It seems inevitable that by the end of this decade, we’ll reach a new level of software automation in our lives never seen before. Between drones and robotics and autonomous cars and possibly the Internet of Things and automated sensor networks, these software systems are becoming more pervasive in day-to-day life with no real end in sight.
The Glass Cage: Automation and Us is a discussion of what all this automation means. It’s published in 2014, and slightly dates itself from a technological standpoint given how much progress we’ve made in even just the past 2 years, particularly around self-driving cars and their future implications. Automation may be one of those technologies that brings about exponential change to society, so it’s not surprising that the thinking has advanced substantially even in a few years’ time.
The Glass Cage mitigates obsolescence, though, by looking at examples from further past; specifically, it explores how aircraft have changed with the introduction of auto pilots, which it has done so since the 80s. In the case of flying a plane, the tradeoff is between the added safety of letting a computer always make the right decisions, versus training human pilots to act in emergencies and to go beyond pre-programmed responses to find creative solutions (e.g., the Hudson River crash landing). At the same time, the economic ramifications of using software over human beings has played itself out over decades, and what used to be a glamorous profession that employed half a dozen people per cockpit is now trimmed down to two underpaid pilots per flight.
The book also looks beyond just the obvious issue of robots-replacing-people. It asks questions about the reliability of both software and hardware, the legal frameworks that don’t currently exist for automated decisions and errors, and the problems that humans would face by relying too much on automated systems. Even if the AI takeover is inevitable, there will still be a long transition period where humans have to co-exist with software and machinery, and there’s a lot we haven’t figured out in this relationship.
I like The Glass Cage. Most of the research isn’t original and I’ve likely read a lot of it via everyday articles, but the book provides a good summary of the current state of automation and raises some good questions about where we are now and how we’ll likely stumble moving forward. As someone who creates software, it has made me stop and think about the one-dimensionality of relentless efficiency via computing, and that there’s so much more to consider.