UNSC's AI: What are their capabilities?

Community Halo Universe
1

I have been thinking about them more recently and the more I think about them, their capabilities and what they could do I realized if they could do what I think they should be able to, then it would needlessly bog the universe down, and it would not be good to rely heavily on AI’s anyway or else that’s just a recipe of your own destruction.

Anyway, this is what I was thinking - Shouldn’t AI’s be or atleast capable of extreme innovation?

Here’s what I’m thinking; they are capable of billions, if not trillions of processes a second are they not? And they are probably capable of making perfect or near perfect simulations of physical simulations of our universe (I am meaning in the sense of experimenting with different properties, actions and reactions and the most probably outcome that would occur).

If they are capable of that, then wouldn’t they be able to innovate technologies by leaps and bounds, far faster than any physical being?

For instance, ultra smart AI’s like Cortana and BB, after they’ve learned all of humanities history and started their AI life out, shouldn’t/couldn’t they look at the slipspace drives, render simulations and eventually come up with a more efficient drive after a while?

As it stands now, all AI’s are really for are assistants which should have way higher IQ’s than their human counterparts they work with. I’m not trying to belittle everyone in the Halo Universe, but I could see the UNSC effectively using the AI’s to do R&D and keep humanity up to pace without us falling into pieces without them.

I can even see them being super effective in ship battles. Just load an AI in a specially designed projectile, fire it off in a rail gun and have it hack Covenant systems over their COMS. It would force them to either shut off their COMS completely in battle and make them less organized or they would be forced to deal with multiple system intrusions, errors and shutdowns during combat.

2

Come to think of it, you’re right, but I suppose that AI’s are pretty hard to make.

3

> Come to think of it, you’re right, but I suppose that AI’s are pretty hard to make.

They are yes, the good ones require a dead genius to make. But If the AI’s could be that effective, they could ramp up production.

And my idea of blasting an AI at the Covenant ships wouldn’t be the death of the AI. They would just have to figure out how to make a good combat intrusion AI that can make a bunch of fragments of itself without severely hurting it.

4

Anyone else have any ideas on this?

5

There are still the dangers of AIs.

If they give AIs too much freedom, who knows what will happen. They could turn on you, malfunction, vent atmosphere, etc.

Look at Roland. “Undid Iridium” was a passcode that completely overrid his systems, allowing a [war criminal] to access classified information.

You don’t want to give AIs too much power if there is such a risk of corruption, rampancy, malfunctioning, etc.

6

> <mark>Look at Roland. “Undid Iridium” was a passcode that completely overrid his systems, allowing a [war criminal] to access classified information.</mark>

I only disagree with you there since Halsey (if my memory serves me well) was pretty much the founder and designer of modern AIs. Since she designed the code that AI’s use, she purposefully added a few extra lines that only she would know in case such a situation was right.

But you are totally right about the risks they pose, look at the forerunner AI Menediact Bias for example.

As far as innovation goes, I think it would be very hard for AIs to perform a ‘human’ task. They will certainly dominate any human mind when it comes to figuring out the physics of the next stage of slip-space drive, however the problem is AIs are meant to be ‘tools’.

As Bob said, they are assistants currently in the Human world whose job is to assist humans. The reason is simple, machines will always treat a situation as a process with a start, a finish and the events in between to get a solution.
If you place an AI-controlled ship against a human-controlled ship, then it is very likely that the AI will have the advantage ‘normally’. However what a AI cannot do, which a human can, is random.
In conventional warfare, an AI would keep their ship at optimum range from the enemy ship to give them the winning advantage. However the humans may use an act of sacrifice to use the ship to ram the AI ship, which the AI would not expect since they are programmed to think of it in one way.

Enough of my babbling, the point you are talking about however almost happened recently … however it ended with the ‘apparent’ demise of the AI Cortana. She is the only AI in human history to have been cloned from a living brain (Dr.Halsey) which theoretically means she can think like a human but with the super computing power of a smart AI.

Throughout the Halo series, it is quite evident that she is somewhat different from the usual smart/dumb AIs. With her encounter with the grave mind for instance, most AIs would not be able to resist the corruption of the Flood for as long as she had and she only just managed it.
Also during the mission where you are escaping the collapse of the core of Requiem, she does do some initiative thinking by using the power of Chief’s shields to boost the Ghost’s boost function.

Cortana was an AI capable of what you are talking about, however it is highly unlikely we will see another model like her for quite some time.

7

>

> As far as innovation goes, I think it would be very hard for AIs to perform a ‘human’ task. They will certainly dominate any human mind when it comes to figuring out the physics of the next stage of slip-space drive, however the problem is AIs are meant to be ‘tools’.

I’m not so sure really on their inability to preform ‘human’ tasks. I think it may just be how they were programmed possibly. As far as I’ve seen most smart AI’s seem very human, and in most cases, they are probably mostly occupied with running the ship’s they’re assigned to. After seeing Roland have difficulty on the Infinity, it almost seems like a too big of a task for him to do alone even.

> If you place an AI-controlled ship against a human-controlled ship, then it is very likely that the AI will have the advantage ‘normally’. However what a AI cannot do, which a human can, is random.
> In conventional warfare, an AI would keep their ship at optimum range from the enemy ship to give them the winning advantage. However the humans may use an act of sacrifice to use the ship to ram the AI ship, which the AI would not expect since they are programmed to think of it in one way.

Once again I’m not so sure on this. Any smart AI I think would be able to calculate scenarios which would be most likely to happen, best outcomes, most efficient means of victory, which would include sacrifice of assets to win.

Part of the problem would likely lay in potential predictability, but if we’re talking about using an AI that was designed for warfare or atleast has a decent affinity to it, I doubt they would use generic tactics considering they have a good couple thousand years of warfare history to go on.

All these AI’s are modeled after humans, used their brains for conception, and they too have emotions and personalities and I believe capable of a lot.

Part of what I think made Cortana special aside from being made from a living human brain from Halsey is that she was specifically designed to be used with the Spartans for covert ops. Her main functions weren’t ship-based but rather highly dangerous ground based operations, and she had a more intuitive civilian aspect to her as opposed to the military based AI’s.

8

> There are still the dangers of AIs.
>
> If they give AIs too much freedom, who knows what will happen. They could turn on you, malfunction, vent atmosphere, etc.
>
> Look at Roland. “Undid Iridium” was a passcode that completely overrid his systems, allowing a [war criminal] to access classified information.
>
> You don’t want to give AIs too much power if there is such a risk of corruption, rampancy, malfunctioning, etc.

You are very right on that how dangerous AI’s can be, human AI’s have already been meddling in our affairs behind the scenes for a while now, I forget what the group is called but they were around prior to 2500’s I think, I believe they kind of kick-started the ORION project.

But I already was thinking about the dangers and benefits of this, just didn’t bother posting them until I got some reception on this.

What I was thinking would be to have AI’s specifically for R&D. Find a way to design them to make them to thrive on developing something new, make them have a desire to certain specific types of development and have a network between a handful of AI’s that are also operating in the same manner, just each working and specializing in different areas.

And all while the AI’s work, there would have to be a human team working with them, learning and developing with each other. Keeping humanity in the loop on developing technology while kicking it in overdrive basically.

But we shouldn’t ever get to the point of the Forerunners where AI’s could wreak havoc like that, though the UNSC is currently in a position like that, if there was ever a way to corrupt ship AI’s, one could probably just set them to self-destruct instantly.

All in all, AI’s are both good and bad, they just have to be properly moderated to avoid heavy reliance and protection if they turn against us.