My good friend Matt and I graduated college the same year. I went off into the work world and he headed for a graduate degree program in nuclear engineering. Much of the research effort in nuclear engineering is centered around developing sustainable fusion technology. Matt quickly realized that something was off.
So he went to his faculty adviser, who had been pursuing fusion research for several decades, and asked him, “I’m in my early 20’s, do you think that we will achieve viable fusion technology in my lifetime?”
The advisor’s answer was an involved discussion of “No.” Sustainable fusion technology involves an entire collection of problems that we are not close to solving. The materials science alone required to construct a vessel to hold the fusion reaction and extract power from it safely is well beyond our current capabilities even decades after my friend had the conversation with his advisor.
Happily for my friend, he had this conversation before he had sunk too much time into his research. Matt bailed out of nuclear engineering, changed his research focus, and has had a highly successful career in engineering education.
Meanwhile, I had been lucky enough to land a job doing Unix support at AT&T Bell Labs. One of the projects we supported was a research group that was working to develop a bespoke system that implemented Karmarkar’s algorithm for linear programming. This was an enormous project that employed hundreds of developers and consumed huge amounts of resources. The customers were the major airlines– scheduling aircraft and the flight crews that staff them is a classic problem in linear programming that directly impacts the bottom line of these companies.
You likely have never heard of Karmarkar’s algorithm, except perhaps for the controversy around it. Initially hailed as a major step forward that would revolutionize linear programming, its detractors claimed that, upon closer scrutiny, this so-called “revolutionary” algorithm was just a combination of known heuristics and speedups. It was not a substantial improvement over existing algorithms of the time.
I never studied the algorithm enough to determine which side’s claims were correct. What I do know is that the airlines pulled their funding and AT&T’s project was scuttled. The IT support team came in on Monday and everybody who was working on that project was literally gone. We moved through their empty office space for the next week collecting computer equipment to be repurposed for other projects. Some of the developers got shifted to other projects as well, but I imagine many people suddenly found themselves looking for work.
The airlines poured millions of dollars into a project that produced exactly nothing of value. Governments around the world continue to pour billions into fusion research with little to show for it and very little hope of fusion power in our lifetimes. Why is so much time, effort, and money being wasted?
These projects have several factors in common. Their goal is highly desirable: a “revolution” that would reshape the world as we know it, or at least an entire industry. The path involves highly complex technology that is impenetrable to a non-specialist: a complex algorithm or deep scientific research necessary to invent things that have never been done before. And they require massive amounts of funding.
This is a perfect recipe for bad decision making or outright fraud. People will sacrifice a great deal to achieve a significant goal. Because the path to that goal is difficult to comprehend, people will fool themselves into thinking the solution is “just around the corner”. Critical thinking skills fly out the window as people focus on the goal and can’t or won’t focus on the process to get there.
And when the project attracts unscrupulous operators who realize that there is money to be made in prolonging the effort, you have the makings of a bezzle. The unscrupulous promise a wonderful new world but use any excuse to keep extracting money from the situation. When challenged about their lack of results they just say, “Technology is complex and unpredictable, but I swear we are almost there!” Technology is a perfect breeding ground for bezzles because we have socialized the idea that computers and technology are inscrutable to mere mortals who must defer to a high priesthood to interpret the signs and omens.
“Generative AI” and “large language models” are the latest techno bezzle. But “AI” is a constant and recurring bezzle that I have seen numerous times in my decades in technology. Remember “machine learning”? Remember “neural networks”? I have lived through too many of these hype cycles and seen too many people lose their jobs and/or retirement funds due to companies that bet the farm on the latest bezzle.
The AI hype is too strong right now for me to convince people caught up in it that they are being conned. But for the rest of you I want you to recognize the patterns at play here and apply your critical thinking skills to any new “revolutionary” technologies that follow a similar path. And try to educate others so that we don’t as a society keep making the same sorts of mistakes over and over again. The resources we are wasting on the current AI hype cycle are killing the planet and could be put to so much better use.
Righteous IT 