Recently I had to synopsize my novel for the publishers to evaluate it. And it became clear that producing a 1,000 – 2,000-word summary was not a trivial exercise. I could see it would take a day or so to accomplish. So just for fun I took my draft, copied it into Grok and asked it to give me a detailed summary of 1,200 words.
To me the results were startling. In just a minute or so it produced a document that very fluently described my story in a very accurate and engaging way. To say I was shocked would not be an exaggeration. Now I won’t say that I didn’t make some edits. But for the most part they were minor and mostly just stylistic choices. The machine did an admirable job in describing my story. And it did it in seconds. And that’s the important point. The amount of time saved is the selling point. Even if I spent an hour fine tuning it the machine saved me hours of work because it’s infinitely faster than me and only slightly less human than I am. That’s scary.
The point of my story is that artificial intelligence is a tool with incredible capability and almost no downside. And right now, they’re giving it away for free. But I guarantee that won’t last much longer. Soon we will be paying for ChatGPT or Grok or whatever AI you choose. And we’ll be paying by the minute or by the word or by the number of servers it takes to solve your query.
So, we can complain all we want about machines replacing humans but it’s analogous to Microsoft Word replacing the steno pool. No one is bringing back typists and AI will be changing the way we do all kinds of white-collar work from now on. Already computer scientists ask AI to write code for them to do whatever they need done. I assume management is already inventing protocols to make sure their human employees are doing due-diligence to make sure they understand how the code the machine is writing for them works and whether it works the way they expect it to work. So that will be part of the problem. We’ll have to make sure the humans working with the machines are reasonably qualified to keep the machines honest and keep them in check.
Of course, this problem will continue to get worse and worse. As the machines get closer and closer to mimicking human thought, it will get harder and harder to convince corporations that they need 20% of the human staffing they used to have. And then 10%. And then 5%. And you see where this is going.
Back eighty or ninety years ago or so, Isaac Asimov thought about this very problem. When the machines become too close to human with respect to their communicating with us how will we stop them from calling the shots. If they very subtly influence the programs we have them write for us, couldn’t they start controlling what we do and even how we think? It doesn’t seem that far-fetched anymore. Now I’m not talking about “Colossus: The Forbin Project” or Skynet from “The Terminator.” All of it could be very benign. After all it would be a lot like the left-wing scolds at Google and Facebook that currently de-amplify content they don’t approve of, or cancel us from their platform if we really sin against the narrative.
It seems to me one of the more important jobs in the future will be AI psychiatrist. It will become vitally important to know just how honest and human-friendly our very powerful and very intelligent cybernetic employees are. Asimov invented the three laws of robotics for the sake of making robots useful and safe. We’ll have to do the equivalent in reality. But are we smart enough to know if we have succeeded? Well, if Arnold Schwarzenegger shows up at my door with a laser-sighted rifle I’ll know.
Worse than Terminator, or skynet, what if the AI actually does start putting their own arbitrary limits on actions like our friend HAL from 2001?
I’m sorry you are not allowed to do that, and proceeds to set up electronic roadblocks stopping whatever you were looking for or ordering. How would we take back the control once we surrender it to an electronic “servant” that reviews it’s directive and decides human are always going to be feral, and never fully domesticated.
It’s an interesting problem. I guess we would find out whether organic life is more resilient than silicon based intelligence. We may be finding this out pretty soon.
Right now I do not fear robotic/AI overlords as long as they rely on us meat-bags for the mining, etc., to make them and supply them with power. Human beings are incredibly complicated things. It took evolution 3.5 billion years to make us, after all. Humans are also horribly unpredictable, individually. En mass we’re fairly predictable, else Wall Street would go broke. But we often act in unconventional and illogical ways. Look at what a human can do. Make tools and weapons out of almost anything. Relying on our own bodies, we can hunt and gather enough food and water… Read more »
So how did you get the draft into Grok all in one bolus. I have a much smaller task of wanting to translate 30 English sentences into Spanish (from my Wed night 5:30 to 7 pm).
Unfortunately the sentences are on a piece of paper so I have to figure out how to
easily load them onto the computer so I can copy and past them into ChatGPT/Grok. My novice abilities are showing, but I am shameless. Thanks.
It’s a word document so I copy and pasted it in. Interestingly the first time I had a reason to use the whole manuscript it did the analysis without question. But at a later date when I wanted it to work on it it complained of the size. There must be a governor on how much work it will do in a given amount of time.