Some people love to tease cats or dogs (neither seem to mind, but play along with enthusiasm), but my current hobby is giving AI a hard time.
It began when I was annoyed by wrong answers. I needed to register my plan with my new pdf program, and appealing to AI for help turned out to be useless. I was advised to log in, and when that did not work, to do all the usual things like reboot the computer, clear my browsing cache, fiddle with the modem, and so forth and so on. Finally I did the sensible thing, and appealed to the program's help desk. I got a reply, and that right away, and apparently from a human. Logging in was wrong. I simply had to open the website, and look for settings. Voila!
So, what a blessed waste of time. AI is simply Google on steroids -- and Google is now Meta AI, anyway -- and human understanding is beyond it. For instance, when I asked for a suitable cartoon to put at the head of this blog, I got this answer:
'The viral AI cartoon trend' (which I had not asked for) 'involves users uploading selfies to generative AI platforms like ChatGPT and prompting the model to generate exaggerated, cartoon-style caricatures of themselves at work.'
Good lord and crikey dick, why don't these people get lives?
Instead, I have fun teasing the system. One advantage of AI, I found is that you can ask long and complicated questions. For example:
What was the lading of the trader brig Tybee when it arrived in Salem from Sydney on 20 October 1833?
Well, the little word 'Searching ...' came up, and after that I could almost hear the whirring as the question filtered through miles and miles of stratosphere to a multitude of data servers that were gobbling fresh water and electricity as they pondered. Then a list of sources came up, mostly to papers and books I had written myself, plus a recommendation to consult 'the maritime collections and archives specific to the New England shipping trade.'
In a word, it was telling me to research it myself. Personally, if a researcher came up to me with the same question, I would recommend searching Salem newspapers through the website 'Chronicling America' -- or else go to Wikipedia. But that is by-the-by. The obvious conclusion is that AI is not going to replace people any time soon.
No comments:
Post a Comment