Learning how to speak machine in today's world
Computers have learned a lot from humans; now we must speed up to keep pace
Once considered novelties, computers have become an essential part of daily life
The rapid pace of change has fundamentally reshaped how computers work today
Computers are now incomprehensible to humans who do not “speak machine”
Most people have outdated views on computing; we need to boost our knowledge
If non-techies remain out of the loop, Big Tech will continue to shape our lives
We need more diverse representation in both technology firms and “thick data”
Today’s computers run on data, not instructions
When my book, The End Of Insurance As We Know It, came out in 2019, I checked the sales metrics regularly to see how well it was performing. I was interested in how many copies sold each month and the split between the paperback (or “dead trees”) and digital Kindle e-book versions. I was also fascinated with the performance of the book under the Kindle Unlimited program, which is a subscription service offered by Amazon. Readers pay a fixed monthly price of $9.99 each month and have unlimited access to over 1 million titles for no additional cost. Publishers must decide whether or not to include titles in the program, and royalties are paid out based on the number of pages read rather than for each copy sold. For example, my book is over 300 pages long, so if someone reads just 20 or 30 pages, the royalty I receive is only a fraction of the amount received for a traditional sale. So as an author monitoring my sales performance, I see the number of books sold and the number of pages read by Kindle Unlimited readers. That’s right - each page read is being tracked and monitored by Amazon and reported to me at an aggregate level.
The ability to track pages read on a digital e-book initially struck me as a bit strange, but in reality, all of our digital interactions are monitored and tracked. This “data surveillance” is how most technology firms work today; they use vast amounts of digital information (commonly called “Big Data”) to tailor services to their customers. Many people are concerned with what data is used, whom it is shared with, and what rights to data privacy we have. However, the simple fact is that we need to consent to collecting and using our information to receive the benefits of these offerings. Lawmakers in some jurisdictions, most notably in Europe and California, have established rules regarding data privacy. There is also a transition underway in the industry with the elimination of third-party cookies, but today’s computational machines run on data - lots of it.
Looking to shape the future? Subscribe free to Forestview and never miss an article.
Speed rules in technology - how do we catch up?
In his book How To Speak Machine, technologist and design pioneer John Maeda shares his recollection of having a stand-alone PET computer in his classroom as a child. (I did too!) The PET was primarily a novelty; it was not connected to any other computers or a network and had limited memory, storage, and processing power, so it was not very useful. The PET required instructions to be provided by a programmer, and Maeda remembers the first program he coded in BASIC fondly that helped manage billing and invoices for his family’s small business. Maeda became fascinated with what he describes as computational thinking when he attended a class on programming and learned that his code could be greatly simplified through the use of recursive loops. Maeda talks at length in the book about how programming a computer obeys logic and rules that do not apply to the physical world, and that the ability to “speak machine” is an important skill for all humans as our lives become so inseparable from the technology we rely on.
Over the course of forty years, computers have grown much more powerful, and perhaps more importantly, they learned to communicate with one another. With the introduction of networking to link computers together and the Internet to connect millions of devices together, computers moved from novelties to necessities. In Maeda’s view, machines now communicate and cooperate together better than humans do. The extreme connectivity of our machines is now paired with the ability to “think” on their own through artificial intelligence. Today’s computers rely less on specific instructions from humans in the form of computer programs and more on machine learning based on neural networks to build “black box” algorithms. The combination of high connectivity and the ability to develop algorithms that cannot be explicitly seen or understood by humans is a worrying tandem to Maeda. Humans are more reliant and less in control of computers than ever before.
Maeda is well-known as a major proponent of the value of good design, but in the book, he criticizes the “gatekeepers” of good design such as museums in favor of a more functional view. Maeda argues that software products and services are fundamentally different than traditional manufacturing: the cost of failure is low relative to the cost of being slow. So firms have embraced the ethos of the Agile methodology and minimum viable products (MVP) over more polished offerings. The ability to collect data, learn from customers, and make refinements quickly and iteratively based on user feedback is the norm today. However, many of these decisions are based on A/B testing and “thin data” and prioritize the needs expressed by the largest group of customers. Maeda advocates for making product decisions based on both quantitative as well as qualitative data. He encourages firms to invest in ethnographic studies, spending time and money getting to know a cross-section of customers individually to build out “thick data” sets and leveraging AI to optimize decisions based on both thin and thick data.
Why inclusion matters in an AI-driven world
Just as speed pushes technology firms to prioritize the collection and use of thin data over thick data, Maeda also observes that speed is one reason for the lack of diversity in Silicon Valley. Building out diverse teams and embracing inclusion may drive innovation spurred by a diversity of thought, but it also leads to more discussion and contemplation that can slow down the decision-making process. This is not necessarily a bad thing: a more deliberate process can ensure that firms avoid making big mistakes and risking reputational harm. But, in Maeda’s view, the shift from a waterfall software development approach to Agile methodology, knowing that mistakes can quickly be corrected in a future sprint, also leads to the deprioritization of building an inclusive team. Mistakes that are made will simply be cleaned up later.
As AI becomes a dominant force in our everyday world, algorithms and not programmers are shaping our destiny. Humans are not constantly poring over the decision-making logic in the same way they once were. AI is not biased on its own; it relies on humans to bring in our baggage. However, once incorporated into AI, factors such as racism, sexism, and inequality can become baked into software solutions. Since programmers aren’t actively rewriting lines of code as the system is continually modified, built-in biases can perpetuate for a long time and not be captured by the massive amount of thin data captured on users. It takes a deeper understanding of under-represented populations - the thick data - to flag issues when they arise. This is one reason why having a diverse team working to develop products and solutions at the beginning is so critical: it helps ensure bias does not get incorporated into the initial MVP.
As I wrote in the last edition of Forestview, the big debates revolving around ethics and AI are not going away anytime soon. Maeda talks at length in his book about our awkward transition to a “Mooreian future” that is governed by power laws and exponential properties rather than incremental changes and linear thinking. To Maeda, the main reason that Big Tech is under so much scrutiny around the adverse impacts of technological change is that they are the only group who truly understand how our world is shifting. Since they are the only people who can “speak machine”, their homogeneity and lack of diversity is a major issue for the rest of society. Part of the solution is to ensure steps are taken to insure inclusion in the development of AI-based products and services. Another part that Maeda advocates for is building up our knowledge among all members of society about how to think computationally so that we become more fluent in “speaking machine”. If we update the operating system in our brains, we will be better equipped to have informed conversations to shape our future together.
Do you feel you are a strong “computational thinker”? Why or why not? How would you evaluate the ability of your organization generally to “speak machine”? How do you balance speed with quality today? How can you ensure inclusiveness without losing quickness in making key decisions? What role does change management play and is your organization devoting enough resources to it? How can you prepare your people for our AI future?