Thoughts on, “Should we really support AI so much?”
As promised, my thoughts and quick hits on your post, “Should we really support AI so much?” Here:
My stream of consciousness replies follows your outline …
ChatGPT in Nov 2022! Yep, like you, that is when I first played with it. So cool that you trained Gemini in Serbian! I am a Grokker nowadays. I use it to critique what I create, but not to create.
So you are using AI for content creation… perhaps also search and text editing, like many, and me. Recently I read that may not be where AI’s real value is. Rather …
The higher economic benefit from AI should be in deploying “agents” to make workflows more efficient and automate workflows. A case study I read this morning: Alpha School. AI agents are doing the donkey work of grading repetitive answer sheets. As in, do we need a human to check all 20 students’ papers if they spelt “dog” and “cat” correctly? The teachers, freed from that task, upgrade their time to coach human-to-small-human how to learn, now that tasks that are really not value add are delegated to machines.
You may be agreeing with some Warren Buffet wisdom with your take here,
“AI will create more billionaires in the next five years than were made since the emergence of the Internet… but who will they be? I am guessing people who are already successful business owners who just need to move them into the new AI architecture… All this reminds me of the Gold Rush in California; diggers rarely got rich, but those who sold spades and other indispensable items did.”
Yes! I saw on old video where Buffet noted that all the US railroads from the boom in 1880s went out of business and did not make net money. Ditto almost all US automakers, from the hundreds in business in the early 1900s and including the three auto makers in the Dow Jones Industrial average in the 1920s (from memory, perhaps Packard, Studebaker, and third, but not General Motors, Ford, or Chrysler). And same with airplane manufacturers and airlines! All capital consumers and losers at the end of the day. Who had NVIDIA on their bingo card a decade ago? The hypotheses being:
Proven businesses should get more profitable by productively incorporating AI agents into their proven business models which make profits today without AI.
We don’t even know if the AI model companies (e.g., OpenAI) will be the winners, versus those who control power and chips.
Given so many competing, plus the observation that winner takes all (history does not repeat yet rhymes), I am keeping my money invested in broad mutual funds (e.g., tracking the S&P 500) rather than stock picking. I am betting for society to win, versus discerning which one of thousand of companies will be King Kong.
Meanwhile, yes I think about:
“the apocalyptic … where humans end up stuck with artificial intelligence that maltreats them”
“To what extent will it replace [folks]? What jobs will my kids have in the future?”
True this:
“tech is going to obliterate some jobs very fast, decreasing the need for certain professionals.”
The questions are, “What to do?” It comes in two flavors.
As an individual / folks in my family unit.
As a society.
N.B. I am recently retired so I can share my crystallized intelligence yet don’t have as much skin in the game.
As an individual, without putting too fine a point on it, younger folks could consider perhaps hedging bets this way:
Career approach A: Learn how to deploy agents in your current profession. Domain knowledge matters! The lowest cost and highest return AI is reducing drudgery using pedestrian agents in currently valuable businesses! Coders who don’t know the business application make tools that dead end. A person need not become an expert AI application engineer, just be able to put one to work.
Career approach B: Focus on a career less likely to be replaced by AI, such as the trades: welding, electrician, etc.
And oh yes, bet on society winning. Invest broad base in being an owner. Use your labor to build ownership, as your labor does not scale and has limited life (literally!). How to invest easiest and smartest in broad based ownership? See the aforementioned S&P 500 mutual fund example.
As for society…
Legislation… hmmm. I am not seeing how legislation protects, really. It is always too late with unintended consequences. An example of pattern recognition: Nuclear. There were less plants built in the US, and at astronomically rising unit costs, after the U.S. Nuclear Regulatory Commission (NRC) was created as by Congress in 1974 to regulate civil nuclear power plants. For technical endeavors, legislation and government intervention tends to slow down societal benefits. See the contrary example of the speed and efficiency of SpaceX.
Universal Basic Income (UBI). You hit the nail on the head that there will be lots of displaced workers. Some may be my children! I am sympathetic to “… strive to ensure that people, whose jobs became redundant due to the AI have some sort of UBI during the period when they are transitioning from one profession to the next.” It is not only a humane impulse, it may be unavoidable. I pray the incentives are such to minimize waste and fraud. I hope the job creation outpaces the job destruction. Even at that, many will be improperly skilled for what is coming. Pattern recognition: coal miners in the UK and the US. Again, what can an individual do? Folks, especially younger ones in or entering the workforce, it may be wise to take agency and upskill.
Maria, thank you for the opportunity to ponder these topics and comment. Like you, I worry and I think about these things. Taking action though is where future-proofing comes in.
If I were serious about walking the talk, I would take a goal in 2026 to personally commission the successful deployment of at least one AI agent. Not that I have to do it myself, just get it done for effect.
Will you hold me to that goal? Check back at the end of 2026! I don’t do it, I wire an UberEats meal to you and yours in Serbia or a bottle of champagne, a penalty for my inaction. If I do deploy an agent, you will get the same. Why? I will have won a lot more!
BTW, this is what I was tied up with earlier: https://open.substack.com/pub/ibvonnet/p/a-non-approach-to-promote?utm_campaign=post-expanded-share&utm_medium=web
This is very well written, with good advice as well!
Thanks for taking the time to answer my questions. I think your answers are insightful.