[graypegg]

> It also appears that the common goal for many leaders in the space is to then deploy and spin up these “beings” dynamically for use in a broad range of conditions, with the vast majority of cases being repetitive labour.

Isn’t AGI considered overkill for repetitive labour? Maybe this is a different point, but for anything repetitive, something deterministic is ideal isn’t it? Why would I want to pay for automated labour with a conscience when I could get cheaper purpose built machinery?

I guess we could suppose a future where everyone gets the same AGI so all you have to do is describe what you want it to accomplish… but I want to attach a combine to it since that would be the easiest way to harvest this field. So am I now trying to explain to an AGI how to interface with a combine? Do I have to track down modules to install that people already trained on my specific combine? None of this saved any time vs just buying an automated combine. The fact that it can come off the field and toast a slice of bread or write a poem about its feelings is kind of irrelevant.

AGI fits nicely into fiction because it gives the machines a voice and soul, but I don’t think those are desirable qualities in an automation solution.

[postultimate]

1974: "No-one would put an AGI in a bomb, that's just ridiculous"

1986: "No-one is going to put an 80386 in a vending machine, that's just ridiculous"

[graypegg]

The vending machine can dial up the distribution facility to dispatch someone to come refill it. The distribution company is happy about consistant ordering. The business with the vending machine is happy they don’t have to do the call anymore.

We’re supposing that a conscious being makes significant advances in use cases where standard automation hardware+software is applied today.

I’m not saying ML isn’t a major shift, but we’re talking about AGI, I don’t think any use case exists unless it’s specifically meant to wow fleshy human beings. A trained ML model in the domain you’re working in, with out the pesky conscious, seems like it’s the boring efficient end state for automation.

[postultimate]

No, that's not what |I'm supposing. The 80386 got used in vending machines because technology moves on, the new stuff gets cheap, and because it's more generally useful, it gains in popularity while the old stuff loses it, even in cases where the old stuff is adequate. The same process will probably happen for AGI - unless non-AGI has capabilities that AGI can't replicate, AGI will probably replace it.

[graypegg]

Well I guess I’m getting at the fact that an AGI is trained mostly on things that ISNT its current task. Is there any reason why I couldn’t train another ML model only on the scope I care about and get better results? No conscious required. Doesn’t even remotely resemble a human being because all it does it control the combine, which is the only thing making me money.

[samr71]

Putting AGI in a bomb may work. You'd have an incredibly clever and motivated targeting system. You tell it where to go and it would be off to races, eager to please!

The Japanese tried this with plain-ol' non-artificial intelligence, and it seemed to have been somewhat effective, if not the most sustainable.

[graypegg]

It’s hard to tell if you’re being sarcastic, but I don’t think the launch procedure is currently the main bottle-neck for launching bombs.

Why would I want the bomb to think? I want the bomb to follow some pretty specific instructions.

[webmaven]

> Why would I want the bomb to think? I want the bomb to follow some pretty specific instructions.

It probably hinges on how rapidly the field of battle is evolving.

Do you want your bomb to at least attempt to adapt in realtime to novel targets? To novel means of camouflage, defense, and even interdiction (rather than wait for reports in the field to eventually prompt a software patch to upgrade the weapons system)? Well, the bomb is going to have to be considerably smarter to do so.

Of course, there are just as many ways that sort of on-board capability can go awry. I imagine that painting noncombatant symbols (or the equivalent adversarial input) on combat vehicle roofs may serve to fool overly smart munitions for a short while, for example.

[graypegg]

That’s a fair point. I think I’m maybe a bit stuck on AGI specifically. As in, indistinguishable from a human being and truly general. You can ask it in natural language to do any computational task.

I don’t know why you’d prefer a “human being” to control the bomb, when most of our technological history in this space is trying to remove humans from the equation considering how many errors they make.

In a true AGI, one advanced enough to make the original poster concerned about the morals of enslaving it, MOST of it’s training is on things unrelated to being a successful bomb.