I’ve always been of the opinion that the computer was invented a long time ago–probably by the ancient Egyptians. You see, a calculating machine need not be driven by electricity and switches, or even by Babbages’ brass cams and gears. A computing machine is a device made up of man-made parts, or even people behaving like man-made parts. Any bureaucracy is a computer, it has input and output functions, a memory, a central processor and a program. The program need not be electronically coded instructions, it can be just a set of rules written down somewhere that tell the human components exactly what to do or say under any conceivable circumstance.
The Egyptians probably had hundreds of these human computers, man-made offices that managed their building projects, gathered and stored their taxes, controlled their agriculture and granaries, their military, even their religion. A bunch of scribes ran the show, all guided by carefully spelled out instructions, and the flow of data was carried out on papyrus scrolls, and stone stelae. This system functioned almost independently of human control, as long as no unexpected situations arose. There was human management, of course, as in any data center, but they were scattered out at key points in the machine to monitor its functions and push the right buttons in case of emergencies. I suspect Joseph, the one of the many-colored coat, was one of these data managers, Pharaoh wants to raise taxes, or ship more rations to his pyramid builders, and Jewish Joe dictated the right memos and saw to it they were sent off to the district offices. No wires, no electricity, no Boolean logic gates, but raw data and canned programs nonetheless. Decision making is either fully automated, or it is passed up the line. The thousands of human workers that actually do the work need not give any thought to the overall task at all. They are programmed to follow instructions.
The ancient Chinese had a system like this, and we all know the Romans perfected the whole idea. These societies were run by bureaucracies, and these bureaucracies were human artifacts, even if their components were also human. The technology involved record keeping, writing, warehousing of records, and it was functionally the same as what happens in a computer: data processing and management. Later on came electricity, transistors, chips, integrated circuits and all the rest. The computer is merely the automation of bureaucracy.
My wife got a letter yesterday from an outfit identifying itself as ‘Highmark Retirement’. Neither of us has ever heard of these people. The letter appeared to be one of those government-mandated bulletins financial houses are required to periodically send to their clients reassuring them their business is being properly handled. My wife does receive a small pension from a now-defunct company she used to work for years ago, so I called the number on the letter which promised to answer any questions we might have. I got the usual run-around from a robot that flooded me with a vast decision tree of options and choices, none of which answered our questions. We could not even determine if it was her job pension that they were talking about.
We need to know this. Our bank could close, or we may be forced to cancel or close our accounts and we need to contact these people so they can wire us the pension benefit. Or my wife could die, and I’ll have to get in touch with them and report it. I am not a beneficiary of her account, I am not entitled to it. Finally I managed to work my way through the maze to the point where I could talk to a real person, but in order to do so, I had a few more hoops to jump through. They would talk to me person-to-person by phone, but I needed to present my “4 digit telephone pin” I didn’t have one, so they said they would text me one. I have a land line, no texting possible. So I kept on entering one-key responses until the software decided I needed to talk to a real person, and after much waiting and endless minutes of elevator muzak, I finally got “Irene”, a demure young female voice with an impenetrable Southwestern Asian accent.
Now I’ve got nothing against Irene. For all I know, she’s a perfectly lovely young lady working her way through to a PhD in particle physics by answering phones in a Mumbai boiler room, and I’m sure her broken English is much better than my command of her Punjabi dialect. But in spite of our communications difficulties, she finally was able to convince me that “Highmark” was indeed now handling my wife’s pension. So I tried to get her to give me a number I could call (she was not authorized to do this) in case I needed to contact Highmark directly to give them a new bank transfer code or report the death of a beneficiary. She would have to give me a “pin number” so I could call directly the appropriate office. She could not help me because she needed to authenticate that I was indeed my wife’s husband. So I asked her if she wanted to talk to my wife, which she did, and satisfied her requirement. (I guess that I could be a scammer working with a female accomplice, but I guess that was not in her programming). So she said she would mail me a new pin number so I could call them. I am now waiting.
All through this exchange it occurred to me that this telephone nightmare was remarkably similar to corresponding with ChatGPT. I was actually talking to an AI, although this AI was a human being tightly restricted in what she could do or say by a pre-existing set of instructions and rules. IOW, a fucking program. Automating these communications serves one primary purpose, it saves Highmark money by making it unnecessary to hire Americans, pay them a living wage and train them to know enough company policy to answer my questions, and enough authority to make any changes that might be necessary. Highmark saves money by my having to do what used to be their job.
I have no doubt there are many worthwhile reasons to develop Artificial Intelligence, and no doubt there are unanticipated consequences in its widespread implementation. But it is not, in principle, any different from any other technology. AI will be just more computer aggravation, and getting to a real human operator will just be harder. It will force us to be further alienated and separated from the everyday business of communicating to commercial or government entities to get our everyday business done. We are losing control of our ability to provide or receive information from the institutions that run our society, and which we depend on. The science fiction view of the future is one where brutal machine intelligences will enslave and tyrannize us. But the world I see coming is much worse than that. It is a future where we are controlled by intelligences indifferent to us, that don’t even know we’re here. The structures and procedures we are creating to administer our civilization are being primarily devised to save someone else money (not us!), and to make our active participation less essential. We are turning over our lives to processes, not intelligences. And those processes, no matter how benign they may have been in conception, do not have any idea we even exist.
Think about it. It is the automation of bureaucracy. It isn’t just a bureaucracy with automation, we’ve had that since Pharaoh’s scribes. This will be a machine bureaucracy.
The distinction is subtle, but no less profound.