Big if true but I'm super skeptical.

First of all how many years have we been hearing about how we'll have fully autonomous vehicles just around the corner? At least a decade.

How many years have we been hearing that human-like AI is just around the corner? At least since the 60's.

I'm skeptical (although optimistic) about massive tech breakthroughs in general. Even more so when it comes to military tech since so many in the mil. industrial complex stand to benefit from the trillions that the DoD sloughs off every few years for the next big thing.

Not an aviator but spent almost a decade in the military. You used to hear all the time about the latest and greatest tech that was going to change the way we fight.

At least in my nearly 10 years nothing really changed in any substantial way. incremental improvements, sure but no massive breakthroughs like are continually trumpeted in popular press.

I hope you're right though!

We won't have general human-comparable AI or autonomous driving (or we will briefly until a few tragic things happen and people come to their senses). Enthusiasts think we will, but people underestimate what real intelligence requires and confuse it with training complex statistical models.

They call it AI, but the "intelligence" there is just a name.

We don't need a human-like AI to beat humans in the air. I can't imagine that the US doesn't already have a fully autonomous aircraft, and a virtual fully autonomous fighter jet waiting for the physical jet to be built, if that hasn't happened already. In any case, we wont be hearing about it for a while. It took seven years from first flight for the F117 to become public.

No matter when it happens, I think that when General AI shows up most of us will be shocked and surprised.

General AI showing up will be kind of like the discovery of quantum mechanics and the nuclear bomb. No matter how much speculative fiction we write about it, we won't be able to predict ahead of time how transformative it will be on society once it actually exists.

not even drones?

I worked in stress analysis in the civilian aircraft world and it was obvious to me that probably 85% of my job could be automated away, and most of what remained could be shifted to the design side if the right software were written and design processes were revised around it.

But the right software would cost a lot of money, and since it would be new, it would need customer buy-in up front (aerospace is in some ways very risk averse when it comes to adopting new workflow tech).

I think there's a revolution waiting to happen in bringing down aerospace design cost and schedule and increasing the ability to iterate on designs. I'm hoping NGAD is the sign of something big that will trickle down soon to the civilian aerospace world.

Building the hardware is easy, it's the systems and software that will kill it, just like the F-35.

Until we break new military developments away from the political process of congress people funneling projects to big companies in their districts regardless of competence we will continue to see failure after failure in every area.

We saw many advances during world War I and world War II because instead of rewarding companies with contracts that voted for politicians, all everyone cared about was defeating the enemy.

My suggestion is that the military should put out specs for what it wants and then put out big cash prizes for getting it done similar to what was done with the Covid vaccine. Say a $1 billion prize for a modern tank design. And then a fixed price for each one delivered thereafter.

Roper is awesome. There’s a video of him describing the role of K8s in the next-gen process.

We are doomed.

Avionics software resembled k8s before k8s was designed. Sure, some of the most problematic bits don't exist due to static assignment, but the difference is not that big. Including presence of virtual overlay networks.

Anyway, k8s is mostly for lower-criticality software to be quickly added/removed, iirc, partly because unlike the normal avionics setup it allows dynamic addition/removal.

I had not seen this, thanks.