Search This Blog

Tuesday, November 14, 2017

Leaps in Artificial Intelligence

One of the holy grails of Artificial Intelligence research has been to understand how the human brain works. The idea is that if we knew how the brain works we could simulate its processing using computers and those computers would then be intelligent. Alas, remarkably little is known about the brain.

In the last few years, AI researchers have tried a different, but related approach. They've simulated neural network topologies in a computer that are sorta based on neural connectivity in a brain. The idea is that even though nobody knows how those neurons in the brain work when connected like that, perhaps those topologies will do something useful anyway.

And much to my utter amazement, that approach has made some really jaw dropping (well, my jaw, anyway) breakthroughs in a wide range of areas from vision to self-driving cars to emerging intelligence to self-directed learning and more. There's not time and space for me to get into all of these areas, but I'll touch on a couple.

The first is image recognition. A huge goal of AI has been to be able to have a computer look at an image and tell you what's in the image (for example, a car or sword or shark or poppy or fighter-jet or ...). And since a human can distinguish between hundreds of thousands of different types of objects, wouldn't it be nice if the computer could also distinguish between that many different things as well.

As of 2010, that level of ability for image recognition by computers was a pipe dream, and nothing more. As of today, a mere 7 years later, computers are now really good at that. Not quite as good as humans, but rapidly closing in as shown by the following graph:



Some background is required for the graph above. In 2006, some researchers got the idea to create a database of 14,000,000+ images (they hope to have 100 million images eventually) of tens of thousands of different objects, each image labeled with the object(s) it contains and bounding boxes of each object. With this database, neural nets can be trained to recognize the objects. Then, when shown an arbitrary image, the neural net will identify the objects in it.

The database is called ImageNet and was first ready for use in 2010. A contest was created to see who, if anybody, could create and train neural nets to distinguish between the tens of thousands of objects in the database. In 2010, the results were dismal with most contestants guessing right less than half the time. But, by trial-and-error and building on the best successes (evolution?), each year the results got better - a lot better. To the point where if you show one of the better nets a picture (and the picture is reasonably clear and a few other minor caveats), it will correctly identify the main object(s) (again, out of tens of thousands of possible different objects) in the image the vast majority of the time. And anyone can download these trained nets and utilize them with open source software such as Google's TensorFlow. While it takes weeks and weeks of cloud computing to train these networks using 14,000,000 images, once trained, a typical desktop can recognize the objects in an image in a few tens of seconds and in less than a second if it has a sufficiently powerful GPU (it turns out that graphics cards happen to be nearly exactly optimal for processing neural nets).

These image recognition nets are called "deep learning convolutional nets" and nobody really knows how they work, only that they do. Sorta like how we don't know how the human brain works - only that it does. Some modifications of these nets has enabled a lot of different applications to be addressed. For example, a while back, an AI beat the worlds Go champion. Ho hum, chess had already fallen to computers, so not a big deal, right? But it got a little more interesting a few weeks ago:
A new paper published in Nature today describes how the artificially intelligent system that defeated Go grandmaster Lee Sedol in 2016 got its digital ass kicked by a new-and-improved version of itself. And it didn’t just lose by a little—it couldn’t even muster a single win after playing a hundred games. Incredibly, it took AlphaGo Zero (AGZ) just three days to train itself from scratch and acquire literally thousands of years of human Go knowledge simply by playing itself.
Self-learning artificial intelligence. Pretty nifty.

Many of these techniques (and many more) are used in self-driving cars. They will soon teach themselves to drive really well - "literally thousands of years of human" driving experience. Bigger nets will be able to incorporate millions of years of human driving experience. It may take years to train them, but once trained, they can be downloaded to all cars. Humans may bested by AI in wide range of applications in my children's lifetimes, not just relatively trivial things like chess and Go (which only 10 years ago were not considered at all trivial).

I'll leave you with what I think is a very interesting video. I'm sure you've all seen faces morph from one person to another, but I think you'll find that the morphing is qualitatively different starting at the 1:50 mark on the video. All of those faces are simulated by the neural net which has been trained to "know" what a face is. The morphing from one face to another, even radically different faces in different poses, tends to stay pretty realistic throughout the transition. And the scene morphing, also completely simulated, maintains a surprisingly realistic rendition even when changing between radically different scenes, for example the bedrooms just after 4:00. Enjoy!
More information on the video is here.

56 comments:

erp said...

I don't even want a self-driving washing machine, never mind a self-driving car. Why do the younger generations want to give over control of their lives to big brother and/or a machine version of same?

Don't get it.

Clovis e Adri said...

Wow, Bret, you just made me feel like a fossil. Thanks for the heads up, I was not aware of the recent progress on image recognition - even less of image rendering like that.

We just can't ever assume again that a face you see in the internet belongs to anyone who ever actually existed (thought that was true before, it is now in a massive level).

Once they develop that to 3D rendering, we'll finally achieve that promised level of virtual reality where you can't differentiate what's real anymore.


Are you employing any of those techniques to your own robots?

Bret said...

erp wrote: "I don't even want a self-driving washing machine..."

Ummm, your washing machine doesn't operate automatically? Or are you still using a washboard? :-)

erp wrote: "Why do the younger generations want to give over control of their lives to big brother and/or a machine version of same?"

Because during those hours per day I'm stuck in traffic, I'd rather be reading or singing or even sleeping. Perhaps you're right that it's dangerous to give up driving ourselves, but I'm a good enough hacker that I think I'll always be able to at least control where my car goes.

Bret said...

Clovis wrote: "I was not aware of the recent progress on image recognition..."

Well, embarrassingly, me neither, and it's my field. It totally snuck up on me.

I was aware that people were fiddling with these deep convolutional nets back around 2010, but there's always been way too much hype in AI and robotics, and they clearly didn't work then, so I just didn't keep track of it. Then, all of the sudden in the last two years, it went from not-ready-for-prime-time to working really well.

And we are now using these nets in our dodder weeder (recognizes dodder weed in a field of safflower and sprays it with an herbicide). As I mentioned, it's all open source and it was fairly trivial to retrain one of these nets (we used google's inception.v3 net) to recognize dodder and then to transfer this net to a GPU based single board computer to operate in real time.

So the bad news is that I missed an important development, the good news is that it's not too terribly hard to catch up, the bad news is pretty much anyone can do it so not much of a competitive advantage.

erp said...

Bret, I guess you don't do much laundry. New washing machines have a mind of their own and you're right I don't drive in traffic much, but I'd rather drive in traffic than have my car programmed by big brother.

erp said...

I'm so far out the loop, I have no idea why meshing faces into facsimiles of human physionamy is anything but a parlor trick.

Bret said...

erp, I do all of my own laundry. Who else do you suppose would do it for me? Even before being tossed aside by the soon-to-be-ex-wife, I did my own laundry.

Bret said...

erp wrote: I have no idea why meshing faces into facsimiles of human physionamy is anything but a parlor trick.

Because it's not "meshing" faces. It has learned what a face is and then creates faces based on that knowledge, much like an artist draws a face that he's created in his mind.

erp said...

Artists brings their feelings and point of view into their work. That's what makes it "art." When a machine, no matter how complex, does it, it's a neat trick. I'm sure I'm missing something very important. If it's not too much trouble, please explain.

Re: Laundry -- newer model appliances have much too much autonomy for me and some of them can't be retrained into doing it my way. A couple of years ago, when our 20+ year old washer and dryer needed to be replaced, I sent back two different ones and finally purchased a refurbished one from a very smart guy who does a land-office business selling old-fashioned appliances he's cleaned up and rebuilt to us geezers. I live in fear of my old dishwasher going and treat it very gently. I hear the new ones have more buzzers and flashing lights than the flight deck of the Star Ship Enterprise.

Peter said...

Bret, I need your and maybe Clovis's help. By chance I've been doing some reading on AI and some of the utopic and dystopic predictions about it. It's pretty clear that robotics, data retrieval, memory storage, and computational capabilities are increasing at mind-boggling exponential rates and that all sorts of miracles and horrors are being predicted. What intrigues me is this notion of the Kurswell "singularity", the point at which machines will become "more intelligent" than humans and AI becomes AGI. At this point, some say, we risk losing control and the machines will go rogue in some way with unforeseen and uncontrollable consequences. Basically I'm trying to figure out whether the notion is a serious one or the science fiction fantasy of overtaxed brilliant minds.

I think I get they aren't talking about machines developing benevolent or malevolent human qualities (like Hal) and they aren't just talking about bad people like Kim Jong Un or Dr. No getting control of them for nefarious purposes, which is not a new worry and scary enough without AI. Still less software glitches or incompetent programmers. The fear seems to be that their memory will be so extensive and their problem-solving capabilities so sophisticated and single-minded that they will take extreme and unforeseen "hyper-rational" measures to "solve the problem" unless they are programmed not to, but the programmers will lose the ability to foresee the scope of potential extreme measures they might take and guard against them all.

Accordingly, computers and robots may develop self-correcting capabilities that amount to self-preservation measures like thwarting all efforts to shut them down. They will "know" (by calculation and memory) what resources they need to achieve their objectives and so may start gobbling them uncontrollably, or "compute" that the best (i.e. most rational) way to protect the natural environment is to get rid of us. Their "creative" capabilities will be untethered to human values, of course, and may result in unforeseen solutions of which there will be too many for programmers to foresee. My favourite nightmare is if we program them to keep us safe and happy, they may lock us all in our bedrooms and stick electrodes in our ears to stimulate our brains' pleasure centers ceaselessly. No, I don't take that seriously, but it does seem to be true that the sophisticated trading algorithms that Wall Street relies on today can cause a serious market crash in microseconds by being too efficient in responding to unwise or panicked human orders to buy or sell.

Thoughts? Leaving aside the question of when all this may happen, are these fears realistic or just another exaggerated human worry about technology and change? Is these going to be some qualitative difference in tomorrow's computers that might leave humans behind and endanger us or will they just be more powerful extensions of what we have today? Put another way, is the video in your link evidence of something truly new or just the latest step in a continuum that began when the camera put all those portrait painters out of work?

Peter said...

erp:

I'm with you. The worst invention in all of history was the multi-functional remote controller. A TV salesman told me a few years ago that they were meeting resistance from seniors who just wanted to be able to turn the damn thing on and off and change channels and who were flummoxed to the point of rage by all the rest of the functions, which they would never use but which meant they would have to call the cable company and pay for a reset at the slightest glitch. The other example of technology-we-didn't-ask-for-and-don't-want is in banking. I had to open a new account recently and was amazed at the rigmarole. A nice banking employee guided me through the computer steps, which were supposed to be child's play even for dummies but which took him 45 frustrating minutes to get right. In a moment of uncommon honesty, he admitted that modern banking is all about their finding ways to never have to see you while at the same time trying to persuade you they lie awake at night worrying about your personal financial health. When I told my millennial kids about it in the hope I might spawn some kind of Luddite revolt in the name of human dignity, they just looked at me blankly and said, "Do you mean that in your day you couldn't pay your bills in your pajamas at 3:00 am?"

Bret said...

Peter,

Given that I totally missed the rapid deployment of these image recognition and related nets, I'm not feeling quite as confident in my prognostications as I have been in the past.

Nonetheless, I still think mankind's demise is most likely to be at the hands of man via future weaponry, probably pathogens or some technology we haven't even thought of yet. Smart robots are waaaaaaay down the list, in my opinion.

Secondly, on the rare chance it's even possible, I think we're really, really a long ways away from having rogue robots (created by some group of brilliant misanthropes) kill humanity or any significant portion of it. It's a really long gap from training a computer to drive a car really well to having that computer somehow jump the bounds of its task and decide to do something else, anything else; that it's somehow going to develop self-awareness or consciousness; that it's going to "decide" to begin solving "problems"; that it's going to act as a biological entity with all of the hormones and cellular interactions of biological entities; that it's going to become part of a culture and have what might be called "rationality" and "reason."

One of the interesting things about the sort of AI I discussed in the post, is that the trained nets have no rationality. They are not programmed. They are just stimulus-response systems. They make no moral judgments. They're very good (surprisingly good to me) at what they do, but they neither think nor are programmed. They "know" but they do not "think." They may sometimes seem humanlike, for example when they talk to you, but they're not. They're just very complex stimulus-response systems that "know" how to respond to anything you say in a way that's meaningful to you, but it's not meaningful to the system - it's just responding.

That's not to say that the military won't create very, very deadly stimulus-response systems, but even those simply won't move outside their bounds, at least not for a very, very long time.

Bret said...

erp wrote: "... it's a neat trick."

Is it also a "neat trick" when a blob of goop (i.e. human) talks? Simply a "neat trick" by evolution or God or whatever? I'd say yes.

Therefore, pretty much everything of interest qualifies as a "neat trick" and I agree that getting a computer to "know" what a face is, is indeed a very "neat trick." That and closely related technologies are quite useful too.

erp said...

What's to stop non-biological beings from doing what we do, i.e. looking around and seeing better, faster, more efficient ways of reaching a goal? If we puny humans are in way, no problema, we are expendable. Asimov got around it pretty well, in fact, he thought of robots as a tool, I concept with which I agreed when I first read it about age 10 and still think is the correct conception, but he didn't live long enough to fully understand how the world would change.

Clovis e Adri said...

Peter,

I am far less qualified than Bret to answer that one, but I will touch 2 points he did not:

1) There is a good reason for general definitions of life to go like "an organismic state characterized by capacity for metabolism, growth, reaction to stimuli, and reproduction". Things are alive when they can look around for its source of nutrients, breaking it up for energy and assembling the next generation that will keep doing it afterwards.

Maybe we will have a hypernet of things in future, where almost all our links in the chain of production will be linked by computers and robots in a grand 'self-driving' AI scheme. Then I can see a powerful AI-like 'being' taking over - otherwise, it can't check all the boxes to 'keep alive'.

Another possibility, which would be less of a takeover and more of a darwinism-on-gigasteroids, would be humans gradually meshing with the machine, by means of cyborg-like implants and so on. We would morph to another species (maybe many different ones), and instead of merely watching the AGI to take over, we would merge with it.

2) I am not quite sure, as many of the discussions on this topic assume, that such purported AGI would want to take over. Being a super-rational 'being', it could as well conclude it is all meaningless and shut itself down.

None of the suggestions above are original, but since you've asked...

Peter said...

Being a super-rational 'being', it could as well conclude it is all meaningless and shut itself down.

Clovis, I absolutely love that. Imagine, all those brilliant minds inventing super AI to save humanity and solve all its problems and the whole thing succumbs to existential despair. "Bob, maybe we should tell our programmers to write in some religion, or at least a good self-help book."

Fascinating stuff. Thanks, guys.

Hey Skipper said...

To the point where if you show one of the better nets a picture (and the picture is reasonably clear and a few other minor caveats), it will correctly identify the main object(s) (again, out of tens of thousands of possible different objects) in the image the vast majority of the time.

This sounds to me essentially no different than data retrieval. Machines are vastly better at storing, sorting, and finding data than people, but aren't particularly good at turning that data into information.

Let's say this net a picture of two adjustable crescent wrenches, identical in every regard save for two -- color, because the second is made of chocolate.

How good is even the best net at discerning the difference, never mind making sense of it?

Many of these techniques (and many more) are used in self-driving cars. They will soon teach themselves to drive really well - "literally thousands of years of human" driving experience.

I read somewhere in the last couple days that GM's work with self driving cars includes the vehicle having a 95% probable circular position error of 4 inches. (The article only said 4 inches, but all the navigation standards I'm familiar with use 95% CEP.)

Assume that car manufacturers achieve that standard, and that ultimately all vehicles will have that positional accuracy.

In that event, cars don't have to know how to drive on roads any more than packets need to know about the internet -- the two problems are essentially identical.

In that event, self-driving cars only have to recognize and avoid non-vehicle obstacles, a far smaller problem than learning to drive as humans know how to drive.

Hey Skipper said...

If all cars were similarly equipped and treated as packets on a network, there is no technical barrier to self-driving cars on limited access highways.

erp said...

Please, not before I leave the room.

Hey Skipper said...

Why? It could save a lot of lives from the grotesquery that is called driver training in the US.

Bret said...

Hey Skipper asks: "How good is even the best net at discerning the difference [between chocolate and non-chocolate crescent wrenches]..."

If it were shown correctly labeled chocolate and non-chocolate crescent wrenches, it would be able to distinguish between them quite well.

Pretty silly though, because if you showed me a picture of a chocolate crescent wrench, I rather doubt I'd be able to correctly identify it as chocolate (as opposed to a brown crescent wrench).

Hey Skipper wrote: "...self-driving cars only have to recognize and avoid non-vehicle obstacles, a far smaller problem than learning to drive as humans know how to drive."

Sure. But they will nonetheless be able to drive autonomously just like (actually better than) humans. In which case, they won't need to rely on any sort of centralized data bases.

Peter said...

Does the advent of self-driving cars that drive better than humans imply that it will eventually be illegal for humans to drive?

Bret said...

I don't know. It may mean that if you choose to be able to drive your car, you'll have much higher insurance premiums.

But I was thinking it was probably inevitable that we'd eventually lose the right to drive anyway. After guns are confiscated, the deaths due to driving are really high and the same logic would indicate taking away the right to drive and only having mass transit and professional drivers.

Hey Skipper said...

[Peter:] Does the advent of self-driving cars that drive better than humans imply that it will eventually be illegal for humans to drive?

My bet is yes.

Ignoring exogenous factors, there are absolutely no technical barriers keeping us from fully autonomous vehicles. The problem is indistinguishable from internet packet routing. So long as all the vehicles are known and controllable, then road space can be used with far (as in probably 1000 times) greater efficiency. There would be no need for traffic lights or stop signs, since the position of each vehicle can be managed to prevent conflict with any other vehicle.

That means human driven cars would be an unacceptable variable.

IMHO, the first step will be enforced autonomous operation on limited access highways, then primary roads, and perhaps never tertiary roads.

We can't ignore the exogenous factors -- hapless pedestrians, dogs, fallen trees, whatever. It is relatively cheap to place sensors along limited access highways, more expensive on primary roads, and perhaps prohibitively expensive on tertiary roads.

So humans would drive the first and last mile, while leaving the rest up to the routing system.

None of this requires adding any sensors other than positional to vehicles, the means to communicate parameters to the router, and the ability of the router to control each vehicle.

Peter said...

Yes, I can't see human driving surviving. The safety standard will be set by self-driven cars and if humans can't meet it, sayonara. There would be fierce lobbying by the last bereaved parents saying tearfully "If we can save just one life..." How could any politician ignore that?

Talk about a cultural wrenching. Think of all those ads that persuaded three generations of males that the highest expression of personal freedom is expressed by hitting the high road with mountains in the distance. And how are our great grandsons going to lure the babes?

erp said...

Peter, your great grandsons won't be interested in luring babes. Sex robots are already available and the mind boggles at what they'll be like in another generation or two.

To be unbelievably vulgar, the human race won't end with a bang, but with a whimper.

Bret said...

That's your best at "unbelievably vulgar"? You are so last millennium! :-) Funny though!

Bret said...

Peter asks: "And how are our great grandsons going to lure the babes?"

Since I have daughters, I had to read to them the ENTIRE "Little House on the ..." series by Laura Ingalls Wilder from the pioneer days. In one of the last books, she's being courted by Alonzo(?) Wilder and she's attracted to him because he has a really hot pair of horses pulling his buggy. It cracked me up because even back then, it was the guy with the hot wheels who got the babe!

So yeah, I have no idea how future generations of guys will woo women.

Peter said...

I think you are right, erp, seriously. I see tomorrow's man as publically proclaiming his total solidarity with gender equality and the latest feminist agenda while privately making creative use of Bret's amazing nets.

erp said...

Peter: Yeah, so boys will continue to be boys, but what will us poor girls do.

Bret: I believe I would have been completely comfortable during the Victorian era.

Bret said...

erp wrote: "...but what will us poor girls do. "

Lesbians should do quite well!

Also, no reason there can't be robots for girls too!

Hey Skipper said...

[Hey Skipper:] So long as all the vehicles are known and controllable, then road space can be used with far (as in probably 1000 times) greater efficiency. There would be no need for traffic lights or stop signs, since the position of each vehicle can be managed to prevent conflict with any other vehicle. 

In Europe, trains are a big deal. They go nearly everywhere, are reliable and clean, and are cost competitive with driving (never mind parking).

Of course, a major factor in cost competitiveness is the extortionate price of fuel here.

But leave that aside. Fast forward 20-ish years. All vehicles are fully road-net compatible. No one needs to own a car anymore; essentially the entire car economy has morphed into Avis vs. Hertz.

As adequate as trains are here — it only takes me about an hour to make a forty minute trip to the airport — why would anyone take a train instead of stepping into a car that will take them point-to-point? Particularly when fully fledged road net management will make traffic jams a thing of the unmourned past.

It isn't just the auto industry that's in for radical transformation.

Hey Skipper said...

[Peter:] Talk about a cultural wrenching …

I see what you did there.

At the risk of social ostracism, XY's are perhaps more mechanically inclined than XX's. One of the ways boy people cultivate that tendency is through car maintenance.

My son can do just about anything to a car save an engine or transmission rebuild. That was fairly common when I was his age. Anymore? Not nearly so much. Along with being able to operate a manual transmission properly.

Not being a girl person, I can't say for sure. But I rather suspect that one of the qualities that enhance a boy person's attractiveness is the ability to deal with that fiddly stuff. As the constructive qualities are no longer needed, what's left?

[erp:] Peter, your great grandsons won't be interested in luring babes. Sex robots are already available and the mind boggles at what they'll be like in another generation or two.


And to think Bret scoffs at my suggestion humanity will self-extinctify within several hundred years.

erp wrote: "...but what will us poor girls do. "

[Bret:] Lesbians should do quite well.


I self identify as a lesbian trapped in a man's body. The world's my oyster.

erp said...

Skipper, you're right us girls do like those persons unlike ourselves who do heavy lifting and fix things whatever you call yourselves. I heard the reason people can no longer tinker with cars is that they are computerized and can't be tinkered with without expensive equipment that can talk to their computers.

The world is so different than it was even in my lifetime, yet most of us adjusted to it. Same will be true of whatever the future brings.

NB - I just read really young kids aren't as enamored of devices as are their elders in junior high, so get ready for what's coming next - playing outside in the sunshine maybe?

Ya think :-)

erp said...

Bret, re: captcha

Now it pops up after I've hit Publish Your Comment, instead of before?

Hey Skipper said...

[erp:] I heard the reason people can no longer tinker with cars is that they are computerized and can't be tinkered with without expensive equipment that can talk to their computers.

That's what people say, but it is bogus.

First off, cars are so much better engineered and put together now that getting to things is far easier now. I'd far rather work on our 11 year old car than the now 25 year old car I gave our son.

Then there is the computer part of it. For very little money -- $50 or so -- one can buy a digital volt/ohm meter to check continuity of wiring and operability of sensors. If they check out, then go to an auto reclamation yard and get the computers at pennies on the dealer dollar. Yes, that reeks of throwing parts at the problem until it is solved, but it works. (Dealer cost to fix an ABS light on our old car? $1500. Throwing an ABS computer from a junk yard -- $90. Time to swap, 20 minutes.)

Or, you can do what I finally did a few months ago. Spend $250 for a decent code reader. Had I done it one maintenance event sooner on our newer car, it would have paid for itself out of the box.

Men have to be useful to women. If technology wipes that out, what's left?

Hey Skipper said...

[erp:] Bret, re: captcha

Now it pops up after I've hit Publish Your Comment, instead of before?


I had that problem for awhile.

As it turned out, it was because I was checking the "I'm not a robot box". At the time, if I wasn't a robot, I was subject to all kinds of Captcha harassment. But once I copped to possibly being a robot, by leaving the box unchecked, all was good.

erp said...

Skipper, you told me that trick a while back and I've following that advice. I also deleted cookies and cache ... still happening and I'm cancelling my subscription to the NYT crossword puzzles because they can't fix it so I can login and stay logged in. The end of 70 years of Sunday puzzles. Yes, I am that stubborn and have cut off my nose to spite my face rather tell Captcha which squares have road signs every time I want to login to my paid subscription. Had a few go rounds with snippy NYT techs. They aren't from Bangladesh, but no doubt, scions of the snippy editors there.

As for cars, we just inherited a 2008 S350 Mercedes from our 20 YO granddaughter who was trading up. Yes, that is how it is in today's world and were told that our mechanic couldn't work on it without special equipment.

It ain't so apparently.

... as for what's left. If you're daddy didn't tell you about it, I sure ain't. More unspeakable vulgarity from the old lady. Oh the shame of it.

erp said...

... that's "your" daddy -- I have to learn to proof read before, not after, I hit send.

Bret said...

Hey Skipper wrote: "And to think Bret scoffs at my suggestion humanity will self-extinctify within several hundred years."

Yep, I'm still scoffing away. Not everybody will prefer sex robots or even if everybody does, some will still want children and lots of 'em and that genetic phenotype will inherit the earth.

Bret said...

re: captcha

I have no idea. Sorry!

erp said...

... children no longer require both parents to ever be in the same bldg even now. Why project today's ideas onto to tomorrow.

Hey Skipper said...

[Bret:] Yep, I'm still scoffing away. Not everybody will prefer sex robots or even if everybody does, some will still want children and lots of 'em and that genetic phenotype will inherit the earth.

So let me get this straight: yet another impediment to having children, and therefore more children?

I think you are missing a step or two.

Oh, and I read recently that the US TLF has hit a new low of 1.77 or 1.72. Can't remember which.

I think most demographers don't see that as the road to a population 'splosion.

Bret said...

Hey Skipper wrote: "I think you are missing a step or two."

You're the one missing the step. Which is that those who don't wish to have children die out leaving those who do.

Hey Skipper said...

You are assuming at least one fact not in evidence.

Clovis e Adri said...

Skipper,

---
Men have to be useful to women. If technology wipes that out, what's left?
---

A good question, and a small part of the bigger picture, which is that humans have to be useful to other humans. If technology wipes that out, what's left?

I am pretty sure technology is wiping that out, and we're going, soon or later, to reckon with the "what's left" part. Nowaydas, I am not optimistic at all about that.

Bret said...

Hey Skipper wrote: "You are assuming at least one fact not in evidence."

As are you.

Hey Skipper said...

Of course, there's a way to put at least part of that right.

Japan has a TLF of something like 1.2. Of course, some of those women are having no children. But of those that are, how many are they having? If 10% of women are having 12 children, that is a powerful argument in your direction.

Conversely, if almost no women are having more than two, then even if your putative fertility gene exists, it would be so rare as to be useless in stopping a terminal population crash.

Bret said...

Hey Skipper wrote: "...it would be so rare as to be useless in stopping a terminal population crash."

Untrue. As I've shown in my simulations.

Hey Skipper said...

Except your simulation relies upon a fact that not only isn't in evidence, it contradicts evolutionary theory.

After all, for your fertility gene to exist, somehow evolution not only selected for it, but also select for a variant that would select itself out of the population.

So it seems only two options are left. Either there is there is a gene that influences how many children women want to have (not whether women want a child in the first place), yet somehow variants of this gene that select themselves out of the population not only still exist, but are predominant.

Or -- until the advent of cheap, easy, and reliable birth control -- this gene, which had absolutely no effect on reproductive fitness existed nonetheless. Even though it couldn't.

Bret said...

Hey Skipper wrote: "... it contradicts evolutionary theory."

No, it doesn't. The basis of evolutionary theory is "Life Will Find A Way."

Hey Skipper said...

Huh? Wait. What?

Taking as read "Life Will Find a Way" doesn't mean that All Life Will Always Find a Way All the Time.

Nor does it cancel genetics. Presuming a gene for which there is no selection value existed, if there is no selection value, it cannot exist for very long.

Because the most basic basis of evolutionary theory is "Mutations".

Bret said...

Hey Skipper,

I took the distribution of number of children per women that exists in the population today and modeled it. Yes, those are phenotypes, not genotypes, but it hardly matters: memes vs. genes. You think Orthodox Jews in NYC who currently have 6 children per women are suddenly all going to engage with sex robots? You are the one assuming facts not in evidence, not me.

Hey Skipper said...

Bret, is the number of children a woman has solely the product of genotype, or might culture have something to do with it?

Mormon birth rates are sinking like a greased safe. Did their genes suddenly change?

Orthodox Jews are the most self confined religious group in the US, by a long shot. So the question you need to be asking is whether that confinement has saved them from the extinction gene rampant elsewhere, or whether women's maternal instinct is far more satiated with the first child, then any that follow.

erp said...

... a better question might be will there be any Orthodox Jews left in NYC or anywhere else after the robots take over.

Hey Skipper said...

[Clovis:] A good question, and a small part of the bigger picture, which is that humans have to be useful to other humans. If technology wipes that out, what's left?

I am pretty sure technology is wiping that out, and we're going, soon or later, to reckon with the "what's left" part. Nowdays, I am not optimistic at all about that.


Neither am I.

It seems to me that much of human behavior -- at the statistical level -- can be explained by game theory. Over evolutionary time, what became humans derived from their more simian ancestors, and in so doing had infants with increasingly large skulls, which could compromise female mobility and survival only so far, which drove longer and longer periods of infant vulnerability and maternal dependancy.

That could go on for another para or so, but the point should be clear in this specific case: females traded access to them for material support. That's a heartless way of looking at it, and all manner of fairy tales paper over it, but universal female hypergamy gives the game away.

Similarly, males ceded other activities to females.

However, relentless modernism has rendered men, in particular, less necessary. Where machines won't suffice, the state can step in. Life is less risky, physically onerous tasks less necessary, and those that remain more amenable to machines.

If what men can provide is progressively less necessary, then why should they put up with men?

That goes both ways.