There was this sudden flurry of people putting up poetry (mostly what poets would dismissively call 'rhymes') and essays and whatnot, leading me to wonder about whether there was a new virus infecting people with the urge to write. Reading the fine print, I realized that there was this experimental bot sort of thing called ChatGPT which was actually doing all this writing with my FB friends egging it on to do so.
Before too long, all this poetry was accompanied by laments about how this AI-powered bot would cost jobs in the millions. The dratted thing could apparently write code, essays and, possibly, stand on its head (if it had one) if asked to do so. IF the said lament was in rhymes, one had a sneaking suspicion that that chap had used ChatGPT to write the lament about how ChatGPT would steal his job!
And then there were others who said, "Not at all! Yes, it will take away some jobs but it will create new ones. Every time when new technology comes in, we fear loss of jobs but, eventually, we find that more jobs get created." Makes one wonder, though. I mean, like, we all talk of technology as improving efficiency which, essentially, would mean that we have to work lesser, right? If we needed to work less, then either number of jobs should go down or the number of working hours should go down, no? And, yet, throughout the history of technology we have, apparently, only increased our total working. Then why are we still chasing this efficiency which only seems to create more work for us instead of reducing it? (Yeah, yeah, it is all because we keep increasing our needs to suit. The point is, no increased efficiency, no having to pursue new needs, no?)
Anyway, that's for other people with brains in their head to bother about. I was sort of happy that ChatGPT was around so that I could offload my blog writing to it and relax. And then...
There is this IIT chap's lecture which I happened upon. He says that it can actually give wrong results. Like, put it to create a tour itinerary and it can give you hotels that do not exist. Ask it to write an essay on economics and it is known to have cited a paper written by authors who exist, in a journal that exists...the only problem being that THAT paper does not exist. So, he says, the thing about ChatGPT is
1. Do you need accuracy? NO? You can use ChatGPT.
2. Do you need accuracy? YES? Do you have domain knowledge to verify its results? YES? Then you may use ChatGPT.
3. Do you need accuracy? NO? Do you have domain knowledge to verify its results? NO? Are you willing to take on the consequences of putting out erroneous information? YES? Then you may use ChatGPT.
4. Do you need accuracy? NO? Do you have domain knowledge to verify its results? NO? Are you willing to take on the liability of putting out erroneous information? NO? Then you should not use ChatGPT.
So, the net result for me is that, apart from poetry and fiction, ChatGPT will be MOST useful for what is known as WhatsApp university. Citing non-existent papers by well-known sources, proving anything you want to prove? RIGHT up their alley!
Accuracy, not needed. But, yes, they still need 'domain' knowledge...as to whether what ChatGPT has put out supports their POV or opposes it...which they probably have. And liability of putting out wrong information? What is that?
So, there! Going by this, the first jobs that it is going to cost is most of those meme-creators. Great!
Great? Not really! Ye gods, the floods of such info a bot can produce...