Thinking about AI
It is very clear to me that people don't know beans about what they want.
People, even very intelligent people, are rarely self-aware and are not capable of articulating an explicit hierarchy about their desires. They haven't thought about the scales (ratio-of-trade between attributes) of what they want and the relationships between those desires.
The videos on Hoe-Math and the videos where the narrator asks women what they require in a romantic partner demonstrate that.
A typical vignette of the second type of video has the narrator approaching an attractive woman, typically outside of an upscale bar, and asking her "What do you look for in a boy-friend?"
The answer typically comes back "He has to be between 25-and-30 years-old. He has to be making at least $150k per-year in finance or marketing. When we go out on a date (at least once a week), he must be happy to drop $250 for my half of the tab. He has to take me on at least four one-week, over-seas vacations a year. He must be fluent in French and have a swanky apartment. He has to be at least 6'-2" and have thick, wavy hair and have perfect teeth and manicured nails. Oh...and most importantly, whenever I call him or text him (any time of day or night), he must drop everything and call me back for a 30 minute phone-call where he pays rapt attention to events that have absolutely no interest to him."
She isn't describing a human being. She is describing a unicorn.
Guys might be slightly better, but not by much. While our needs in terms of a romantic partner might be simpler, there are other fields where our needs are more complicated.
Why this is important
AI answers questions. If you are unable to articulate a meaningful question, the responses will be equally not-useful.
How we got here
Think back to a pre-civilization village. A very large one might be 120 people. Books did not exist. Knowledge was passed down by observing your elders and your neighbor and copying what they did. Footwear is a good example. What did they make their sandals or mukluks from? How did they cut the hides or braid the straw? Fur in or fur out? How did they stitch them together? How did they care for them so they didn't rot? At what point did they start making a new pair?
Many of our behaviors are patterned on the "copy our peers" thinking. Picking a mate was a trivial exercise when there were only three available candidates and two of them were a day's walk away.
To be overly-flowery in my language, "copying our peers" is a form of "cloud-sourcing the logic" or "wisdom of the crowd".
Weaknesses of the follow-the-crowd model
First, a snarky observation: You don't need AI if you are going to rely on social media to tell you what you want.
Other weaknesses:
Computers will not make assumptions for you. If you gave that very long list that included $150k/yr to a computer, it will not automatically filter out candidates who beat women, are addicted to cocaine, have Monkey-pox and other sexually transmitted diseases, are compulsive liars, make a living denying insurance benefits to widows and orphans or are $1.2 million debt and have no assets.
Many of the things that we think "are a given" are not. Computers deliver a geeky karma of "Be careful what you ask for, because you are likely to get it."
A possible employment niche in the post-AI world
A consultant who does scenario-testing with the client to solidify specifications. Going back to our romantic-partner example because it is something nearly everybody has experienced:
"Suppose you and your new romantic-partner are dropped onto a deserted island or a remote, wilderness paradise. What attributes are you looking for?"
"Suppose that you had to relocate to a city where there was no internet or that it was so slow and unreliable that it was effectively unusable; What attributes would you value most in your new romantic-partner?"
Both of those scenarios attempts to fire-wall off the "performing for the applause of the audience" grand-standing.
"Imagine you and your new romantic-partner together five years from now, would there be any additional requirements? Would any of your current requirements have softened or become less important? Which ones?"
"Now imagine you are together ten year from now..."
"How about twenty and forty years from now..."
Then... trade-off studies:
"Do you prefer Jeff who makes $150k and is 6'-4" or Mike who makes $175k and is 6'-2" tall?"
"Do you prefer Sean who makes $200k a year but can only take two weeks of vacation or do you prefer Ian who makes $150k a year but is allowed four weeks of vacation a year?"
"Do you prefer Prakash who is a surgeon making $200k who cannot take phone-calls while performing surgery and must enforce strict sleep-hygiene (i.e. no phone calls during his sleep hours) or do you prefer Jimmy who makes $150k and drives a hazardous material truck and can talk you on the phone (but not text) any time?"
How porous is the barrier between "Allowable" and "Not allowed"?
"So...we
found a potential romance-partner who meets all of your criteria except
for one. He lives 10 minute travel-time from where you live. The
closest candidate who meets all of your criteria lives 90 minutes away.
The flawed candidate is 35 years old...would you date him?"
Of course she would date him. She pulled the "25-to-30 years-old" out of thin air. There was nothing magical about that range except she thought there should be an age requirement.
This
deficiency comes up over-and-over through life. Which house should you
bid on? Which job should you interview for? Relocate or stay? One more
kid or not?
Incredibly, most people have never given a thought to the fact that having a lot of one thing (like income and status) means that there will be much less of something else (free time and flexibility). Critical questions are "How much of each is ENOUGH". Do you value lots of time with your romance-interest and eating balony sandwiches or do you value infrequent times with your romance-interest and Instagram worthy selfies more?
All of this scenario testing is really a "Sims game" but it is exploring the client's real-life desires and expectations. Bonus thought: The person who can write a Sims-like front-end to make a game that combs out the clients real requirements (and not the generic, cloud-sourced ones) will make a lot of money.
Change of topic: Update on the last part of yesterday's work-ticket
The last part of yesterday's work was a piece of cake.
I watched Quicksilver so Southern Belle and Handsome Hombre could assemble the pool.
Then I made a parts-run to Lansing for some odds-and-ends needed for the assembly.
Then I watched Quicksilver some more.
I tapped-out at 7:30 p.m. I was tired and dehydrated. The young people were still hammering away at the project.
