Thursday 6 June 2024

AI, friend or foe?

Playing with it

I’ve tried working with AI and I’ve had a bit of fun:

  • I’ve got it to read out some of my work and although it does this with all the finesse of a sat nav font, it is gradually learning and may soon produce something very acceptable.
  • I’ve used it to help me get all of my bullet points in order. I’d been writing some simple tip sheets and I couldn’t recollect all me wont arguments about that advantages and disadvantages of both traditional publishing and self-publishing. It found them for me, suggested a couple of things I hadn’t thought of and a couple of things I didn’t agree with.  I couldn’t just copy and paste though; it didn’t sound at all like me.  I guess eventually it could learn my voice.
  • I’ve tinkered with art but I’ve not been all that satisfied with the result yet.  A story I recently had published on line had an AI generated picture attached to it; it looked as if my protagonist had three legs, and she was looking at two pairs of boots – which didn’t quite fit with the story.

Robotic code

I have quite a bit of AI in my WIP. It was in the previous novel in the series and will also be in the next one, the final one. All of my machines obey robotic code: they are there to serve mankind and will do humans no harm. I’m sure it’s not beyond the wit of those producing AI to build in some safeguards like these.

Is it really all that new?

We’ve lived with Amazon and Google ads algorithms for some time. Today I’ve been working with an author on her cover for a new book. There has been a lot of to-ing and fro-ing between us, the cover artist and her agent and I’ve spent some time looking at other appropriate book covers on Amazon.
Scarily in the middle of the latest round of emails Amazon suggested I might like to buy her previous novel.

Previous panics

We’ve often feared that something might put us out of work or look like cheating.  Every time these things have actually enhanced what we’ve been able to do.

Think of the reactions to:

·         The printing press

·         Paperbacks

·         Television  

·         Word processors

·         Spellchecks

·         The Internet

·         Google

·         Google translate

·         E-books

After a period of adjustment, these have actually offered us more opportunities and made us more efficient.

We might say the same about the cotton mills, Ford Motors, the automatic washing machine, the dishwasher and the way you can now tune pianos etc.  

The real danger may be in not having a human as a last resort

      I have three tales of talking to machines recently where the machine couldn’t cope and there was no easy way of contacting a human.  

·           Amazon: there was a problem about book supply. Actually if you persist you can get through to a call centre and they will sort the problem. Most of the time the human are more effective than the machines. But how to contact them isn’t obvious. The tick-boxes aren’tt fine-tune enough.  

·         Facebook: I’ve been on Facebook for over twenty years and am not doing anything differently from what I’ve done in the past. I’ve had two clashes with them recently

·         This week I was accused of spamming. Hmm, I’m just mentioning some of the stories that we’ve posted recently on our e-zine.  People follow our page to see his sort of content.  None of the tick boxes allow you to say that.

·         A couple of weeks ago I commented on the balance between good stories and good writing in a writers’ group I’m in. I’d read a fabulous story but it needed much more editing.  I pointed out how the editor and even the writer should be taken to task.  I used some figurative language – no swear words – and the Facebook machine just didn’t understand the metaphor. I did not name the book, the author or the publisher. Facebook did not accept the appeal. I was also barred from the group for a week. The group administrator contacted me and agreed it was stupid but he couldn’t do anything either.   

·         HMRC refunded me some CGT I’d paid. The robot couldn’t work out why and whether I should repay it and how.  It recommended I should talk to a human being. Except that human beings weren’t answering the phone for several months. I had to quote IT problems to get through. I suspect there was an error in their programming to do with accepting one-time direct debits.   

AI will get more sophisticated as it learns and when it comes to writing words, we’ll be the ones teaching it.  

Most tasks it will do more accurately that we can. But it won’t know when it’s doing something daft or brilliant; it will still need us to make that judgement call.      

No comments: