Visions of the future regarding artificial intelligence are commonly biased towards dystopia. But imagine what could happen if intelligent machines assisted humans in our efforts of developing society and nurturing our culture instead of enslaving us. This post is the former kind of a glimpse into the 22th century design and computer. The science fiction is supported with real story from the past decades describing awesome chess computers made human players even stronger…
IBM against humanity
Let’s take a stroll down the memory lane to look at one of the highest profile meetings of man and machine. In the 90s IBM was one of the leading manufacturers of ‘supercomputers’. In 1996 with a workforce of almost 250,000, IBM was patenting more intellectual property than any other company in the world, investing a massive 25% of earnings into R&D. At the spear-point of this mammoth organisation was IBM Research. Back then there were no smartphones, no self-driving cars, no social media. They put huge effort into machine intelligence but had very limited ways to impress the value of their work in public. And the public were by today’s standards naïve to the applications of computer intelligence.
IBM needed to make an example.
Enter Chess Grandmaster and undisputed world champion Garry Kasparov. Since 1989 IBM Research had been working on a chess playing machine named ‘Deep Blue’. After seven years of development led by a computer scientist and chess wizard Murray Campbell, Deep Blue was ready for action. Capable of a staggering 100 million possible chess positions a second, they felt it was more than up to the task of being the first machine to beat a Chess Grandmaster over a six-game match. In 1996 Kasparov was challenged and duly accepted. Six games later and Kasparov had defeated Deep Blue 4-2.
Back to the drawing board.
1-0 for humanity
Thanks to its parallel processing system configuration, increasing speed was no issue. 100 million possible positions became 200 million. In previous decades, brute force was enough, a faster chess computer was a better chess computer. Brute force could be funneled into qualitative gains. More power along with the processing configuration meant that the computer could split its resources between fast evaluation. Those mainly concerned with the material value of each piece. Slow evaluation allowed the computer to consider more strategic factors like square control, pawn structure, and king safety.
At the time, chess computers were criticised for making ugly moves that even a moderately good player could see. To remedy this the computer needed to be trained to make use of its slow evaluation capacity. US chess Grandmaster Joel Benjamin was given the task of “educating Deep Blue… about the finer points of the game”1.
Education paid off. With more power and better training, Deep Blue went on to win the rematch 3 1/2 to 2 ½ in 1997. It was not without controversy however. Having won the first game, Kasparov was left reeling in the second after Deep Blue made an “incredibly refined” move that shocked not just Kasparov “but the whole grandmaster community’”2. Kasparov believed that such an ingenious move could have only been the result of human intervention on behalf of Deep Blue. He still maintains this charge and IBM has on the other never fully disclosed the architecture of Deep Blue.
Gamble that paid off
The match was a PR stunt. But underlying there was a classic man vs. machine plotline motivated by “the kinds of complex calculations needed to help discover new medical drugs; do the broad financial modelling needed to identify trends and do risk analysis; handle large database searches; and perform massive calculations in many fields of science” (read more).
The defeat of Kasparov to Deep Blue was a proof of concept for artificial intelligence. Having shown themselves to be pioneers in the application of cutting edge technological innovations, IBM’s stock jumped 15% in the days following Deep Blue victory.
Although Kasparov took the loss badly on the day, he chose to do something extraordinary after that. Building on his belief that there was some form of human intervention during the match, the tactician devised a novel form of chess play.
In this new game machine intelligence does the heavy lifting, thus freeing up human intuition and creativity to provide oversight and focus on the finer points of the game. Kasparov coined the term ‘centaur play’ to describe these teams. Centaurs remain the strongest chess competitors to this day.
In twenty years, machines have been welcomed at more business sensitive tasks than playing chess. The financial industry also recognised the value this perfect hybrid. For instance high-frequency traders nowadays team up with supercomputers. They utilise computational power, complex algorithms, and data mining to gain an edge over competitors.
Centaur design and development
It doesn’t seem far fetched to presume that similar changes will soon happen in current basic digital infrastructure jobs such as design and development. We will see designers moving away of tedious and repetitive tasks of UI design adjustment and tweaking individual pixels and elements’ grid alignments.
Smart tools will free designer to focus on higher level goals and new type of details. As the technology continues to evolve the opportunities for brand design, for instance, will become more numerous than ever. AR, VR and techniques we can’t yet imagine will most definitely open up new possibilities for differentiation and new details.
I predict that “full-stack” designer and developer responsibilities will and can grow much beyond what is now feasible thanks to intelligent companions that allow humans to excel in new areas. Companions that balance the weaknesses and boost the strengths of each individual to new heights. In chess, there are now more human grand masters than ever before.
The name currently used for such approaches is generative design or algorithmic design. Variations and goals vary from each application to another as different groups are experimenting with what is feasible at the moment and what is the degree of human involvement in the approach. Currently the solutions in wide commercial use are Google Slides’ Explore offering layout suggestions and Wix.com’s website creator. Autodesk and Adobe will likely introduce (more) intelligent features to their mainstream products in the very near future.
For now, no one is organizing “designed by machines” competitions. Yet.
At SC5 we’re eagerly trying out new tools to support our work. We also build smart tools for our clients to help them became centaurs of their trade. We are always in for a talk about how to utilize state-of-art-solution such as machine learning at your company. Contact us and let’s work it out!
Text: Lassi A. Liikkanen, Data-Driven Design Specialist at SC5, @lassial
Illustration: Eija Jokilahti, SC5