not artificial, barely intelligent
sounds like my biography but it's not
humans have peaked.
intelligence has been commoditized. earth is now fully connected. we’re venturing into outer space. but the average person asks: will ai take my job? fair enough. the non-techie highbrow faction of silicon valley has the perfectly stoic answer.
stay calm and carry on. we shall see ads!
what comes of ai?
more ads. new ways to cull human attention span now that we’ve got another online oracle since google1. a lot of automation that increases consumer demand across markets2. new white-collar jobs as usual to operate at the ever-ascending levels of automation and abstraction.
we may not see an ai apocalypse. no cart-to-car type disruptive automation. and nothing that requires ubi (universal basic income). that’s the commentary.
at the current rate, ai technology built on transformers (large language models) seem to be increasingly capable at automating sophisticated jobs including creative, high-iq work like proving math conjectures or composing music. agentic ai is currently capable of automating pretty much every software development job. it feels like software engineers dug their own grave.
there are a few big-name researchers working on building real-world models (as opposed to large language models) to create human-level intelligence or artificial general intelligence. large language models are capable of compressing information into patterns found in large corpus of text. general intelligence is essentially compression. a baby who recognizes a pattern of attention when wailing or cooing is compressing different examples of being fed, coddled, spoken to, or touched. and llms are doing well at compression. why is that not enough?
bipedalism made apes intelligent
freeing up two limbs that previously supported four-legged locomotion is considered to be the trigger for developing brain areas responsible for higher-order thinking related to pattern recognition, abstraction, spatial reasoning etc. maneuvering hands to use a weapon or rotating objects demanded more data center space in the brain, just the same way ai industry is demanding brick-and-mortar data centers today.
the increased demand for compute capacity in bipedal brain required more physical space (bigger brains) and higher energy (brain consumes 20% of our energy). language processing abilities alone wouldn’t get artificial neural networks to advanced intelligence. but no one knows what that would look like although the general idea is to build “real-world models” to mimic human learning.
is ai disruptive to human well-being?
we don’t know. and we can’t know.
alan kay, the computer engineer at the erstwhile xerox parc that inspired much of what apple built once remarked that the best way to predict future is to invent it. current prediction of ai related job losses are a good linear projection, but useless. causal reasoning cannot predict disruptive changes. no one could have predicted that steam engine or electricity could produce this much wealth via industrialization saving most of the world from abject poverty. black swan events and their impacts are never linear or predictable. the outcomes only make sense in hindsight, never in projection.
every step along human progress leads to higher abstractions of doing things and better levels of comfort. we no longer hunt, forage, or have to pray for health. there’s always food, shelter, and healthcare. our office jobs exist simply to solve contrived problems of civilizational splurge from an evolutionary perspective. but progress has a price. it accentuates our inherent faculties - the intellectually fittest survive and the rest suffer or perish in the absence of faculties to adapt to progress. so democratization of intelligence through ai might actually only help the fittest despite being available to everyone.
intelligence is the ability to get what you want out of life given the constraints.
that’s the idea postulated by cognitive scientists like pinker and aphorists like naval ravikant. if you can’t get what you want out of life, not necessarily in the exact way you planned but perhaps in some concoction of serendipity, resolve, and skill that makes life livable, you’re probably not going to make it anyway.
evolution doesn’t care for kindness. the fittest to survive survives.
ads essentially paved the way for every other technological revolution since the internet. cloud, big data, ai and every other trend in between that’s less popular to the non-tech crowd mostly came out of google in the last 25 years. and what powers google? ads! google engineered the online infrastructure behind farming human attention and became unbelievably profitable to fund and create almost every innovation that underpins modern software development. unlike most companies that succeed and then become bureaucratic, google is known to hire with a maniacal obsession to avoid false positives (bad hires) at the risk of false negatives (rejecting extremely good people). well that and jeff fricken dean.
jevon’s paradox: when a technology or automation makes something easier and cheaper to do, the demand for it increases (as opposed to automation destroying the economy and jobs like luddites think). higher demand creates more jobs in high-paying skilled trades, and that percolates into demand for lower-level services that require more maids, truckers, nurses, mechanics and so on.