Sounds like they fucked up training the AI then. For a user it doesn’t matter whether the AI is designed poorly or trained poorly, it’s behaving poorly.

On one had we’ve got links to the department of energy and to Wikipedia. And just some hand waving on the other hand.

If we’re talking approaching fundamental limits, Hydrogen fuel cell is not a great comparison. A high pressure tank can only get so light, even with linerless ultra high strength carbon fiber pressure vessels, the mass of the vessel is maybe 6-10x the mass of the hydrogen it carries. To increase specific energy there you need to go to cryogenics which is a whole technology leap and has its own set of challenges.

Battery tech has been improving more than you have seen, clearly. Since '08, lithium batteries have increased energy density by 8x (https://www.energy.gov/eere/vehicles/articles/fotw-1234-april-18-2022-volumetric-energy-density-lithium-ion-batteries). The best LiPo batteries are around 0.9MJ/kg right now, but there’s no fundamental reason a battery couldn’t achieve 9MJ/kg. Lithium-air batteries could theoretically achieve way higher energy density than that even (https://en.m.wikipedia.org/wiki/Lithium%E2%80%93air_battery), and have already been demonstrated in a lab to achieve more than 5x what current commercial automotive batteries are doing.

I think the point that is counter to yours is that we are nowhere near the fundamental limits of energy density for batteries. It’s probable we are near a fundamental limit for LiPo, but the point is that battery tech improves by changing technologies/chemistries. BEVs couldn’t exist at all when the best rechargeable battery tech was lead-acid, but were enabled by LiPo. Theres most likely a type of battery you can’t even imagine that has yet to be invented that could store >10x or more energy than current LiPo per unit cost or mass.