Anonymous #1 1d

That’s a different kind of AI as well PAL. Try to keep up.

-2

Anonymous OP 1d

go ahead and explain it then

-1

Anonymous #1 1d

Okay. Generative AI is prompt based and uses human research and training to generate its own responses. Artificial Intelligence (which has been around for a LONG TIME) is programmed to do specific tasks and cannot ‘think’ for itself. ASI is similar to generative, but much smarter. It’s meant to be like a human brain, but significantly smarter, with its main goal to “outperform” humans in any way possible.

-2

Anonymous OP 1d

none of them can 'think' lmao

-1

Anonymous #1 1d

“Artificial Intelligence … cannot ‘think’ for itself.” When did we lose our comprehension skills?

-1

Anonymous OP 1d

yeah, it cannot. it's a program. it follows code and instructions. do you think AI programs are sentient?

0

Anonymous #1 1d

Glad to hear you agree with me. And no, I don’t think it’s sentient per se, but it can come across that way since everything it’s learned has been from humans. However, robots and AI are not able to access our 5 senses and feel emotions, so again, no.

-2

Anonymous OP 1d

what is your point with any of this?

0

Anonymous #1 1d

Just that there’s a difference and I want to educate people on their differences. Why do you seem so upset about me spreading information? Are you against learning or something?

-1

Anonymous OP 1d

because AI people are annoying as fuck

0

Anonymous #1 1d

Well yeah, AI is very annoying. I made this post because I was frustrated that people think frequent use of ChatGPT is the equivalent as playing a video game that was made with artificial intelligence. There’s different kinds of AI, some more harmful than others. AI should obviously be avoided as much as possible, but using coded AI is MUCH DIFFERENT than generative AI.

-1

Anonymous OP 1d

do you mean like, when people use the term AI to refer to the NPC behaviors in a game? or games that were made with generative AI tools?

5

Anonymous #1 20h

Whoops I fell asleep, lol. I’m not referring to either. I’m saying that there is artificial intelligence used in code, it is very different from what people think of when they hear AI. It literally is just code than can run by itself but people believe that it is the same kind of AI as chatgpt.

1

Anonymous OP 20h

Please give a specific example of AI that is used in coding but is not generative

6

Anonymous OP 17h

Your definitions are mixing things up. "Prompt based" isn't what makes something generative. What makes it generative is that the model outputs new content instead of classifying or predicting. Lots of non-generative AI isn't hand-programmed either: spam filters, face unlock, recommendation feeds, and fraud detection all learn from data. The real split is rule-based vs learned, and within learned, generative vs not.

1

Anonymous #1 17h

ASI doesn't exist. Superintelligence is a hypothetical level of capability, not a kind of model anyone has built. So "it's called Super Intelligence now" is wrong, and so is describing ASI as a type of AI that's around today. Nobody has built human-level general AI yet, let alone something beyond it.

1

Anonymous OP 17h

"Code that runs by itself" is just automation, not AI. A script or scheduled task isn't intelligent. Real non-generative AI includes spam filters, fraud detection, recommendation algorithms, face/fingerprint unlock, GPS route planning, and NPC pathfinding. In coding specifically, examples are older autocomplete tools that use ML to rank suggestions, or models that flag likely bugs. Copilot-style tools are generative.

1

Anonymous #2 17h

I was making fun of Trump "declaring" that AI is now called SI, not endorsing the term

1

Anonymous #1 17h

ML models aren't written line by line. Engineers write the training process, but the behavior comes from billions of learned weights nobody hand-coded, which is why these models can surprise the people who built them. That doesn't make them sentient. But "it just follows instructions" isn't accurate either, and plenty of AI does take in sensor data like cameras and mics.

1

Anonymous #2 17h

How does taking in data from cameras or mics mean it isn't following instructions? It had code written, or generated, to interpret those inputs

0

Anonymous OP 17h

As someone who has been studying computer science for over a decade and researching AI for longer than ChatGPT has existed, your definitions are very painful to read.

1

Anonymous OP 17h

And the nerve you have calling everyone else idiots is very telling and not helpful. Make sure your own ducks are in a row before you worry about other people’s.

2

Anonymous OP 17h

Your actual point is fair, but you're drawing the line in the wrong place. The problem with generative AI isn't that it's "prompt based." It's what it costs: data centers draining power and water, models trained on artists' work without consent or pay, and jobs cut to save money. A spam filter does none of that. If you want people to stop lumping it all together, argue the harms, not made-up categories. Otherwise you're just handing AI capitalists easy ammo.

7

Anonymous #1 17h

The sensor bit was replying to OP's "5 senses" claim, not the instructions point. On that: yes, there's code, but it's a generic recipe: multiply numbers by weights, pass them along, repeat. That same code can run a face detector or a chatbot. What it actually does with an image comes from weights learned in training. Nobody wrote "if these pixels, then cat." So it's executing code, but not following instructions anyone wrote for the task.

1

Anonymous #1 17h

And that's why researchers say we don't fully understand these models. There's a whole field (interpretability) just for reverse-engineering what the weights learned. Grokking is a wild example: train a small network on modular arithmetic and it first memorizes the answers. Then, long after it seems done, it suddenly figures out the actual rule and generalizes. When researchers dug in, they found it had invented its own trig-based algorithm. Nobody programmed that.

1

Anonymous #2 12h

1. My definitions may seem painful bc they’re extremely simplified due to the character limit on here. 2. I called everyone who doesn’t think there is a difference idiots. If you’re offended by that, then that says more about you than it does me. 3. I never even stated why I think AI is bad, why are you assuming that I don’t like AI just bc it’s “prompt based”? 4. The categories are not made up, they may all still be AI, but there’s still subcategories regardless of how similar they are.

-1

Anonymous #2 12h

5. You’re mixing a lot of what I was saying and twisting it to make it seem like I was wrong, then you’re just agreeing with me in different words. That’s a weird way to debate.

0

Anonymous #2 12h

6. I never said that ASI exists. It’s still a concept being developed. I used the words “meant to be” because the functions I listed is how they want it to perform, if it was actively doing such things, I would’ve talked about it in the past tense like I did with the other categories.

0

Anonymous OP 12h

Character limit isn't the issue, I fit correct definitions in the same 500. And you did say why: "generative AI and their new models suck" and AI "should be avoided as much as possible." On ASI, you wrote "ASI is similar to generative, but much smarter." Present tense, listed as a category. Subcategories exist, nobody denied that. Yours just weren't real ones. Calling people idiots and then getting defensive when corrected isn't a great look.

4

Anonymous #1 12h

Sure! The game HALO uses something called Traditional Game AI which is non-generative.

-1

Anonymous OP 12h

Agreeing that subcategories exist isn't agreeing with yours. You said generative AI is defined by being prompt based: wrong. You said regular AI is programmed for specific tasks: most of it learns from data. You said AI in code is "just code that can run by itself": that's automation. Those are corrections, not different words for the same thing. If you said cats are reptiles and I said they're mammals, I'm not agreeing with you just because we both said cats are animals.

1

Anonymous OP 12h

Halo's enemy AI is a real non-generative example, but #1 asked about AI used in coding, not in a game. And when #1 asked you hours ago if you meant NPC behavior, you said you weren't referring to that. Now it's your go-to example? Pick a lane.

2

Anonymous #2 11h

So games actually have coding and that non-generative AI I mentioned is in that coding. I disagreed with 1 because I wasn’t referring to games made with generative AI or NPCs specifically. I admit that HALO may not have been the best example bc of the NPCs, so I’ll change my answer to Pac-Man.

-1

Anonymous OP 11h

So AI is not just anything automated lmao the Pac-Man "AI" is the ghosts. They're actually just NPCs. Each one follows its own hand-coded targeting rule: Blinky chases you directly, Pinky aims ahead of you. It's probably the most famous NPC “AI” in history. You dropped Halo for being too NPC-based and replaced it with the most NPC example possible. And when people complain about AI today, they mean neural-network stuff like ChatGPT, not Pac-Man ghosts.

5

Anonymous OP 11h

If you want to draw moral lines between kinds of AI, generative vs non-generative is the wrong axis. The more useful ones are cloud vs local and open source vs proprietary. A model running on your own machine doesn't need a data center, doesn't harvest your data, and doesn't make a corporation richer. A closed cloud model does all three. Who owns it and controls it matters way more than whether it writes text or sorts your spam.

1

Anonymous OP 11h

And Generative AI isn't automatically immoral. That idea rests on assumptions that don't always hold: that the person did zero thinking, that there was no input, editing, or quality control, and that every output is just stolen work. Sometimes those are true, and those cases deserve criticism. But a lot of people use it as one tool in a process they direct and refine. Judging the person without knowing how they used it is presupposing the verdict.

1

Anonymous #2 11h

Being defensive and responding to your claims are two different things, I was only participating in the conversation. That wasn’t supposed to be reasons as to why AI’s bad, it’s just my opinion, and it’s too vague for assumptions like that. I know how horrendous AI is for the environment, economy, and society; however that wasn’t the point of this post. And it’s listed as present tense bc it’s actively being developed. The other types of AI are already around so their functions are past tense.

-1

Anonymous OP 11h

Your tense excuse doesn't work. You wrote "Generative AI is prompt based" and AI "is programmed to do specific tasks." Present tense, and both exist. And AI itself isn't what's wrecking the environment. Profit-driven cloud data centers are. Run an open model locally in LM Studio and it uses about as much power as gaming on your PC. "AI is horrendous for the environment" blames the tool for what the business model does.

1

Anonymous OP 10h

You realize that if 15 years ago someone said "wow halo reach has really good AI," no part of that statement, or the technology used to code the enemy behaviors in halo reach, has anything to do with the current concept of the technology called AI, right? they are 100% completely different things

2