to 42 positions. Maybe five times more complex? Not quite. Connect Four has 4,531,985,219,092 possible games. It’s nearly 18 million times more complex than tic-tac-toe.

Computers can still easily play a perfect game of Connect Four. Even at 4.5 trillion combinations, it’s well within their reach. And checkers? Checkers has 500 quintillion possible games. Still solvable. If you play against a computer, it will just look up which game it’s in and know the next best move.

But what happens when we cross the line into games so complex that we cannot know all possible outcomes? Chess is one such game. It is estimated to have between 100 quindecillion and 100 quinvigintillion combinations. Those are unimaginably large numbers that flirt with the number of atoms in the known universe. The computer can no longer know beforehand which move is best. It needs to decide independently. It needs to learn to play the game.

In 2010, a little research lab in England named Deepmind showed up to help do just that.

Previous attempts to “teach” a computer a certain skill, such as playing a game, required the programmers to explain all the rules of the game, known strategies, etc. Deepmind built a new system that enabled the computer to learn by doing. Their program would play a game over and over and eventually got better after thousands and thousands of rounds. And here’s the kicker, it wasn’t built for a single specific game. It was a general game playing intelligence.

By 2013, their gaming AI could play a handful of classic Atari games (think Space Invaders, Pong, Breakout) at superhuman levels. In 2020 they announced that it had now surpassed human performance on 57 Atari games. It had learned how to play games in general and done so simply by playing them.

Now the world has seen successful chess computers at this point. Already in the late 90’s the famous matches between Garry Kasparov and IBM’s Deep Blue computer proved that AI could topple even human grandmasters. However, Deepmind set their sights on unconquered territory: the game of Go. They would use their self-learning technology to build an artificial intelligence capable of beating the Kasparovs of Go.

The game of Go is infinitely more complex than chess. Add a couple hundred zeroes to the number of possible moves in chess and you’re getting close to the complexity of Go. It’s wild.

There were several versions and improvements to Deepmind’s technology to accomplish this but in 2016, their AlphaGo system beat world champion Lee Sedol in a series of five games. Lee Sedol is a 9-dan ranked Go player, the highest possible rank in the game. There are less than 200 in the world. This was an incredible outcome. Until this point, the best Go computers in the world could be beaten by amateur Go players. The sudden advancement in computer capability sent shockwaves through the community. It was not expected to happen for another decade. Something truly new had been unlocked.

Self-learning is the natural way to learn. We try something, we see a result, and we learn. We try again. However, once a computer is capable of self-learning, the playing field is tipped very steeply in its favor. It comes down to speed.

A human can get good at chess by playing chess. One estimate to achieve the top rank of Grandmaster in chess places it at an investment of 12,480 hours. So if it were your job, you would play chess for eight hours per day, five days per week, 52 weeks per year for six years to reach that level. Now how long would that take a computer?

The Deepmind computer began with no information about chess at all, only being told if it had won or lost a game. In nine hours it had played 44 million games against itself and learned enough to play at a superhuman level. Time and experience have a different relationship when it’s a silicon brain doing the thinking.

So far these “general” intelligence agents have still been limited to related tasks. But what happens when they become truly generalized? What happens when we have a single AI that can play chess, write a song, file your taxes, be your therapist, and tell you how to cook a mean omelet? That’s not AI, it’s AGI: Artificial General Intelligence. Once we crack AGI, the singularity may be peaking around the corner.

The ability for a computer to learn anything nearly instantaneously walks the thin line between exciting and terrifying. Good thing they’re tied to our desks and power outlets, right? As long as no one loads this stuff onto a robot or a drone we’ll be fine. Wait…what’s that? You did what now? Oh boy…

Vol. 1, No. 2 — Yes, the light was red. Yes, you were a bit distracted but wow that was a fast yellow. No, it didn’t really endanger anyone. And now, a minute later, it looks like you got away with it. No harm, no foul. Lesson learned for another day.

But wait, what’s that?

Those familiar red and blue lights, approaching with some urgency from behind. You know the routine: pull over, engine off, dome light on, and hands on the wheel. Some things never change.

The officer approaches your window and asks your name. As soon as you respond, you can just make out a tiny green light as it briefly flashes behind her sunglasses. “Is this your vehicle?” “Is it insured?” “Where are you going?” Each time a little green flash after your answer.

But then, the formalities are over. “Do you know why I stopped you?” You hesitate, unsure if she actually saw the infraction. An orange glow starts behind the glasses and she slowly reaches up to squeeze a button on the brim of her hat, never breaking eye contact. “Did you run a red light?” A small chime sounds and she moves on to the next question before you can answer. And the next. And the next. Each more specific than the last.

Then with a bit of whiplash, she thanks you for your time and is gone, back in her car and driving off. As the patrol car disappears around the corner your bank notifies you of a $215 transfer to the local authorities: THANK YOU FOR YOUR COOPERATION.

Humankind has been able to read minds since 1875. Only rabbit, monkey, and dog brains to start with but then in 1924 the first human mind reading was performed with the invention of the electroencephalograph or EEG. You’ve probably seen one in movies: a hairnet full of wires glued to your scalp with an umbilical cord attached to a briefcase-sized device. As it operates, the electrical signals from your brain are recorded onto a roll of paper.

So can we deduce the thoughts of the person using an EEG? No. Not really. But wait, aren’t EEGs lie detectors? I thought so too but, no. Polygraph machines look similar but don’t read electrical signals from your brain. They instead measure your pulse, respiration rate, blood pressure, and perspiration. Telling lies should cause discomfort and a polygraph is designed to track that discomfort.

The fictional situation above is about withholding information, lie detection, and ultimately genuine mind reading all being accomplished without any physical contact. How much if this is already possible?

Digital cameras have advanced so rapidly in the past decade that completely new uses for them are being discovered. Reading someone’s pulse and respiration rate are two of them. Modern camera sensors can deliver such high fidelity images that it’s possible to accurately track fluctuations in a person’s skin color as their blood circulates. Impossible for the naked eye, these fluctuations previously went undetected. And while we may be able to see someone’s silhouette change as they breathe in or out, it becomes impossible if they’re moving. Not so for a computer. It can estimate someone’s pose and chest shape many times per second. Combine all of this with a forward-looking infrared camera (FLIR) and you get skin temperature as well.

We’re mostly there for our lie detecting Judge Dredd scenario. What about true mind reading? Within the last year, there have been two major breakthroughs in this area. When they were announced, the artificial intelligence and neuroscience communities collectively freaked out. Let’s see if there’s reason to.

In late 2022, a research group from Osaka, Japan published a paper that described how they had reconstructed “visual experiences” from human brain activity. Participants in their study looked at images while hooked up to an fMRI (functional magnetic resonance imaging) scanner. These scanners work by tracking blood flow changes in the brain. Having more blood in an area means it has higher activity than neighboring areas. The scanner data was then fed back through an artificial intelligence model they had developed which generates an image trying to guess what the person saw.

Their results are astounding. The original vs recreated comparisons are jaw-droppingly accurate. And these aren’t the typical Zener cards used in telepathy experiments (think black and white shapes and wavy lines from that famous Bill Murray scene in Ghostbusters). No, these are color photographs of real things like a posed teddy bear, an airplane taking off, a snowboarder doing a trick, etc.

Human views image, we scan their brain, we ask an AI what the human saw, it generates a color image. Incredible.

The other Earth shaker happened in May of this year. In a process similar to the Osaka visual group, a team in Austin, Texas developed a method to turn brain scan data into speech. And unlike the Japanese imaging group, it wasn’t limited to a single word or sentence that the participant had heard. It managed to create continuous speech for extended periods of time. In addition to recreating the words a person was hearing, it also worked on original imagined speech. When the subjects were recorded as they watched silent films, it even generated a logical narration for what was happening on screen. The research group believes they’ve stumbled on something deeper than language which is hidden in our brains.

A couple of gotchas and a reassurance. These most recent breakthroughs require an MRI machine which is room-sized and not very budget-friendly. Work is being pursued to achieve these results with more portable equipment such as fNIRS (functional near-infrared spectroscopy) but has so far been unsuccessful. It simply can’t measure deep enough into the brain. Also, they have only been tested on a limited population. Perhaps their results are not as effective as we think. Oh and in case you’re worried: participation is always required. Attempting to read a subject’s mind against their will failed spectacularly.

Now Webster defines telepathy as “communication from one mind to another by extrasensory means.” That doesn’t necessarily mean paranormal woo woo charlatans. We may not be able to directly sense someone else’s brainwaves but they’re there and we’re starting to accurately interpret what they mean. As Arthur C. Clarke so famously wrote: “Any sufficiently advanced technology is indistinguishable from magic.”

Imagine a slightly different scenario than our opener: you running the red light has caused an accident and the victim is on their way to the hospital aware but unable to speak. The EMT can ask “Can you hear me?” “Do you want me to call someone?” “That’s your mom?” “What is her number?” The victim can communicate. They can be reassured.

Extend that for all neurodegenerative diseases. All stroke victims. Anyone with speech difficulties. The medical applications alone convince me of the value of this kind of research. The future is not inherently dystopian or authoritarian. To me, it will be magical.

Vol. 1, No. 1 — The future is going to be extremely weird and will arrive way before we’re ready for it.

This being the opening line of a new set of columns, I wanted to punch it up a bit with some slick synonyms. For “weird” you find things like “unnatural” and “eerie.” You could also pick “spooky” or “uncanny” but I don’t think they’re accurate. It’s just going to be really really weird.

Example from the present: it’s probably weird to someone born in the late 20th century (umm…1982) how we listen to music now. Back in the day, the most direct route to hear your favorite song might be to call Magic 93 and convince John Ford to play it or hope Mr. Kasem already had it on his countdown. If you were really on top of your game, you’d then stand by and hover over the play/rec buttons on your boombox to save a copy for yourself.

Not even 150 years before that the ability to record sound hadn’t been invented. Period. Let alone have that technology available to everyone at a reasonable price. Listen to the music live, make it yourself, or nothing.

Now what can we do? In mid-breakfast-mouthful, we can grunt out a song name, and a speaker in our room will play it in spectacular fidelity. Any song you can think of. Nearly the entirety of all music, ever, as fast as you can say “Hey, ding-dong, play The Safety Dance.” That’s weird!

No? Not weird for you? You’re young and hip? How about if you didn’t have to name the song? What if you could just hum it silently to yourself and your stereo would jump in? Or while brushing your teeth say “Hey, ding-dong, play that song I heard in my dream last night.”

You may be wondering about the title: Singularity Dispatch. The Technological Singularity is an oft-debated idea in computer science and science fiction. There are several versions and interpretations out there but basically, it boils down to what happens when our technology becomes so advanced that we’re not able to comprehend it? And by “we” I don’t mean the users or even the inventors. What happens when we, as a species, become completely outpaced by our own creations? What happens when our technology is capable of improving its own design?

To me, this hypothetical “intelligence explosion” event feels like “the internet” did in the mid-’90s. Back then, I was a teenager pouring over a book by Bill Gates called “The Road Ahead.” It attempted to explain various technological advances in relatable ways. What is digital vs binary? What can using e-mail change for us? Is it better than a fax? What is online shopping? These things were not clear at the time. It was all brand new.

My aim with this series of columns is to do the same for our freaky, inevitable future. How good is Artificial Intelligence (AI) getting? Which jobs may or may not be going away? How many of all these “advances” are truly game-changing and how many are snake oil? (The media plays pretty fast and loose with “AI Armageddon” headlines.) And what external factors are we missing? Remember the recent (and continued) global chip shortage? It's pretty hard for the disembodied AI voices to take over the world if their silly human day laborers can’t make enough chips or generate enough electricity to keep their brains on.

Also, why does this all seem to be happening so suddenly? An unrelated set of technological improvements has come together in serendipitous ways resulting in a drastic adjustment of timescales. Estimates for certain capabilities have been shortened by decades and then shortened again. We’re already doing things with AI, robotics, energy production, and space travel which were considered pure science fiction a mere 20 years ago.

It’s not the first time this has happened. In 1903, the New York Times wrote “the flying machine might be evolved…in one million to ten million years.” Less than three months later, the Wright Brothers had their first manned flight. In 1910, the opinion was that flight would remain only for the rich and never be faster than a train or car. By the 1930s, commercial airlines with hundreds of thousands of passengers were up and running. Then we orbited the Earth in 1961, landed on the moon in 1969, and three years later it became so mundane that people stopped tuning in to watch it happen.

It goes without saying, I’m no Bill Gates. But I will attempt to bring this high-flying techno hype train back down to Earth. One prediction for the singularity places it before 2045, let’s see how far along that road we are. If not for you, then for me. I mean, how else will I know when to expect my jetpack and flying car? The Jetsons promised us those, you know. Long, long ago!

So what is my first topic? I briefly mentioned what sounds like telepathy a few paragraphs back. That’s what we’re going to look at next week: How close are we to actually reading someone’s mind? And since we’re likely using a computer to help us read that person’s mind: how long until they can do it on their own?