amen zwa, esq. on Nostr: For more than four decades, I have liked #AI (rule based, connectionist, statistical, ...

For more than four decades, I have liked #AI (rule based, connectionist, statistical, the lot) as a computational tool, not as a religion.

What I find alarming with the present trajectory of AI in #IT is the complete lack of a common, consistent, equitable, enforceable code of #ethics. This starts from the very top of the IT executive tier, all the way down to the lowly data loader.

Another troubling matter, I find, is how most AI practitioners in industry sees AI is nothing more than API call fest. They cannot be bothered with learning the underlying mathematical and computational concepts. They invariably develop intense crushes on specific model or architecture, and apply that one favourite model to any and all data sets. They possess no statistical skills to analyse the problem domain and the collected data. This lot is using AI, often indirectly via some form of ChatGPT-style textual interface, in a way not too different from a young child who just learned addition, and applies the addition every time he sees a pair of numbers.

I have no objection to laymen using an LLM to conduct their business, as long as that LLM was developed and supplied by conscientious, ethical folk with keen insights into the problem domain and the solution techniques. But that is not at all the case, at present.

These trends have harmful potential, not just against the IT industry, but also against the whole of humanity.