giardini
Joined 19 years ago
Joined 19 years ago
How many books did you read in your education? Far fewer that what LLMs read, no doubt. Of course we "have a life" and so we interact with hundreds of people, which is also a form of learning.
But in any case there's are fundamental differences here between these two systems (LLMs and humans) that we have not yet discerned clearly.
I likely would have been better off taking Spanish but I avoided that b/c half my class were fluent in Spanish already and I had to maintain a decent grade.
- how to tie his shoelaces,
- how to make/change his bed,
- how to make change (currency),
- how to cook breakfast: boil oatmeal, fry and poach eggs, make toast w/butter,
- how to wash dishes w/o a dishwasher (the value of soaking dirty dishes),
- how to sweep an inside floor, how to sweep large outside areas (e.g.,driveway),
- how to mop a floor and with what (e.g., don't mix stuff!),
- how to change/patch a tire on a bicycle, car, scooter, et al,
- how to use a fire extinguisher,
- when he gets older: how to mow and garden a lawn safely (safety glasses, heavy shoes, keep hands away, proper tool for the job, etc.) and
- the weak points of the body (with an intention of self-defense),
- How to sand and paint,
- the perils of household fluids, e.g., chlorine bleach, alcohol, gasoline, powdered cleaners, etc. and their mixing,
- My least favorite: how to dig and fashion soil with spade, shovel, hoe and rake.
- How to throw balls, rocks, a handful of dirt, etc. Spear/bow/arrow is fantastic but probably best left to organized instruction. Emphasize safety (for the balls and rocks too!8-))
By then he should be old enough to set his own path.
We don't even need machines that can answer any question in a specific topic area. But we could use machines that can answer "I don't know." when they cannot compute the answer.
Consider the possibility of much-smaller LLMs that answer questions only in limited domains (physics, chemistry, math, medicine, etc.)?
Could not such LLMs exist? If not, why not? Would they not likely be more useful and far, far easier and cheaper to create and run (possibly even with far less hardware) than the current "know-all" LLMs? Also might they possibly avoid certain problems ( e.g., hallucinations, deceit, apparently undesirable behavior, etc.) associated with the "know-all" LLMs we now have?