Talk about anything: Page 72 - Godot Community Forums

Talk about anything

Bimbam

I should probably clarify the context, I was not referring to people labelling something AI and it not 'being' AI, as being AI is a low bar to achieve. Instead, I was just implying that often things advertised as 'Artificial Intelligence' are far from 'Intelligent'.

AI is the most ridiculously overused umbrella buzzword in data science, that depending on how you bend the definition could include something as simple as a calculator. Even Machine Learning is dubious as a definition, as many of the processes that fall under this term ultimately predate machines (regression analysis is 200+ years old).

It's almost as bad as when people throw 'Big Data' around. As far as I've seen, people usually just mean 'It crashed Excel' >.>

At least in gamedev the term AI generally refers to one thing and not a slew, that being NPC behavioural mechanics.

The thing is, if your NPC behaviour WAS actually learning the same way an ML model does... your players wouldn't win.

Audiobellum

@Bimbam @cybereality @fire7side

Now my stance on language is usually that it’s reality that shapes or perception of reality, and not the other way around. So replacing a word with a “bad” connotation with a “good” sounding word, doesn’t change how we perceive what it’s describing. Instead, it just changes how we see language as the “good” sounding word will even be seen as neutral or bad. However, I’m just going to make an exception for words like artificial intelligence for a few reasons. Most people don’t really see it as being “good” or “bad”. Also, people who misunderstand the word are necessarily separated from it’s reality. The “problem” with the term artificial intelligence, is the expectations it creates in people who don’t know how to code and make PC games. The only way this “problem” can fix itself overtime is if everyone became some sort of gamdev coder but, I’m not sure that would ever happen( though it would be cool if it did). To my mind, all “artificial intelligence”( even the good ones), are just a series of conditional statements reacting to what’s going on around it. Could the same be said for our intelligence? I think a better turn to use is just script. Instead of saying enemy AI, we could use enemy script. Then again, I anticipate that could create confusion.

cybereality

Well AI is the appearance of intelligence, not necessarily real intelligence. Is an Excel macro AI? In some cases maybe, since it can replace what someone's job was before computers were invented. Is the search on your computer hard drive AI? Possibly as well. It can mean a lot of things.

cybereality

@Bimbam As a kid, I thought the AI in Street Fighter II was a learning machine, so I would constantly switch up my style so it didn't figure out my technique. At least it seemed that way, because my cheese moves would stop working after a few rounds, but now I understand that the SNES processor was not that advanced. But since I believed it to be true, I guess it was convincing enough for a kid.

Bimbam

@cybereality said: Well AI is the appearance of intelligence, not necessarily real intelligence. Is an Excel macro AI? In some cases maybe, since it can replace what someone's job was before computers were invented. Is the search on your computer hard drive AI? Possibly as well. It can mean a lot of things.

Going way off topic (if that's possible in this thread) this issue is something that has come up recently for me at work, as the government in my country are finally trying to develop policy/frameworks about how AI is developed and used appropriately. While fundamentally I agree these policies/frameworks should exist (and should've done for decades), the incredibly loosely defined terminology makes them subject to ridiculously broad interpretation.

Their words were "Artificial Intelligence, or AI, is intelligent technology, programs and the use of advanced computing algorithms that can augment decision making by identifying meaningful patterns in data. "

Firstly, including the word 'intelligent' or 'advanced' in a definition is a recipe for disaster. Neither of these are quantifiable as they are relative, and the definition does not declare to what. 5*5 is advanced to a 3 year old... Secondly and perhaps more importantly, the above definition is literally saying that not just the code, but the USE of any code that aids decision making by identifying patterns in data, is AI.

By this definition, anyone who has used something as simple as Excel's flash fill function is included. And thus may? be subject to these new policies and limitations.

Anyway may seem like an incredibly moot point to most (not like I can do anything about it, tis what it is), but this may affect my ability to win work in future if someone decides some code I wrote is actually AI and therefore can't be used for some reason :(

Then again, if that happens I guess I quit and go full solo gamedev

OpinionatedGamer

A lot of people seem worried that AI is going to become too advanced and take over the world, like we are in some sort of science fiction movie. I'm not sure if they are serious or if it's all talk but either way I think we have more pressing matters to worry about right now lol =)

fire7side

@cybereality said: Well AI is the appearance of intelligence, not necessarily real intelligence. Is an Excel macro AI? In some cases maybe, since it can replace what someone's job was before computers were invented. Is the search on your computer hard drive AI? Possibly as well. It can mean a lot of things.

I think so, as long as it's processing information rather than just sitting there. We used to have switchboard operators that were really just an advanced switch. As long as it's making some kind of hierarchical decisions, I consider it AI.

Audiobellum

@OpinionatedGamer said: A lot of people seem worried that AI is going to become too advanced and take over the world, like we are in some sort of science fiction movie. I'm not sure if they are serious or if it's all talk but either way I think we have more pressing matters to worry about right now lol =)

What would need to happen for AI to take over the world would be a motive; this involves emotions and desires. I think the closet thing that AI could have to emotions and desires, would be a group of arbitrary goals set by it's creator, that the AI reinterpreted in ways that the creator didn't anticipate. I don't think it's impossible for AI to take over the world but, it's hard to image that we'd ever let that happen. What would change my mind is if the AI were to have a successful civil rights movement, then you know we're screwed. I don't think that AI would ever take over but, I think it might be cool if they did in my opinion.

fire7side

@Audiobellum said:

@OpinionatedGamer said: A lot of people seem worried that AI is going to become too advanced and take over the world, like we are in some sort of science fiction movie. I'm not sure if they are serious or if it's all talk but either way I think we have more pressing matters to worry about right now lol =)

What would need to happen for AI to take over the world would be a motive; this involves emotions and desires. I think the closet thing that AI could have to emotions and desires, would be a group of arbitrary goals set by it's creator, that the AI reinterpreted in ways that the creator didn't anticipate. I don't think it's impossible for AI to take over the world but, it's hard to image that we'd ever let that happen. What would change my mind is if the AI were to have a successful civil rights movement, then you know we're screwed. I don't think that AI would ever take over but, I think it might be cool if they did in my opinion.

I remember ages ago there was this show which had a big computer that was taking over the world and they ended up unplugging it. AI is only a nemesis because it's become a million spam bots. The future really hasn't ended up what we predicted it would be. We have self driving cars. We have spam.

some_lame_kid

@Bimbam so you're telling me, this janky bash script I wrote to convert my flac library to mp3 and export them to external storage is, by your country's legal definition, a robot? Can I put that on my resume?

some_lame_kid

Old dudes in office will never cease to make funny legislature about technology so long "intelligence agencies" aren't involved.

cybereality

@OpinionatedGamer said: A lot of people seem worried that AI is going to become too advanced and take over the world, like we are in some sort of science fiction movie. I'm not sure if they are serious or if it's all talk but either way I think we have more pressing matters to worry about right now lol =)

I hate to break it to you, but the AI already took over. The Singularity happened circa 2020, and AI now runs the world. Nothing looks any different, but basically everything is controlled by computers (except for old school forums like this, with no algorithm). Watch my book if you want to know why.

Audiobellum

@Bimbam

I’m not too sure if I’m in favor of the government regulating artificial intelligence. Even if we were to ensure that the policy makers in question could define AI in a sensible fashion, I’m not sure if definitions can content the meaning of AI indifferently. Though it can be legitimate, I usually think that the attempts of the deconstructionist does not accomplishes much in the long term; especially when dealing with descriptor of physical reality. However, the idea of AI just seems so abstract in comparison. Of course, the assumption I’m running with now is that it’s not really appropriate to define AI in the way I did previously in the thread( or at least in the context of government regulation). I’d also agree with you that the definition provided to you by your government does seem to board.

Audiobellum

In the context of government regulating "AI", how do you guys think AI should be defined?

some_lame_kid

@Audiobellum said: In the context of government regulating "AI", how do you guys think AI should be defined?

I don't think it's possible, it would work better to regulate the use and function of electronic systems. Trying to regulate "AI" is like trying to regulate the production of magnets to prevent the presence of illegal files on a hard drive. It's not logistically possible. Instead what's regulated is the process of creating illegal files that the magnet happens to be a part of. You could start by asking why does "AI" need to be regulated.

fire7side

Did you know most moons are like ours in that they rotate at a speed that one side always faces the planet? That question finally occurred to me and I looked it up. It must be heavier on one side and gravitation eventually slows it down so it's always facing the planet.

cybereality

Regulating AI is a mistake. Even the top researchers at Google don't even understand how neural networks work. It's basically a black box. No way some judge is going to be able to understand, when they fail to understand pinch to zoom on an iPad or how encryption works, which are much simpler concepts.

DaveTheCoder

An exception may be for autonomous vehicles.

If it detects that a person is in the road in front, there are also people on the sidewalk to the right, there's oncoming traffic in the other lane, and there's not enough room to stop, should a software engineer be free to decide what to code?

cybereality

People always use that argument, but it's irrelevant. Humans have to make that decision every day, there are about 17,000 car accidents every day in the US. And to get a drivers license, they never ask you ethical questions. They leave it up to your own judgement, which is what we should do with the AI. And the AI would honestly be able to make better decisions than most humans (who are driving drunk or doing drugs, watching YouTube, or sexting, when they should be looking at the road).

DaveTheCoder

But AI doesn't have judgement. It's a computer algorithm written by a programmer, who imparts his judgement into it. The programmer has to somehow decide on the rules that the algorithm uses for decision making.

« Previous Page Next Page »