Resident Evil 1 - Isn't it the Red Queen's Fault?

[EvilWeasel35]

I just searched the zombie film forums but coulnd't find anything wabout the first Resident Evil film wich surpised me, so if i missed it plese move it to the correct thread!

I'm watching the first film and suddenly I realised that they they shut everything off with halon gas which killed everyone. If the red queen knew what would happen and the consequences then couldn't she have equvuated everone that was far away enough to be uninfected? The fact she exposed everyone to halon gas over-compensated and created more re-animated corspes. Wouldn't she have been advancd enough to know this would happen? Am I seeing holes or am I missing them?

[neoflux]

Red Queen is akin to 2001's HAL. The system is an adaptive AI... and as such likely functions based on heuristics.

A heuristic, is an algorithm which produces a reasonable solution to a problem that doesn't have 1 clear answer. This isn't always the most reasonable answer or the answer which generates the best results... it's the first random combination of variables which seems to match all of the criteria that the problem presents. The smarter the AI, the longer the list of criteria it will set and as a result the fewer possible combinations of results there will be... thus making the random one it selects more likely to be a wise course of action.

Computers can not think or infer one piece of data based on another... they can simply look for pre-programed cues which help them to define problems and then select a course of action which will resolve those problems. The flaw is that if the system does not identify enough of the factors when defining the criteria needed in order to resolve the problem then its results could cause even more problems.

OK, now in English for those who might not get that.

Say you have bags full of groceries and then need to be put away.

AI 1 might set the criteria as:
Where do we put groceries?

And it will generate an answer such as (cabinet, fridge, or freezer).
As a result, it might put all your groceries in the cabinet and you'll come home to find rotting meat and melted ice cream in the cabinet where you keep canned food.

AI 2 might set the criteria as:
Where do we put frozen food?
Where do we put cold food?
Where do we put canned/ packaged food?
Where do we put fresh food?

It will generate an answer for each of these questions independently.
As a result it will put the frozen food in the freezer, the cold food in the fridge, the canned food in the cabinet and the fresh produce in a bowl on the counter.

There is another factor which we are forgetting.... order. In this example, it might crush the eggs by putting the milk in on top of it, it might crush bread by stacking cans on it or squish peaches by dropping a watermelon on top.

AI 3 might set the criteria as:
Preset 1: address heaviest/ densest food in first.
Preset 2: do not stack food.
Where do we put frozen food?
Where do we put cold food?
Where do we put canned/ packaged food?
Where do we put fresh food?

You get the idea... the more advanced the AI the more refined it's criteria... but it never really thinks. But the most factors it takes into account and the more criteria it sets the slower the response. If your AI takes in too many factors, your ice cream might end up melting before it puts it away. As such the first criteria that most complex AI systems would likely look for is urgency.... how many factors can it take into account. Being in a research facility, it is likely that a containment issue would likely be priority 1... meaning it only can factor in 1 criteria.

In that case the only criteria might have been:
Stop All Potentially Infected People From Escaping

Now the AI will select the first course of action which fits that criteria. In this case it's lock the doors and gas them all.

It should also be noted that depending on what criteria the computer sets to define someone or something as 'people'... the infected might no longer be processed as people and so an AI might just as quickly open the doors once everyone is infected because then there are no longer people left to contain. This wasn't something Red Queen has done... but it is a reasonable possibility that an AI might do something along those lines.

The short version is that a computer never thinks... the intelligence in artificial intelligence are the size and complexity of it's dictionary (what it recognizes as factors to address), the complexity of the equation (number of criteria it can force an answer to meet), and the speed (rate at which it can find an answer which meets all criteria.) We simply personify anything which can take in information, pose a question or equation from those cues and generate an answer to that equation as being intelligent.

Does that clarify?
-Mike

[ZomCom]

Except, every automated system I can think of that has the ability to kill or hurt a human has an override. In fact, I'm not familiar with one that can hurt a human without human interaction. It's not some grand protocol that was agreed upon by an almighty AI cabal, it’s really just common sense. Allowing a computer to make decisions regarding human life and very expensive property is silly for the reasons you outlined. Maybe the folks that built The Red Queen never saw 2001.

[neoflux]

Except, every automated system I can think of that has the ability to kill or hurt a human has an override. In fact, I'm not familiar with one that can hurt a human without human interaction. It's not some grand protocol that was agreed upon by an almighty AI cabal, it’s really just common sense. Allowing a computer to make decisions regarding human life and very expensive property is silly for the reasons you outlined. Maybe the folks that built The Red Queen never saw 2001.

Well, first thing to consider is technologically we aren't quite there yet. So there isn't that many real world instances in which an AI could be in charge.

Now as far as Science fiction many simply concede to Asimov's 3 rules:


1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.
2. A robot must obey orders given to it by human beings, except where such orders would conflict with the First Law.
3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.

For every 'U.S. Robots and Mechanical Men' there is a 'Sky-Net'. These rules do not always apply to AI.

Even Asimov added the 0th rule:

"A robot may not harm a human being, unless he finds a way to prove that in the final analysis, the harm done would benefit humanity in general."

In reality, adherence to these laws would be quite difficult to script and even if you could program in enough scripts that it 'valued' human life (or at least didn't generate solutions which resulted in destroying it) then in practice that number of prescribed conditions would cripple the system's turn over speed. Because of this in many instances the operators would opt to shut off such protocols for productivity.

An instance of that is windows automatic update. Newer ports of Windows are designed to have internal components which automatically update. Though these updates are designed to protect the computer from outside threats in practice they often cause unexpected internal conflicts with different applications and the fact that simply by allowing it that's 1 more program sapping your resources many advanced users shut this and other 'protection' tools off.

Now you could say, 'You're talking about file security on your laptop... with Red Queen we're talking about human lives.' But I think that's neglecting the nature of the Umbrella Corporation. I mean this is a group which tailored viruses, blew up cities full of people to cover their asses, infected others so they can play saviors and have done nothing but manipulate people through the series of games, movies, books, etc... Do you really think such a company would even risk a reduction in Red Queen's productivity in order to protect a few human lives?

-Mike

[ZomCom]

Yes, we can construct an argument that features intelligent AI and a controlling agency that has no regard of human life. So, given those parameters, The Red Queen having the ability to kill humans is possible in that framework. But, it’s still not reasonable. Think of the decision making that would go into that programming:

***SPOILERS***

If there is an contamination in the Hive,

the Hive will be locked down, allowing no access into or out of the facility (reasonable)

the automated systems will attempt to isolate the contamination inside the Hive by closing airtight doors, etc. (reasonable)

the Hive will then kill all the scientists working in the Hive, which means killing the very people with expert knowledge who are developing the virus and it’s antidote (unreasonable)

the automated system will defend itself from any attempt by the corporation which created it to re-enter the Hive and find out what he hell happened (unreasonable)

the automated system will include a laser defense system that will defend itself from an attempt to shut down the automated system, and it cannot be overridden (unreasonable),

and once shut down, all hell breaks loose because there is no degraded mode, just open the gates (unreasonable)

and once reactivated it will attempt to autodestruct and destroy an entire city (unreasonable),

and the automated system has the ability to make decisions without the approval of the corporation, and it has the ability to ignore the corporation’s attempts to override its decisions (only reasonable in a video game).

I don’t care how callous and evil are the creators, even Dick Cheney would balk at this decision matrix.

[neoflux]

That is the flaw of heuristics. Each question is answered based on the assumption that all previously selected answers were correct. With such a system, it could in theory make a billion right decisions... but once it makes one flawed decision... it will begin to make all related decisions based on that flawed decision. Each answer only has to be a reasonable progression from the previous answer not reasonable to an outside observer or fitting with any kind of logic.

We've all played those logic games.

There is hair on the couch.
Couches do not have hair.
Something shed hair on the couch.
Animals shed hair.
Dogs are animals that shed hair.
There must be a dog.
A Labrador Retriever is a dog.
Labrador Retrievers like to swim.
...

All of which are reasonable on a mechanical level... but that doesn't mean you won't come home to find the AI drowned your cat.

You're trying to apply ideas of what is reasonable from a human prospective... we aren't talking about a human.

The only decision made by a human was based on observation:
- Red Queen has shown to always make reasonable choices.
- Protocol slow down it's productivity and have showen to be unnecessary.
- The more productive Red Queen is, the faster they will see results.
- The faster you see results, the greater the potential profits.

I mean if those were the facts as observed... it isn't unreasonable for the corporate head to put the system in charge.

As far as preventing people from coming in:
Humans have flawed reasoning.
I always make logical choices.
I should prevent humans from interfering with my process.
-=God Mode=-

In short, people depending on computers is reasonable. Computers run things in a way which is logical. The reality of any situtation is frequently neither reasonable or logical.

-Mike

[ZomCom]

You aren’t going to come home and find the AI drowned your cat because no one is going to build or buy an AI system that can drown your cat.

The Red Queen initially kills people using systems that are not designed to be lethal, the halon extinguishers and the water sprinklers. That would indicate she is not supposed to be able to kill people. Or, else the evil corporation decided to give her a fire suppression system that also has a lethal capability. Either that is a huge mistake (it’s unlikely an automated system would be so poorly designed it kills people), or evidence of Umbrella’s evil genius (in which case, if they are so intelligent why did they allow the computer to be able to take over and do things the corporation does not want to have happen). Then, they gave the computer the ability to override any human input (explain that decision to the shareholders). They give the computer a lethal laser system (obviously Umbrella played off the OSHA inspectors). They also gave the computer the ability to blow up an expensive facility and the town under which it lies (you know that’s unrealistic because the guy who designed the computer lives in that town with his family).

That’s a lot of unrealistic decisions. Generally speaking, the more totalitarian is the system, the more centrally controlled it is. I think the evil geniuses at Umbrella would be less likely to cede lethal power to the computer. I’m not arguing that a computer as smart as the Red Queen could not, or would not be capable of taking lethal actions. I’m saying that the designers would not give her the ability to do that. It’s not a matter of her going nuts. It’s the fact she would simply not have autonomous control over lethal systems. Why? You said it yourself, “Computers run things in a way which is logical. The reality of any situation is frequently neither reasonable or logical.”

[neoflux]

You also have to factor in the "Astroboy" complex. Red Queen is designed after the programmer's daughter. The programmer likely extended a bit more liberties to what he saw the AI as a daughter. If he started to really see her as his daughter, it's completely possible he gave her the tools needed to defend herself... even without the knowledge of the company.

Now honestly with new technology we take plenty of things into our homes which later prove to be dangerous. How many products are recalled months or years after hitting the market when they found issues which only arose after long term use etc. This wasn't even opened to mass market so who knows which beta version they were on.

Look at the HAL system from 2001 (inspiration for Red Queen).. It's easy to say that after it goes crazy for daisy that we wouldn't use it... but before hand... it does the job.

-Mike

[ZomCom]

It's like that Star Trek episode where the guy programs the computer after himself, and he can't admit that he and the computer are nuts.

The idea of a creation that goes awry can be traced back through Battle Star Galactic, Star Trek, the Alien series, the Terminator series, Fail Safe, Doctor Strangelove, Frankenstein (the movie and the early 1800’s book), and even further back. The story of Adam and Eve is also about created beings having and misusing free will, except the warning inherent in the story is for the creations, not the creator. Will we ever learn?

[neoflux]

Will we ever learn?

The scorned creator goes back to the dawn of time.
Hepheastus... the creator of technology... the god with an ugly face.

The whole reason all of our old smith/ craftsman/ forge/ creators myths going back to our earliest history shows the creator as lame and unsightly is to say that no matter how beautiful the creation we must keep in mind that it is of flawed hands and nothing is ever perfect.

-Mike

PS: Looking over this thread... I think I've leveled-up in my geekdom by hitting Programing, Mathematics, Logic games, Sci-Fi, Anime, Horror and Greek mythology in the same argument.

[ZomCom]

You be da Man!:clap: