PERFORMANCE RIGHTS, page-23

"Software replaced filing cabinets in 1980s" – it replaced humans, not filing cabinets. Filing cabinets don't send invoices and letters, or process payments,

or any of the millions of other mundane procedures that software took on. But without the discretion and intelligence of a human, software also ushered in the age of "computer says, 'nah!'", sometimes described as "I'm sorry our system can't accommodate your perfectly reasonable request".

"this isnt software its an algorithm" – pardon? How do you suppose the "algorithm" is implemented?!

"that learns over time" – absolutely incorrect! Once trained the model is static. This is mainly because nobody has yet worked out how to incrementally train a neural network... not even a small one... and these "AI" models are enormous. It's also because there is no feedback source for the model to continue learning from. When you chat with ChatGPT for instance, what is it going to learn? Ask ChatGPT how current its knowledge is and you can verify that it isn't learning over time.

You're one of the "Happy non-engineers" from the illustration above. You've been seduced by the term "Artificial Intelligence" into thinking that the trained model is somehow intelligent. It isn't. The "AI" is in the training. Once trained the model can be knowledgeable, and can behave in complex and sometimes unpredictable ways. But it can't "think". It can't solve novel problems. It can't circumvent its training. If it produces a result that is nonsense or inappropriate it has no way to realise this or adjusting its behaviour.

"this time the adoption will be at warp speed 5 imo" – something we can absolutely agree on... and the result will be carnage.