How to Speak Machine is written for non-computer people who want a solid grasp of artificial intelligence, machine learning, and the overall digital environment that seems to be controlling much of today’s world. John Maeda has written many fine books on technology design such as The Laws of Simplicity where he explores how hard it is to make something inherently complex simple so that it can be automated.
In the early 1990s John was one of the first electronic artists to become famous and since then he’s held many roles traversing art, design, technology and business including president of the Rhode Island School of Design, a professor at the MIT Media Lab, CTO of Automattic, and a venture capitalist. He is currently VP of Design and Artificial Intelligence at Microsoft.
Maeda describes computers as powerful invisible interconnected “soft machines”. Soft machines continuously collect, collate, and store data generated by us and by the environment in real time. Unlike hard machines, which have driven much of humanity’s advancement over the last two hundred years, we might only notice a soft machine when it fails. Soft machines produce experiences, hard machines goods. Maeda demystifies complex computing concepts, like processing loops and recursion, with common examples from our everyday world. Recursion is like a mobius strip or Russian stacking dolls. Exponential growth is based on multiplication. Neural networks are like a 3-D cube where points can reference each other across dimensions. Neural networks learn by detecting patterns in huge data sets and continuously evolve by processing new data, often in real time. To demonstrate loops and recursion (which allows computers to “think”) he shows simple computer code snippets in readable English. Computational thinking is not simple, but through his examples and good teaching he puts you in the ballpark so you can get a basic understanding of AI, its benefits and threats.
Maeda cites the short film “Powers of Ten” by Ray and Charles Eames which explores physical reality as humans know it. Though this movie was made a half century ago it draws thought-provoking parallels between the world we live in and artificial intelligence. In the opening scene, there’s a shot of a couple having a picnic in a Chicago park. The movie then zooms out by powers of 10, showing the city, the earth, finally coming to a full stop at the outer universe 100 million light years away at 10 24 before zooming back to earth in powers of 10 to the couple on the blanket, into the skin of the man’s hand, to the double helix of his DNA, all the way down to a carbon nucleus within a molecule of his skin at 10 -13. So, physical reality is recursive, infinite, fixed, and mathematical!
In the second half of this book Maeda looks deeper into at the nature of artificial intelligence, machine learning, and neural networks and how these advancements are different from human intelligence. He compares human capabilities and output to the French “au levain” bread, which is made with naturally occurring yeast. On the other hand, AI and machine learning are like “la levure” bread, which is made with synthetic yeast. While these two types of bread look identical, the French can immediately tell the difference by how they taste.
When working with machine-generated intelligence, technology creators should focus on producing a perfect understanding of the end user vs. producing a perfect design. This perfect understanding of the end user comes about by combining qualitative user research with machine-generated insights. The machine-generated insights tell you what people have done and their patterns but not “the why”. It’s essential to also understand their context, motivations, emotions, or other driving forces behind their behavior. Those developing the technology need to know how people are reacting to synthetically produced experiences and data. We’re seeing the very serious effects on society of profiling and algorithmically-controlled social media, but most of us just react. We don’t know what is going on because we don’t understand this new technology.
Much of society’s functioning will soon come to depend on AI because it is becoming more ubiquitous and reliable. But we’re at a point where we can still bend the arc, so technology serves human interests vs. those of big tech. To guide the future of AI we need to have a common-sense level understanding of how it works. That is, we need to know how to speak machine. While his book is a bit repetitive and windy at times, for me “How to Speak Machine” is one of the most interesting and useful non-fiction books I’ve read in several years.