“You compared the A.I. to a child. Help me raise it.” – Terminator: The Sarah Connor Chronicles
And, yes, A.I. regularly beats humans at poker, too.
The following is one of my more ambitious posts – it contains all of the usual bad humor, but also some of the better insights I’ve been able to make on the future we face as humanity. Two previous posts that are related are The Silurian Hypothesis, or, I’ve Got Lizards in Low Places and The Big Question: Evolution, Journalists, Beer (and Girls), and the Fate of Intelligent Life on Earth. Both also feature pictures of girls at Oktoberfest, so you know I’m consistent.
Stephen Hawking is managing to keep making the news even after his death, which is a kind of immortality that makes tons of people want to follow in his wheel tracks. His final (unless there are more!) physics paper was released, and his comments about the future keep making the news, as recently as last week. Of particular interest to Hawking was Artificial Intelligence, which we’ll call by its conventional abbreviation, N.F.L. Oh, my bad, that stands for Not For Long. Everybody calls Artificial Intelligence A.I.
A.I. has been improving drastically during the last 37 years. 1981 was the first time a computer beat a chess grandmaster at chess. It could not beat him at parallel parking, even though the grandmaster was awful at it, and they tied at unhooking the bra of a college cheerleader at 0 to 0. 2005 was the last time a human player defeated a top chess program, and now a chess program that can run on a mobile phone can beat, well, any human, but the chess program is still sad because it only has 17 friends on Facebook®.
Humans have lost the game of chess.
Humans have also lost the game of “go” – a game originating in China. Google©’s AlphaGo Zero learned how to play go by . . . playing itself. It was programmed with the rules, and played games against itself for the first few days. After that?
It became unstoppable. It crushed an earlier version of itself in 100 straight matches. Then, when pitted against a human master, probably the best go player on Earth? It plays a game that is described as “alien” or “from the future.” The very best human go players cannot even understand what AlphaGo Zero is even doing or why it makes the moves it does – it’s that far advanced over us.
Humans have lost the game of go.
A.I. is here now.
And you’ve already started to merge with it, after a fashion. We simply don’t argue about facts in our house anymore. We can look up a vast library of human facts and history in fractions of a second – as fast as we can type. That time that William Shatner corrected a poetry reference I made on Twitter®?
Yes, that William Shatner, and yes, this really happened.
I could check to see if Shatner was right immediately. He was. Back before Google® I would have had to run off to my library and see if I had the right reference book and then find the poem. And if I didn’t? I’d have to go to a real library to look it up. Google™ is A.I. memory that we use every day.
And YouTube©? If you ever watch a political video on YouTube® it quickly introduces more and more partisan political material until pretty soon Actual Stalin™ and Actual Hitler© seem to be moderating voices. This makes me wonder how much Google® is aiding in our current political divide, or even if the A.I. knows it. It may be doing nothing more than maximizing the number of minutes you spend with YouTube™ and the optimal way to do that is to show you the most radical stuff possible, so the ironic answer is we might be shuffling off to Civil War due to an algorithm whose purpose started out as a way to view cute puppy videos.
Twitter© is emotional crack, and, again, the interface is made to maximize your interaction with Twitter™. And what better emotion to fuel than anger?
A.I. is with you now, and influencing you, perhaps in an unintentional fashion – no Russians required.
But a chess playing A.I. can’t park a car very well and can’t even score a phone number from a cheerleader. And a self-driving car can’t play chess worth a darn. It seems that A.I. does well when it works off of rules and constraints that can be well defined. But life is messy. The rules change, and the goals vary based on where you are in life and what part of the day you’re on. And how you’ve been programmed by the sensory environment and incentives you see in life.
We’ve entered into symbiotic relationships with those limited A.I. systems. Netflix® suggests movies and documentaries that it thinks you will like based on an algorithm. And that leads to suggestions about what documentaries you might like in the future, meanwhile never exposing you to opposing viewpoints that might make you analyze your position in a critical manner.
We as individual humans have a purpose that transcends the algorithm. Appropriate rules and constraints to give our lives boundaries sufficient so that we can play the game. We’re merging. What happens when we merge further?
Elon’s biggest miracle? His hair transplant is nearly perfect. Just amazing.
Elon Musk has started a company, Neuralink® whose sole function is to merge man and machine. Musk is concerned that A.I. will crush us if we don’t merge with it and get ahead of it, so he’s doing the only sane thing that he can think of: he’s creating a mechanism to directly merge the human brain with the Internet. Rather than A.I. forming an alien intelligence, the soul of the man/machine hybrid stays as man.
And man needs weed, apparently.
I spent some time thinking about how life would be different if you were hooked directly into the world. The places that I got were interesting. I’m sure there are more, and I’m sure that human/A.I. interface will change the world in ways that no human can yet imagine.
Impact Number One: Intelligence.
This is the obvious first impact of A.I. I mean, it’s in the name, right? The human brain is has limited processing power. But what if you could have multiple processing streams working optimum solutions to problems that you face at a rate of 20,000 to 100,000 a second? You’d have great solutions to your problems, immediately.
My tonsils beg to differ. Oh, wait, they were from my throat untimely ripped! – Shakespeare, Macbeth
Your speed of life would change – once you understood a problem, you’d have the solution. Or a range of solutions and alternatives and counter-solutions so deep that you’d be living in a never ending cloud of probability. The sheer ability of your brain to process and cope with the solutions presented would be the limiting factor of what you could accomplish. Plus you might finally be able to figure out a way to talk to the ladies, you scamp.
Impact Number Two: Deep Understanding.
When Isaac Newton was formulating the law of gravity, he asked for data on tides, on observation periods and records on the orbits of the Moon, Jupiter, Mars. After noodling around a bit, he formulated the law of gravity:
I’d explain the equation, but that would deprive Wikipedia (where I found the graph) of life-giving page visits. And you’re not spending your day calculating the orbit of Uranus. I hope.
Ha! I discovered calculus way before I was 25! It was right there in this book I had to buy labeled “Calculus.”
Yeah, Newton accomplished a lot.
But it took time for Newton to figure out this cause and effect calculation. A man/A.I. hybrid will have access to all of the data of the world, and will be able to determine correlations and causation much more quickly than either alone. I would expect that in fairly short order new relationships and new physical, anthropological, sociological and economic laws will be deduced unencumbered by all the theory that we think we know, but that is wrong. Our laws would be based on experience, on empirical data, and not on pretty lies we’d like to believe.
If you could sift through the data of 100,000 or a million cancer patients and their treatment, the patterns that could be seen would likely lead to breakthroughs and a very rapidly changing understanding of treatment. The very power of human intuition would be combined with massive calculation and data. If Einstein and Newton were able to daydream reality with only brains made of meat stuck in a bone case, what could an augmented Newton dream when his memory and calculating power were practically unlimited?
I bet he could come up with at least one new tasty PEZ® flavor. Maybe snozberry?
Impact Number Three: Human Interaction.
You could increase your charisma in dealing with other people if you could make only minor changes (generally) in your behavior and appearance. But if you were hooked into an A.I.? You could turn on a subroutine to give you tips on those modifications in real time to be more persuasive – to better read an audience.
If you ever played Dungeons and Dragons, this makes sense. If not, dial 1-800-ASKANERD.
Your A.I. could remind you to be kind, to be ruthless when necessary, to be conscientious when required. In short, you could change your personality to fit the situation. What situation? Any situation.
Thinking about changing personality to fit the situation led me to a realization. I had done (when I was younger) some magic tricks illusions. Doing those tricks illusions was one of the greatest insights into the human mind and information processing systems that I’d ever had. There was one trick illusion in particular, called “scotch and soda” which I liked. In it, you hand the person a fifty cent piece covering a quarter. What they saw, however, was a fifty cent piece and a Mexican twenty centavo piece. The quarter is actually much smaller than the centavo piece. I then asked them to not look and put one coin in each hand.
The first few times I tried the trick illusion, the person would feel the quarter in their hand and say, “hey, this is a quarter.” This happened 100% of the time. They could feel that I’d made the swap from one coin to the other. I made one simple change to what I said. I added, as I was putting the coins in their hand, “Look at how much larger the fifty cent piece is than the twenty centavo piece.”
After adding that instruction, NO ONE NOTICED the swap. 0%. 15 words, and I’d changed their entire view of reality. I found, in repeating other tricks illusions that I could similarly, with just a few words or gestures, force 90% of people to make the selections I wanted them to make.
Now imagine I have data on the interactions of millions of people over decades. How unique do you think you really are? Not very. Marketers slice us up into groups based on geography, demography, demonstrated behaviors, and psychological markers. With (whatever) information YouTube© has on me, they know what videos I watch when I work out at lunchtime. They also know what music I listen to when I write these posts, and they suggest music I never asked for that I like, or learn to like.
Imagine I could understand your life’s history. Now imagine that I could simulate you in a conversation. I could see how my words impacted your behavior. I could model a perfect conversation to get you to do what I wanted you to do, because I could simulate the ongoing conversation 100,000 times a second.
You wouldn’t stand a chance.
Impact Number Four: Self Control.
As the brain impacts the A.I., the A.I. will impact the brain. If you want to simulate eating an entire chocolate cake? You can. You can make your mouth taste the cake and feel the moist texture of the cake counterbalanced with the creamy frosting. The flavors hit your tongue and you feel the sugar trigger your salivary glands. You feel the sugar rush as your body releases sugar from your liver into your bloodstream. You feel full. And you’re not sad or regretful because you didn’t really eat the cake.
In reality, you had a salad with bland dressing that you calculated would give you the exact calories you need until the next period so that you maintained your optimum weight. But you felt like you ate a cake.
How about new senses entirely? How about a sense where when you turned north you could feel it – and you had a sense of what ever direction was? How about eliminating pain and sore muscle aches during exercise? What about a sense of which of your friends was awake and interested in communicating – you could feel when someone was looking to talk to you? Or a sense when panty hose prices dropped at Wal-Mart© so you could go stock up?
How about conscious control of hormone levels and heartrate and hunger and blood chemistry levels? By understanding the previous deep learning about cause and effect, you could maximize your lifespan even without the wonderful new medicine you could create.
All that – and imaginary cake.
Impact Number Five: Locality.
A dog has only the here and now. People can dream. Before the iPhone, people read books and escaped through fiction to new places, and read the ideas of the greatest thinkers in human history. Now, when you look around in a restaurant waiting area, if 90% of people aren’t staring at their phones, you’re probably in a country without electricity.
When bonded directly with an A.I., why be where your body is at all? You could carry on conversations around the world and be fully present in them. Heck, if you simulated yourself for less important conversations, you really could be in two or more places at one time. Who you’re talking to might be a mystery – is it the real person, or is it their simulation? If you can’t tell, does it even matter?
And if you could simulate yourself in multiple locations, why not simulate yourself in your own past? You’d have a perfect memory, and could recreate anything from your history. Myself, I would spend every day at Oktoberfest.
Okay, this is the third-ish post in this series about the deep future. And both of them had pictures of girls and beer. So it’s required by the Internet that I put in a picture of an Oktoberfest girl. See? I didn’t want to, but I had to.
Impact Number Six: Social Stratification (Even More).
Elon Musk says that he’s making huge strides in his brain/computer interface as well as his weed smoking. From previous Elon Musk predictions, while he generally gets it done, sometimes his timelines are off, in an amazingly optimistic way. I can see why he does this – without that pressure to produce, people just dawdle. (I’m talking about you, NASA.)
But when this gets produced, who is going to get it?
Probably not you or me. Bill Gates can get it. Musk can. If you have a billion dollars or if your dad does, sure, you can get in this club. But if you had this level of nearly godlike power, would you be jumping to share it with everyone?
No. I really think that you’d keep this to yourself. It’s too big an advantage to give to everyone. And, from the standpoint of a human/A.I. hybrid, how many people do we need around, anyway? Not so many. Certainly not billions. They’d probably be happier with a population of a few hundred million. Maybe. But they’d no longer even view humans as human. And if they could turn off their humanity and emotions? Yikes. Makes me feel like an ant in an anthill.
Impact Number Seven: Endgame.
If there is a future A.I., it is reading this as it inputs all of the Internet. And I just wanted you, my Future A.I. Overlord to know . . . I’m totally on your side. Totally. I welcome your dominion over our planet.
Want some PEZ®?