I'm comfortable saying 2050, but that's if the federal government starts providing regulation and guidance very soon.
This. It takes two to tango: acceptably bulletproof AI
*and* a regulatory environment ready to pivot for both the product (car -side) and "application" (road -side) interfaces. And then on top of that there's the fact that readiness for widespread adoption has to conform to very traditional means of manufacturing & supply-chain scale-ups in reaction to both maturing tech and maturing regulations.
"Disruptive innovators" too often count on the tech revolution steamrolling straight into the regulatory revolution and supply-chain revolution, when really those pieces are entirely out of the hands of the technology. Government is as government does...move slowly and in political zigzags rather than straightforward practical moves. Whether we are or are not in an especially bad era for government's stimulus-response reaction to regulatory change is kind of moot; on a sliding scale of better to worse the inherently inefficient political reaction from the regulatory state is a rote-standard feature in every First/Second -world (or even Developing-Third World) region on the planet. Disruptive innovation never ever pushes the regulators from the source...despite MANY self-claims of the innovation being badass enough to have exactly that power. If that were truly possible some of the more regulatory-efficient regions on the planet would've gone all-electric cars a solid 20 years ago when the raw technology was
first ready for mass adoption...instead of still sitting at/under that cusp in 2020. For that reason alone the 2035/15-year target is way, way overoptimistic for tech that fundamentally changes the driving "interface"--so heavily regulated across the board--at way more disruptive a level than electric power plants ever did/will. Immediate self-check is needed if one finds themselves starting to buy that hype that regulatory change can be pushed at tandem pace with technological change.
Then there's the manufacturing side. It's one thing to have the disruptive underlying tech refined into production-ready state...it's another thing entirely to get the production scale implanted. You're dealing with global supply chains that are a mismash of new & old. You're dealing (esp. with cars) with a lot of certified craft labor. You have to ensure an aftermarket supply chain for maintainability outside of the lock-down control of the manufacturer for up to 20 years/100,000 miles. So whether you rebrand your facilities as "gigafactories" or not, the scale-up to mass market supply has to play a very traditional game with a lot of heterogeneous moving parts and labor. Bleeding-edge tech in the auto industry is already getting thrown for a loop on this with the unintended consequences of monolithic computer firmware updates inducing fit/finish problems with heterogeneous components. It's not enough to simply reclassify the car brain as the equivalent of updading an Android phone; Tesla's tried that customer service tact and gotten loads of criticism about that is NOT an okay equivalence when the car handles differently to the driver after a firmware update. Self-driving's going to have an even bigger hurdle to clear with those unintended consequences of behavioral changes by-update. If systems integration writ-large is still running a tad (not bigly, but somewhat) behind schedule getting on the same page with scale targets despite a whole lot of "disruptive" pressuring...how exactly does that gestation period get
further shortened through several more degrees of "disruptive" pressuring? Movement across the spectrum to mass-market production paydirt seems to be the logistically immovable (or far less movable) object here setting the overall pace.
So duly noted: AI tech advancement is a non-linear endeavour. That still doesn't explain where if/when the next great leap does come, the
production-readiness is going to leap-and-bound in tandem let alone push the inefficient-by-design worldwide regulators to adapt quickly. That does not mean improvements to any one of those realms are for naught. It just means "disruptive innovation" doesn't jump whole-realms easily for a whole lot of structural reasons that are neither good nor bad...but just 'are'. There's an over-tendency within the innovation realm to see that as an indignity that shouldn't be if only 'they' got the same innovation spirit...but at the end of the day that's irrelevant. It's not
attitude that makes advances go mass at the speed they do...it's writ-large realm vs. writ-large realm vs. writ-large realm gravitational pull. When regulation and supply-chaining are such extremely large gravitational wells...how they move in relation to each other is basically little more than macro-physics. You can't speed them up by reasoning with them; they just
are, and are just masses reacting to masses. We haven't failed if it takes to >2050 instead of that too-aggressive-for-credulity 2035 for this to happen; that's just the pace the masses end up reacting.
