Generative Design

Sometimes the whole people and technology debate can seem somewhat polarising, particularly when they're represented as mutually exclusive opposing forces when of-course the most interesting scenarios to work through are how one can augment the other. IA rather than AI. Intelligence Augmentation rather than Artificial Intelligence.

With machines beginning to demonstrate not only an increasing ability to think for themselves but greater creativity and unpredictability I think there is a more opportunity for reinforcing learning loops to be created not only in machine to machine scenarios, but also machine to human.

In a machine to machine situation, two programmes might be set against each other and each improve through taking it in turns to work through a problem, make decisions and set a new challenge for the other. As happened with AlphaGo, behaviour is learned over time based on a myriad of different contexts, but also perhaps increasingly from new or unpredictable situations or moves generated by the programmes which teach both of them strategies that they wouldn't have originated on their own through the need to adapt to new scenarios created by the other.

In a machine to human situation, an algorithm or software might work iteratively to suggest potential solutions using an evolving set of criteria set by a human, so that each challenges the other in a reinforcing loop. In a process that simulates evolution, this idea of 'generative design' means that software can 'evolve' novel component designs based on a changing set of parameters set by a human which can then be assessed, dropped or used.

Reading about Autodesk's use of generative design in their computer aided design software Dreamcatcher shows us just one way in which automation might augment the creative process:

'Dreamcatcher’s simulated evolution process begins with the software offering a set of potential solutions to a problem posed by the user. The software then “breeds” a new generation of solutions by recombining the best of those designs, as chosen by the user, and adding some random variation. The software can repeat this process to produce thousands of designs'

We're only at the start of-course of understanding how we might establish the kind of machine-human relationships that enable the most productive and challenging types of learning but there is something powerful in the idea of a continuous learning process where machines and humans each play their part in augmenting the capability of the other. The challenge of-course is how we shape the balance between human and algorithmic interaction to make sure we are reinforcing the right kind of learning. 

There's some who say that data and algorithms have no part in creativity. My own view is that there are some increasingly interesting opportunities for technology to inform the creative process and vice versa. But it's IA rather than AI. It's creativity augmentation rather than artificial creativity.