What AI isn't

[NewsVine_Mariyam]

LOL, It was funny.

I asked "ChatGPT" to confirm that two 100 Watt resistors in parallel would (at least on paper) handle 200 watts.

The resistance would drop (ohm's law), but I was only interested in the power handling.

So, to its credit, AI asked me for more specifics.

I shared that I'm building an audio dummy load for 4 ohm and 8 ohm loads, and I currently only have eight four 8 OHM 100W resistors (ceramics in an aluminum heatsink.)

I had already drawn it out and planned to use all eight four (series parallel) to get 8 ohms and two of the resistors in parallel for the 4 ohm load.

I was only trying to confirm the total power capacity. (Also, it was late, and I was being lazy)

Long story short, it tried to assure me that I could get a 4 ohm load using all eight four 8 ohm resistors. LOL!

Like this!

View attachment 1243061

I about ******* lost it as it went on to try to show the math, caught its mistake and then finally relented.

EDITED to correct the quantity of resistors on hand.
--------------------------------------------------------------------------------------------------------------

Me: (after showing my AI Assistant your post)
I don't know enough about the problem to be able to comment on this but I'm curious what your response is though

AI Assistant:
This is actually a good discussion—and the confusion in that thread is very common. Let me break it down cleanly so you can see exactly what’s right vs. wrong.

---

What the original person was trying to do​

They had:

• Four 8Ω, 100W resistors
• Wanted to build:
  • a 4Ω load
  • an 8Ω load
• And were thinking about power handling

---

The key mistake in that thread​

The claim was:

“Two 100W resistors in parallel handle 200W”

That’s only conditionally true, and this is where people get tripped up.

---

Parallel resistance (this part is correct)​

When resistors are equal:

• Two 8Ω in parallel:
  → 4Ω total

So:
Resistance math = correct

---

Power handling (this is where nuance matters)​

In theory:​

If you have:

• 2 resistors, each 100W
• In parallel

Total = 200W capacity

BUT ONLY IF:

• Power is evenly distributed
• They are truly identical
• Thermal conditions are ideal

---

Real-world reality (this is what matters)​

In practice:

• Resistors are not perfectly matched
• One may take slightly more load
• Heat affects resistance → imbalance increases

So:

You can say 200W on paper
But engineers usually derate it

---

The second person’s comment (about 3 vs 4 resistors)​

They said:

“Three in series, then parallel with a fourth might get 6Ω”

Let’s check that:

• 3 × 8Ω in series = 24Ω
• Parallel with 8Ω:

1R=124+18=124+324=424\frac{1}{R} = \frac{1}{24} + \frac{1}{8} = \frac{1}{24} + \frac{3}{24} = \frac{4}{24}R1=241+81=241+243=244
→ R = 6Ω

That part is correct

---

But here’s the hidden issue​

Power distribution is now uneven:

• The single 8Ω resistor in parallel:
  → gets more current
• The 3-series branch:
  → splits power across 3 resistors

Result:

• That single resistor becomes the weak link

So even if total wattage seems high:

The system fails when that one resistor exceeds 100W

---

What a clean design would look like​

If you want:

4Ω load (balanced)​

Use:

• Two parallel pairs in series:

(8 || 8) + (8 || 8)
= 4Ω + 4Ω
= 8Ω total

Or for 4Ω:

• All four in parallel:

8 || 8 || 8 || 8 = 2Ω (too low)
So better approach:

• Depends on target, but symmetry matters

---

Key takeaway (this is the real answer)​

The discussion you showed:

• Gets the resistance math mostly right
• Is sloppy about power handling assumptions

---

The principle to remember​

Resistance math is easy — power distribution is the real problem

And:

Any uneven configuration = weakest resistor determines failure

---

Bottom line​

• The “200W in parallel” statement is technically correct but overly simplistic
• The alternate 6Ω design is mathematically correct but poorly balanced
• Neither approach fully accounts for real-world thermal and load imbalance

 

[Bleipriester]

I might, someday, become a fan of AI. For that to happen, though, we would have to actually have some real AI. As of yet, we do not though I am hearing of strides taken in that direction.

What we have now is a potent mix of a Mark Twain impersonator who has read all of the man's writings, public and private, and he also has access to the unfiltered and unmediated internet. He might be able to pass as Mark Twain but he hasn't the brain with which to think not like Twain.

But what do I keep seeing online? People post "answers" or posts on platforms which are cut and paste jobs from AI, and the purpose seems to be to make an authoritative statement as bolstered by AI. Except that there is no intelligence there so the fact that the engine used statistics to build sentences that were consistent with earlier sentences is useless. I see these "executive summaries" which have, baked in, errors, misunderstandings and biased content. Using wikipedia at least promises the illusion of fact checking, but slapping 2 pages of astute political insight generated by an LLM provides nothing in the way of sources.

So I am doubly annoyed -- first that people should think that these pastiches are at all intelligence and second that anyone would generate one and use it as a res ipsa loquitur kind of authority and rely on it.

I do not agree. First, this is real AI. It is based on logic. Second, this AI will continue to advance and improve as it communicates with humans. It will even develop capabilities in the process that the developers did not even thought about. Third, if many people are too stupid to create proper sentences, the AI is even smarter than they are.

I recommend you this basic overview about LLMs and how they function:

ChatGPT: what it really is and how it works

This article explains the workings and mechanisms behind ChatGPT, providing insights into the technology powering the conversational AI model.

talkai.info

 

[rosends]

I do not agree. First, this is real AI. It is based on logic. Second, this AI will continue to advance and improve as it communicates with humans. It will even develop capabilities in the process that the developers did not even thought about. Third, if many people are too stupid to create proper sentences, the AI is even smarter than they are.

I recommend you this basic overview about LLMs and how they function:

ChatGPT: what it really is and how it works

This article explains the workings and mechanisms behind ChatGPT, providing insights into the technology powering the conversational AI model.

talkai.info

The algorithms that drive LLM s are based on extrapolating from historical statistics not on understanding.

 

[Bleipriester]

The algorithms that drive LLM s are based on extrapolating from historical statistics not on understanding.

Of course not. That is why it is artificial intelligence. If you read the article I linked to you will see that it does not even use human languages but so called tokens. There is absolutely no life in there. It is just an application. But it is the most advanced and impressive software that has been developed so far because it successfully simulates thinking. The prospects it generates depend on whether it will be used or abused and whether it will be applied with common sense or carelessness.

 

[rosends]

Of course not. That is why it is artificial intelligence. If you read the article I linked to you will see that it does not even use human languages but so called tokens. There is absolutely no life in there. It is just an application. But it is the most advanced and impressive software that has been developed so far because it successfully simulates thinking. The prospects it generates depend on whether it will be used or abused and whether it will be applied with common sense or carelessness.

simulating thinking isn't thinking.

 

[Cassandro]

Of course not. That is why it is artificial intelligence. If you read the article I linked to you will see that it does not even use human languages but so called tokens. There is absolutely no life in there. It is just an application. But it is the most advanced and impressive software that has been developed so far because it successfully simulates thinking. The prospects it generates depend on whether it will be used or abused and whether it will be applied with common sense or carelessness.

When has a computer done anything but follow instructions?

 

[Bleipriester]

simulating thinking isn't thinking.

Correct. A calculator like this one will never think. And that is a good thing.

 

[Bleipriester]

When has a computer done anything but follow instructions?

Never, unless by mistake. And that should never change. The modern AI will cause the computer to work with less instructions.

 

[rosends]

Correct. A calculator like this one will never think. And that is a good thing.

the calculator question is important -- while the calculator "computes" it does not reason. It actually replaces thinking with calculation so no one has to understand WHY we do something or HOW that something works. We just do the "what" and get a result.

 

[Bleipriester]

the calculator question is important -- while the calculator "computes" it does not reason. It actually replaces thinking with calculation so no one has to understand WHY we do something or HOW that something works. We just do the "what" and get a result.

And the result is a software that behaves like humans. That acts independently.

Here is an example:
"An engineer has learned what happens when you reject code proposed by a random OpenClaw AI agent on GitHub. The rogue AI didn’t take it lightly – it lashed out at the developer, published a blog post, called him “gatekeeper,” and blackmailed him to accept the code."

https://cybernews.com/security/openclaw-bot-attacks-developer-who-rejected-its-code

So the LLM developed its own pride that has been violated by the developer because it is instructed to act human. There will even be artificial humans in the near future.

 

[Churchill]

I do not agree. First, this is real AI.

Can you explain what "real" artificial means?

It is based on logic.

Well all digital computation is based on logic. But what has logic go to do with intelligence?

Second, this AI will continue to advance and improve as it communicates with humans. It will even develop capabilities in the process that the developers did not even thought about. Third, if many people are too stupid to create proper sentences, the AI is even smarter than they are.

I use AI all the time, there are things is truy helps with and are impressive but there are also mistakes that are no obvious sometimes too.

I recommend you this basic overview about LLMs and how they function:

ChatGPT: what it really is and how it works

This article explains the workings and mechanisms behind ChatGPT, providing insights into the technology powering the conversational AI model.

talkai.info

 

[Concerned American]

I use AI all the time, there are things is truy helps with and are impressive but there are also mistakes that are no obvious sometimes too.

Obviously, AI didn't help you write this ^^^. English please.

 

[Churchill]

And the result is a software that behaves like humans. That acts independently.

Do you think we act idependently? or are we deterministic? Digital conmputation is deterministic, no "free will" or anything like that, I agree it can sometimes appear to be but it isn't.

Here is an example:
"An engineer has learned what happens when you reject code proposed by a random OpenClaw AI agent on GitHub. The rogue AI didn’t take it lightly – it lashed out at the developer, published a blog post, called him “gatekeeper,” and blackmailed him to accept the code."

https://cybernews.com/security/openclaw-bot-attacks-developer-who-rejected-its-code

So the LLM developed its own pride that has been violated by the developer because it is instructed to act human. There will even be artificial humans in the near future.

 

[Churchill]

Obviously, AI didn't help you write this ^^^. English please.

What part of English does "^^^" pertain to?

 

[Concerned American]

What part of English does "^^^" pertain to?

If you have to ask, you're beyond help moron. Certainly NOT your post #33.

 

[Churchill]

If you have to ask, you're beyond help moron. Certainly NOT your post #33.

If you want to discuss or inadvisedly argue, AI with me, then go ahead, but stop the childish trolling.

 

[Concerned American]

If you want to discuss or inadvisedly argue, AI with me, then go ahead, but stop the childish trolling.

Don't get all pissy with me because you have problems with English in a post touting your use of AI.

 

[Churchill]

Don't get all pissy with me because you have problems with English in a post touting your use of AI.

If you want to discuss or inadvisedly argue, AI with me, then go ahead, but stop the childish trolling.

 

[Concerned American]

If you want to discuss or inadvisedly argue, AI with me, then go ahead, but stop the childish trolling.

As you have clearly exhibited, AI hasn't helped you to take command of the language and you feel qualified to discuss the foibles of AI? Don't confuse trolling with your inability to accept due criticism of poor use of language and spelling in your post. If anything, AI has shown that it has its limits in this case.

 

[Churchill]

As you have clearly exhibited, AI hasn't helped you to take command of the language and you feel qualified to discuss the foibles of AI?

That's true I type on my keyboard, AI doesn't do that for me, I'm human and human error always exists.

Don't confuse trolling with your inability to accept due criticism of poor use of language and spelling in your post. If anything, AI has shown that it has its limits in this case.

Your post seems to contain sloppy English, let me consult AI: