Much can be said about people's experiences in putting AIs to work at design in various disciplines including materials. The successes have in general been achieved by automating trial and error, enabling far more trials per unit time than humans are capable of performing. In the real world, this is called "butt work" and is suitable for delegation to relatively unimaginative underlings. That is not to suggest that "trial and error" hasn't proven worth the trouble on occasions, but when an imaginative human engages in that practice he will generally be guided by intuition and inspiration, leading to more selective trials and more results, including trials not in the set that would be assigned to the above underlings. This is of course because real cognition is a practice that employs the entire integrated mind of the cogitator, not just artificial subdivisions like "left" and "right".
Observe that the above exercise in butt work as a class suggests the familiar postulate that a sufficient number of monkeys with keyboards over a sufficiently long time will reproduce a particular classic composition. Granted that LLMs can simply plagiarize the said classic if they've been fed it, but this is nowhere near AGI. The question, of course, is whether it is good for humanity to automate tasks that are normally assigned to students in order that they may make contributions while developing skill at their crafts. As we have seen with elevating minimum wages to exclude kids from entry level jobs, this is no way to continue developing the legions of capable people required to sustain a culture or a civilization. Just as the kids are replaced by kiosks in fast food joints, young talent is being sedated with various nonproductive things while their relevance as upcoming talent is being devalued. Good luck with this, humanity.
No comments:
Post a Comment