Tag: chatgpt

Okay, I’ll give you my thoughts on ChatGPT

It’s the topic du jour, and since I was the person who predicted the path we’re now on, I’ve been getting some requests for my thoughts about OpenAI’s ChatGPT, which is currently busy cranking out thousands of poorly-written articles about itself for publication by every news outlet in the world.

I’ve ignored every request to comment on ChatGPT, including on whether it should be welcomed into schools or blocked, whether this means the end of Google, how it will change SEO, and whether ChatGPT will mean the end of human-to-human relationships as people find out they enjoy being lied to by a computer much more than by another person.

I’m not quite ready to write my thoughts on how ChatGPT will cause the end of the world or bring about a work-free future of leisure and bad novels that never end, but I have some observations about my reaction to playing around with it that I’d like to share.

Observation #1: I don’t want it

As someone who has just come off a year of writing and editing three books about computer programming, at a rate of approximately $10/hour, I hate that any reader or editor might look at what I’ve written and say, “Why did we pay Chris $10/hour to write about JavaScript when we could have had ChatGPT do it for free?” Or, even worse, I fear that the day will come when an editor or publisher says to me, “You write so fast. Are you having ChatGPT write your books now?”

To attempt to head off both things: Can AI make spelling and coding errors like I can? I don’t think so! I can type really fast (thanks 7th grade!) and I’m sort of dyslexic, so I don’t notice that I’ve reversed words or substituted “futon” for “function” (as in “JavaScript futons are objects”), and my writing not only keeps me employed, it also keeps a giant fleet of expert copyeditors, technical editors, proofreaders, and Amazon reviewers gainfully employed. Give my projects to artificial intelligence and you might as well kiss your job goodbye too.

Observation #2: I’m an old fuddy-duddy

As someone who tries to keep an open mind about things, I struggle with the following probabilities and conflicting feelings:

  1. A world in which someone can type any garbage prompt into ChatGPT and get back a grammatically correct, if not factually correct, response, or where someone can tell ChatGPT to express their half-baked ideas (or worse) in a coherent way will likely lead to a world in which learning to write is no longer valued.
  2. A world in which ChatGPT is baked into Gmail and MS Word is a world in which I won’t have to gag from seeing so much bad writing every day.
  3. Kids today are dumb and they’re just going to get dumber.

I never wanted to be the person who grows up to say that kids today have it easy, but there you have it. I’ve worked with people who think tracing a cartoon or otherwise copying someone else’s drawing is the same as making their own art. People have claimed to be “computer geniuses” in job interviews with me who list the fact that they know how to post on Instagram as proof. Arrg.

Observation #3: I could use this

If you know what you’re doing and how machine learning works, and how to distinguish truth from fiction, and how to write code, and how to write English, AI tools can actually be helpful for generating ideas or for triggering creativity.  There are plenty of possibilities that I’ve considered, and I’ve even used AI to assist me with writing the type of mindless code that counts more as typing than actually coding. It’s even saved me time. But, would I trust AI to write a computer program in a language I don’t know inside and out? Would I let AI write something I’d ever submit to a publisher? Would I ever accept that giving ChatGPT a prompt and pasting the result on your web site or into MS Word is in any way a creative act? No.

Conclusion

A lot of articles about ChatGPT today end with conclusions similar to “Well, it’s a great tool, but you can’t rely on things it writes to be factually correct. So therefore, we’ll all still have jobs forever.” That’s not the way I see it. The “factually correct” part is the easiest part to fix, and it will be fixed as soon as someone hooks an AI like ChatGPT up to Google.

Unlike every ChatGPT-generated article I’ve seen, I’m not going to tie this up in a neat little summary. I’ll write more and talk more about my latest end-of-the-world scenario soon.