Very happy the flight worked as planned! The more people, and companies, in space the better.

They're already testing AI fighter jets. Why is an AI dog-with-gun so beyond the pale?

VASTLY less dangers of friendly fire in the areal space, super easy to isolate valid targets. *And* you would only activate that in a hot war Zoe with zero other valid air presence for hundreds of miles. And also there the ability to withstand high-gs matters, and also there in fact time does matter in a way that simply does not on the ground, and never will.

The robot dog can and will be used in mixed civilian populations.

"The human is too slow and sometimes makes mistakes. Let's put an AI in control for the next version".

Nope. That is what I am telling you is a GIANT step the military is unwilling to take. Not a giant step technically, but a giant step in terms of effect and possible negative consequences.

We have proof, because this robot dog exists and it does not work that way, even though it obviously could have.

And what does "human is too slow" even mean? The robot itself can only move so fast, the decision time of the human to fire is already near-instant. Because the robot dog is in harms away and not a human soldier, you actually do not have AS MUCH of a concern about time, because once the dog sees a target it does not matter if that dog is destroyed. At that point the soldiers nearby know there is a hostile, know where it is positioned, and can leisurely pluck it/them out.

Time is the least point of concern where the robot dogs are concerned.

I think in a lot of ways the government is starting to lean away from such contracts having been burned too often recently,

Maybe, but SpaceX won the other portion of that fixed bid (ironically a smaller portion than Boeing) and they delivered on time and made money...

I think you just can't expect a company structured to work via "infinite bid" to ever switch to a fixed bid format.

That said companies like ULA were never under any real competitive pressures before and they are clearly struggling adapting to that,

I read a long article on this issue (maybe one linked to in the summary) and they mentioned a pretty huge issue was this is a fixed bid contract, which Boeing had not done before for space stuff... them having to eat cost overruns and fighting with contractors has taken a lot of willpower away from getting it finished. Every cent more they spend is a cent lost.

If you look into details, it has a human controlling where the dog goes, and if/what it shoots.

Yeah maybe it has an aimbot to help hit a marked target, but I see nothing wrong with making a remote extension of a soldiers will. You can imagine saving a lot of guys from getting hurt with remote robot dogs, like for urban building clearing. Is it really any different than other kinds of drones with remote feeds, which are used today?

It's only when you hit the point where it's not a human deciding to fire that things get even questionable at all to my mind. But we are not at all close to that point because the military is unwilling to go there anytime soon.

Anytime you see someone use the phrase "Late stage capitalism" you can be sure they have no idea what capitalism even is, because they think government-controlled and supported corporatism is capitalism.

Time to expand my posting of AI-poisoning bafflegab to Reddit. Off I go...

The reason why Reddit is such a perfect source of AI training, is that it's largely immune from single person attacks like this.

The reason is Reddits high level of segmentation, and ease of voting.

If you go into some tiny community and post nonsense, the AI probably will not even notice your stuff because it's related to a very out of the way subject.

In any community you post nonsense in, you'll be voted down hard and AI won't train on you, except to recognize failure or as a counterexample - and that downvoting will be more rapid the more members a community has.

Reddit is by far, the absolute best source of data for broad-ranging AI training.

I tried to point out the value of this to everyone pre-IPO, but all of Slashdot po-poohed the idea, apparently never having used Reddit or comprehending what it is.

However, if you earn enough gold and karma, you may be eligible to receive a payout!

Well that sounds kind of promising! It will probably be some very large amount, but maybe possible... just being helpful and/or funny on Reddit seems to get you quite a bit of Karma.

I can't see myself buying Gold on Reddit (why buy that would I can buy real Gold!).

However, if I am awarded gold for my many insightful and helpful posts, will I be able to convert that back into money I could use to buy real Gold?

Part of the reason the thing is huge you have to think is something to do with required range abilities, maybe mandating something like a 100 ft range.

That's a tremendous amount of power because of how rapidly electrical energy drops at a distance.

But what if you could guarantee being about four feet away before activating, without a need to aim? Then maybe the device could be small enough to be mounted on a drone that used AI and cameras to get as close as possible to a marked target before going off. It would probably ice the drone as well, but if it stopped the bike/scooter - mission accomplished.

A large rock can substitute as a 'hammer' in a pinch. Fashioning a recognizable hammer from a stick, scraps of animal hide and a stone is not difficult.

Ask me how I know you've hardly ever used a hammer. I've framed whole basements....

Your stone-age tools make OK weapons, terrible "hammers" even in a pinch. Stone is an awful medium to use for that purpose.

But even if I gave you that a stone hammer was even really a hammer at all, even then there STILL are more people alive right now that could build you an AI than could properly strap together sticks/animal hide/rock to make a working hammer that would survive one swing.

You are just dead wrong on this one buddy.

This is my last post as you just aren't even making enough sense to respond to.

Thanks, that is what I remember seeing!

We were thinking of using Unity in a non-game app and that fee structure put it right out of contention.

I could have sworn I looked a few months ago and there was still a significant runtime fee.

Not really, hammers are trivial to make, citation: they've existed for thousands of years

AI has not existed for thousands of years, but it is now equally trivial to make custom AIs, very powerful AI models are now running on Arduino.

In fact I would wager that MORE people currently have the skill AND EQUIPMENT needed to build an AI, than do have the metalworking skill needed to make a hammer!

therefore one can logically assume it is not trivial to make

WTF? Are you even a developer?