Global Guerrillas

Owners Are Now Training Their Cars to Drive Autonomously

Tesla's autopilot went live a couple of weeks ago (it's one of the first car brands to do this).

Unlike the autopilots and cruise controls of the past, it's an autonomous system.  This means it isn't limited to the capabilities you get when you pop it out of the box.  It gets better as you train it and provide it with experience.  

Tesla's AP

Here's some first hand feedback from Tesla drivers on how fast the autopilot is learning:

So far I have a little over 300 miles on autopilot, mostly 20 miles at a time on my commute to and from work.  The first day when I was in the right lane, as I approached exit ramps, it would dive for the exit ramp. I quickly learned to apply torque to the wheel to hold the car on the interstate until I had passed the exit.  Each day the system seems to have less tendency to follow the exit ramps as I pass. The last two days it only gave a momentary wiggle and moved over maybe six inches towards the exit ramp then it recovered and moved on down the road. This morning it gave only a very slight hesitation, so little that I did not have to correct it at all. I find it remarkable that it is improving this rapidly.  I wonder if it is getting more information on this section of road or if it is changing how it reacts to any exit ramp?
(John Robb: the answer is both..)

Here's another driver training the autopilot to navigate tight S turns:

I noticed that on sharply curved ramp connecting I-80 west with CA-113 north in Davis, the first time it took the curve at full speed and wasn't able to stay in lane resulting in a "take control immediately" alert.  After a few more times on this curve with firm pressure on the steering wheel it's now learned to slow down and today had no issue taking the curve. Definitely learning.

Here's another:

AP is definitely is learning. Each weekday morning I drive home on Route 50 in the Virginia countryside after dropping my kid at school. Watching the display while driving manually last week, I could see that AP lost lock on the road where Route 50 drops from a 4 lane divided highway to 2 lanes. Where the left lane merges into the right, the dashed line disappears for a while. Later on in the 2 lane section, a right turn lane comes off during a left hand curve in the road. I could see the projected course bend right and then snap back left: I had been using AP on the 4 lane stretch before the lane drop and was glad I had turned it off.  During the last couple days I have seen the projected course tracking correctly in both problem locations. The time of day, lighting and weather has been the same every time, and there were never any cars ahead to follow (I’m traveling against morning traffic). Then this morning I tried AP the whole length of my trip on 50 and it tracked perfectly.

These drivers aren't alone.  Already, Tesla drivers using this system are adding a ~ million miles of experience to it every day.

Here's a heads up 0n what this means...

Most of the systems we currently use aren't built to improve through use. They have locked in performance and capabilities. These systems can only improve through revisions and patches made by technical experts.

That approach is on the way out.

Systems can now be improved operationally ....

Further, for the most complex activities, this will be the only type of system you will be able to buy.

I can't emphasis enough how big a change this is.  

Note the Animation

 

Sincerely,

John Robb

PS:  This suggests that we will soon be categorizing everything in the military inventory into dumb systems and autonomous systems.  Systems that you improve through use -- from the weapon you carry to the weapon you fly to the weapon you work with -- and the systems with locked in performance. 

PSS:  Few car brands will survive the transition to electric vehicles.  Fewer still will survive the transition to autonomous vehicles (and the shift from car ownership to driving services made possible by autonomy).   I'm not convinced any existing US car brand will survive the next 20 years.  

DeepFace

Facebook has an autonomous facial recognition system called "DeepFace" 

It currently "knows" 800 million people.  

DeepFace is autonomous, it makes decisions based on cognitive capabilities it built on its own.

If I take a picture of you, and you are one of the people DeepFace already knows, it can identify you in five seconds.

It can do this at a level of accuracy as good as human beings ~97%.

DeepFace can run on the processor in a cell phone.

The data required to do this for ALL of the ~8 billion people in world can fit on one thumb drive.

Frankly, if there has ever been a photo of you taken (or you have been seen by CCTV), you will be known to these systems.

Journal: The Knife Intifada

"The terrorists, except for the cell that carried out the attack in Samaria, were apparently unorganized. All of them, it has been reported, had no security records, never spent time in an Israeli jail and were not affiliated with any armed Palestinian organization."

This is classic open source warfare... The new generation coming of age today is predisposed to self-organization.

______________

Australian Junk Guns:  "The weapons are made using instructions found ­online and include so-called pen and slam guns that use pipes to replicate a shotgun.  They account for about 10 per cent of all firearms being seized by police.  Firearms Squad commander Mick Plotecki said some pen guns sold for as ­little as $100 and were often as dangerous to the user as they were to other people,  “Most of them are very unsafe weapons, they can blow up in your hand,” he said.

>> Embarrassing level of innovation demonstrated here.  

 

 

Autonomous Weapons

Autonomous weapons are on the way.  They are going to radically change the way wars are fought >>

A weapon is autonomous if makes its own decisions.  

This is not as straightforward as it sounds. Although many of the weapons we use today are smart, they don't make their own decisions.  They are simply following the instructions of the human beings that made them. For example: when a precision guided munition (PGM) is fired at a target, it simply follows the complex instructions it was given (relentlessly, Terminator style) until it finds its target.

In contrast, an autonomous weapon develops its own way of solving problems. Currently, cognitive systems can accomplish this by mining big data sets, interacting with simulations and through real world exercise. This should sound familiar. Think for a moment about how you learned to throw and catch a ball.  You didn't need a formal education to learn this or knowledge of physics or math to do it. You learned to do it by doing it -- through experience. These new weapons can do the same thing. Like human beings, they can learn to do something new by doing it.

A weapon is autonomous if it makes its own decisions with cognitive capability that is built through experience (learning, training, experiments, exercise, etc.) .

Beyond Automation

There's been a big change in the wind for a decade.  Many of us have heard and felt it rumbling in the background, shaking the foundations of an already decrepit global socioeconomic system.  

Depending on how you view the world, you see it differently.  I see it as as technological transformation.  A technological transformation that will upend everything. 

Here's a taste of what is driving this change forward.  It's real and it's coming.  

 >>

Up until three hundred years ago, the world relied on the work being done by people, largely by hand.  The skills and methods required to do this work were largely inside the minds of the people doing that work.  We created organizations to aggregate the people needed for doing work on a large scale and guilds to protect this knowledge. 

To overcome the limits of a world made by hand, we developed something new:  automation.  We've transformed the world by building machines (in both hardware and software) that do work for us. Automation is based on a scientific process that figures out how things work and an engineering process that turns these scientific ideas into machines that actually do work.    

However, we've now reached the limits of automation.  How so?  Automation is limited by the ability of human beings to construct the cognitive models (both scientific and engineering) needed to build the machines that provide it.    

To overcome these limits, we're now building cognitive machines that can build their own models for how things work and how to accomplish tasks. Unlike the machines that provide us with automation, these machines aren't built in the traditional way and they can tackle problems far more complex than anything done by automation.

The big change is that these machines build themselves.  They bootstrap their abilities in the same way human beings do: through learning, training, and experience.   However, they can learn it MUCH faster (deep learning) than we do and once they do, they can share their new abilities with other machines all over the world instantly (cloud robotics).  

If you don't think this is a big deal, you are wrong.  It's the biggest shift in technology we've seen since the rise of automation over three hundred years ago and it's going to change everything.  In particular, it's going to upend the rules of economics, warfare, and politics we thought were immutable.