AI

A couple of days ago I signed up to The Human made Webring, a ring for people actively avoiding AI on their own web sites. That is not all of the philosophy of course, but I have a confession. Not a deep or in-depth confession.

I use AI.

I have been thinking, how do I use it?

Most of the time it is entirely involuntarily. That said, I do try to avoid it wherever I can:

  • I switched AI off on my iPhone. Being honest, when I last tried it, it was not much help, especially attempting to summarise incoming messages or grouping alerts. Now there may be a time when it’s unavoidable, you know how fashions go.
  • It’s switched off at Github.com. I do not code much but when I do I prefer to do things myself (yes, with help from StackExchange.com).
  • I use the Kagi.com search engine (without its AI). It simplifies search, no AI summary, no ads, no sponsored content. It finds stuff without the guff.
  • I haven’t touched Facebook for a while. It’s sometimes necessary to find out what’s happening locally or for shop opening times. I do use WhatsApp because I’d struggle to keep in touch with people I care about – those who don’t use Mastodon.
  • Twitter is occasionally a necessary evil for hyper-local news.
  • My car is over 16 years old and not even the clock updates automatically. I don’t know if I’d have auto updates by spending the extra £1,000 for the top of the range version. But even that just had more things to potentially go wrong.
Stella cat sat on a window ledge. Through the window a quite old red car with not much in the way of modern tech, and certainly no AI.
Stella cat sat on a window ledge. Through the window a quite old red car with not much in the way of modern tech, and certainly no AI.

When I cannot avoid AI:

  • I scroll past it at work when I’m using Google or Bing. (I’m not paying for Kagi at work). I’m only enough to remember an internet without Google. The problem is, I could probably save time by using the data in the AI summaries, especially some of the more technical stuff.
  • I check sources sent to me from colleagues. It’s not that I don’t trust their judgment, it’s just I’ve learned a lot through bitter experience.

What I’m unsure about:

  • Translations.
  • Everything else online. This one’s a biggie isn’t it.

The thing is, how can AI and large language models be defined? I’m old enough to remember (you’ve heard this phrase before) much simpler stand-alone and connected systems:

  • When computers weren’t powerful enough to do anything but simulate an intelligence, e.g. ELIZA, a natural language processing program. It was interesting but after even casual use didn’t really give the impression it understood the conversation. Illusion.
  • Infocom adventures – interactive fiction. They used a limited parser to navigate a path through imaginary worlds full of monsters and traps. Spellbinding stuff. If you’ve ever seen a variation on “You are standing in an open field west of a white house, with a boarded front door. There is a small mailbox here.” then you have a glimpse of how important the Zork adventure was.
  • The first telephone banking automations, the “Read out the 16-digit long number after the beep”, and the “Say yes” kind of things.

I suppose one could lump these two into a small language model type? And I’m probably using the word ‘simulate’ incorrectly too.

Dunno what’s next as I slowly remove AI.

Obsolescence?