Saturday, August 12, 2017

What Is Post-Intelligent Design

Recently, New Atlas posted this article about machine-optimized design engineering. I immediately recognized it as a manifestation of what Daniel Dennett refers to in this video (and many others) as Post-Intelligent Design.
That's a load bearing engine block, optimized using a generative design algoritm
Engine Block designed using generative algorithms (New Atlas)

Intelligence-Free design consists of natural processes such as evolution and natural selection which result in incredibly complex and highly optimized solutions. Bird bones for example are highly optimized for strength-to-weight, elasticity, and flexure. The process of natural selection is demonstrably purposeless and non-goal-oriented.

Intelligent Design is the deliberate arrangement of components or materials to achieve a specific set of performance goals. It typically results in highly simplistic, geometric regular forms, owing in part to the necessity of making things easy to produce, and in part to only simple forms being amenable to manual analytical methods. Cars and bridges represent intelligent design processes.

Image result for bird bone section
Cross-Section of a bird bone
Post-Intelligent Design (PID) means that the engineer allows unintelligent computer simulations to recursively modify a design to achieve optimal performance, typically the lowest weight or size that achieves a specified strength. It can also refer to self-learning computer algorithms or computer systems that develop their own optimized code or networks. In many ways this is Intelligence-Free Design revisited, but now with purpose and goals which we give it.

Intelligent design blends smoothly into PID; there is no crisp transitional line. For instance when I design something really critical, I use analytic techniques ranging from paper-and-pencil, formulas & a calculator, right up to 3D FEA on a computer, in order to identify areas that are critically stressed, and areas that are under-stressed. An optimal design under the maximum design load case should be uniformly stressed, indicting that every bit of material is fully contributing to the device's function. I will then remove material from under-stressed areas and add material to over-stressed areas, and do the analysis again. I may even make new material selections (carbon fibre, high-performance alloys etc) to get the required result.  This cycle repeats for as much time as I have or until the goals are achieved within a tolerated margin.  Post-intelligent Design simply automates this process and gives the computer wider latitude to come up with optimal designs that meet the given constraints and performance goals.

One of the enabling technologies for PID is 3D Printing. This removes one of the constraints faced by Intelligent Design: the need to make designs in simple geometric figures that are able to be produced in real life.

The final obstacle to overcome is that 3D-Printable materials are not the highest-performing materials that we have. The relatively poor specific performance of thermoplastics, sintered metals etc is a major problem that makes generative designs that can only be 3D printed actually less useful than intelligent design using simple geometric shapes.  Naturally, a lot of smart people are working on exactly that problem.

In a PID world, it may be that no one person or even group of people knows how a piece of technology works, what certain features are for, or why something looks the way it does.  One only knows that it is optimized for a specified purpose. In such a world we become literally the god-like Minds whose values and desires guide and direct the autonomous evolution of a technological ecosystem.

Also in such a world, it will be possible, easy in fact, to deduce what those values are, since they will be reflected in numerous ways. In that world of our creation, it becomes incredibly important that we decide upon those values and ensure that they are objectively good - by which I mean supportive of our long-term survival and quality of life.

In the Intelligence-Free natural world, there is absolutely no evidence of any sort of guiding values at work. One can see this in humanity's past.

The Intelligence-Free universe gave homo sapiens lives that were, to quote Hobbes, "solitary, poor, nasty, brutish, and short." Our happiness and comfort or the "sanctity " of life obviously are not values held by the universe. But at some point around 10,000 years ago, give or take, we developed culturally-transmitted and ever-evolving thinking tools (Dawkins' "Memes") and immediately began Intelligently Designing our lives and our world. Human population then exploded.

A few hundred years ago, those thinking tools underwent another growth spurt, and Science was born - the meme that suggests that the most powerful and efficient way to determine fact from fiction was to try to disconfirm the hypothesis. Not coincidentally, in that short time not only has humanity completely overrun the planet but individual lives are now twice as long with vastly more interesting things to do and marred by dramatically less suffering.  We are no longer nasty, brutish, or short.  Well, most of us anyway.

We did that.  We did that ourselves.  We did that by Intelligent Design.  And now, it may be time for Post-Intelligent Design to step in and take over.  But what will our new job be?  We get a promotion.  We become the ones who decide what is important and what isn't.  We need to start taking that job seriously.

No comments:

Post a Comment