An impractical guide to hyper-automated systems
As those who have read this column over time will understand, I have a workbench that includes authors, both academics and consultants, who manage rather than teach. Unfortunately, “The age of invisible machines“
Follow in this direction. It is presented as a book that describes how artificial intelligence (AI) chains and networks, mainly topped by chatbots, will be of great benefit to society. What it does do is provide reasons why executives are good at replacing people with machines while not discussing the real impact of that change.
The good part of the book is the initial discussion of chatbots. The author gives an excellent description of why systems are still evolving and that the main problem is that a chatbot and an intelligent chatbot are two different things. I regularly come across retail chatbots that are no smarter than Automatic Phone Call Distribution (ACD) systems that just ask a series of questions to direct the call. Supposed chatbots as a series of limited questions and if a prospect or customer can’t address these simple questions with simple answers then Natural Language Processing (NLP) can fit into a few boxes then the system is useless.
The book provides a description indicating that a useful chatbot must be linked to back-end systems in order for it to understand and address a wide range of problems. Most of the systems are not there yet, but progress is being made.
The first major problem with the book is that the above is hidden in a six-chapter first section that appears to be very ridiculous nonsense that exists to justify a book rather than an article. It also includes the problem I mentioned in the first paragraph, best exemplified by a quote from Chapter 5, “Humans may be affected by faxing, but they are unlikely to depend on a machine that simply replicates something they can already do themselves.” The problem with that is, simply put, that it Error. Executives would be all too happy to replace humans with replica files that do what those humans can do themselves, so long as the faxes save money that could go to executive bonuses and stock value.
There is also a glowing section about China’s social ranking system. The book makes a good point, that limited use of this would be better than existing customer rating systems for companies and products. While I can see companies buying into such a system for reviews, the statement that “this outcome will be accessible to all, influenced by our continued interaction with private companies, individuals, and even government.” So everyone and everything? I’ll pass.
The rest of the book is about how to use consultants to help you build superior automated systems. This is not surprising, given that the author is such a consultant. However, there are two major problems with it. The book, of course, discusses current technology. This is what business should be concerned with. However, many technologies have shown that progress avoids intermediate solutions. The advent of the GUI and fourth-generation languages demonstrated this, as just two examples. The processes listed here are fine for now, but any outlet should be willing to keep scanning the horizon, as newer, faster, and more powerful solutions will continue to emerge at a fairly fast pace.
The second and larger issue is mentioned above. Inventors have a habit, since long before Alfred Nobel, of ignoring the consequences of their inventions. The excuse is the same as scientists often make, it’s not up to them to decide on user and societal impact, they just figure out and invent things. While this is true of theoretical science, it is time for technologists who focus on applications that directly impact society to give up trying to excuse themselves from societal impact.
The ethical AI movement is just an extension of the systemic movements in society, movements that try to understand how change will affect those societies and do so from the start. Any good programmer considers system problems from the design stage. Waiting to debug is too late to create an efficient system. It is clear that AI will affect society in major ways. It will redefine who can work and how society must address change in the definition of work. It relates to the overvaluation of stocks because of the promise of solutions, and in most people’s lack of understanding of what those solutions mean, and of a very few, a real understanding of what they mean.
Society in a New Gilded Age. The first has led to regulations and protections that have been destroyed over the past 40 years. This new one has more challenges and risks than those that arose in the industrial revolutions. No one should talk about systems that have an enormous potential impact on society without talking about those impacts. This book takes the age-old approach of ignoring society while pushing for great societal upheaval. This means that I cannot recommend this book.