Will frivolity sustain the development of the new AI?

Google’s Bard has been rather a disappointment for me so far.  Several days after promising to try to edit my book draft (uploaded to my Google drive) it does not even seem to have tried. I have also asked it to find current examples of ‘bootstrapped’ (self-financed) new businesses and its answers were entirely useless.  Chat-GPT’s responses projected more confidence but were equally wrong.

I have been using Google Assistant for several years, and it too cannot answer more than the simplest queries.

Will more training make these chatbots more reliable or useful?
I am skeptical.  Think of the “support” personnel who are supposed to help you resolve problems with high-tech products, travel arrangements etc.  They are equipped with “scripts” and presumably human intelligence, but they simply cannot deal with anything out of the ordinary.  Yet I (and presumably others) usually contact them with problems that we cannot resolve after considerable google searches on our own.

I am likewise terrified to read that ‘bots’ are now writing software code, that may end up as components of important applications like air traffic control systems. Rather like poorly supervised labs in Wuhan playing around with viruses.

Yet the capacity of the new AI to make up stuff on the fly, e.g. write ballads, stories, etc. and create images is surely breathtaking.

So perhaps the continued evolution and refinement of AI proceeds through fun and entertainment.  This is a time-honored pathway.  Some new technologies do indeed start off solving practical problems – the first steam engines, first commercialized by Newcomen and then greatly improved by Watt, were used to pump water out of coal mines.  But many new technologies are initially developed by and for hobbyists: bicycles, gliders (that laid the foundation of powered flight) and automobiles for example.  VCRs, I am told, found a foothold through X-rated titles. Fanatical ‘gamers’ have supported the development of advances in computer graphics and image processing.

Similarly, the harmless, frivolous story-and joke telling capabilities, rather than “serious” – and risky — business or technical applications are how the technology advances. But can these capabilities justify the enormous investment? We shall see.