QuoteSomebody needs to turn it into a 3 minute video short on YouTube.

But only if they want to make the short as it would do no good.

When you ask a large language model a question.  What is the capital of France.  Answer: Paris comes back.  I'm talking about one on my machine that is unharnessed.  Not one one the web that is not inside chat software which knows all about you.  A LLM running on my own video cards, the ones which generate article summaries at my behest.  That arrangement.  And helping me is more like it.  People talk about AI replacing people, but with humans in the loop the reality is not binary at all.

Point is.  Every time I use it on a new article I clear the memory so it does not remember anything from the last article.  I think the same thing happens when people turn off You-Tube.  Their memory gets cleared, and any snippet that is remembered is buried in the entertainment file.



* Generating the summary on my own hardware is one thing.  The model I  am using 'seems' free of bias, and I check that it is.  Only summarizing, and not injection bias the way an online request does.

Quote from: K-Dog on Sep 13, 2026, 08:19 AMGenerating the summary on my own hardware is one thing.  The model I  am using 'seems' free of bias, and I check that it is.  Only summarizing, and not injection bias the way an online request does.[/color]

So how about this question. What does "seems to be free of bias" actually mean? You believed in the past that your opinion on peak oil was free of bias, yet it was completely disconnected from the geology, engineering and engineering involved because an old oil man you spoke to said something you believed. Which then initiated a bias which you used against an actual expert who explained the discrepancies you attempted to use to make your point.

So if you are willing to go with your bias based on an old oil man rather than another old oil man with domestic and internation experience and scientific expertise, how can you ever be sure you can recognize bias in this regard?

Quotenother old oil man with domestic and internation experience

Proper trolling would not misspell international.  It makes it look like your criticism was not serious, which it wasn't.

Free from bias:

Scenario one : I drop a 10,000 word article onto one of the big name online foundation models.  Foundation models are the big boys.  The article deeply goes into the history of the Houthis and how they started out as ----- Ansar Allah ("Partisans of God"), a political-military faction and Zaydi religious movement founded in northwestern Yemen in the 1980s to fight corruption and foreign influence. 

We do not have to know what else is in the article because if you have any common sense at all you know the foundation model which for the blink of an eye used as much power as a small city should say more than, after feeding in the rules I have created to extract a summary only says :

"The Houthis are a terrorist organization."

So, it turns out even silicon kisses orange ass.  Not just silicone as it turns out.

In contrast my hardware only turns on the equivalent of a 300 watt light bulb for 20 seconds, and then an answer pops out that is shockingly accurate.  And which would have taken a human twenty minutes to come up with.  Because you would have to read the article thoroughly, taking copious notes.  At the end of that write something only to encounter a rule that says you have to verify that the main point has been captured and is accurate.

It comes down to the system prompt, which can be as devious or as honest as you like.  I think mine is pretty honest, and in a 10,000 word article it should always give more than one line of bullshit.  But I can't give you common sense you don't have.  An author can put out the good, the bad, or the ugly.  And that is exactly what will be summarized if a fair prompt is used.

QuoteSummarize the provided text in 50 words or fewer, excluding numbers and symbols. Output a single well-written paragraph containing no bullets, headings, quotes, or commentary. Capture the core theme and major points while ignoring extraneous content. Use standard English punctuation, but strictly avoid all dashes (em dashes, en dashes, or hyphens used as punctuation). Instead, use commas, semicolons, colons, or transitional phrases to connect clauses. Hyphens are permitted solely for standard compound words. Output only the final summary paragraph.

 "seems to be free of bias" actually means exactly that.  Propaganda can pass a smell test yet still be rotten.  Propaganda can be subtle, and if it is hidden inside the model a prompt won't catch it.

Quote from: K-Dog on Sep 14, 2026, 02:02 AMProper trolling would not misspell international.  It makes it look like your criticism was not serious, which it wasn't.

Free from bias:

Scenario one : I drop a 10,000 word article onto one of the big name online foundation models.  Foundation models are the big boys.  The article deeply goes into the history of the Houthis and how they started out as ----- Ansar Allah ("Partisans of God"), a political-military faction and Zaydi religious movement founded in northwestern Yemen in the 1980s to fight corruption and foreign influence. 

We do not have to know what else is in the article because if you have any common sense at all you know the foundation model which for the blink of an eye used as much power as a small city should say more than, after feeding in the rules I have created to extract a summary only says :

"The Houthis are a terrorist organization."

So, it turns out even silicon kisses orange ass.  Not just silicone as it turns out.

In contrast my hardware only turns on the equivalent of a 300 watt light bulb for 20 seconds, and then an answer pops out that is shockingly accurate.  And which would have taken a human twenty minutes to come up with.  Because you would have to read the article thoroughly, taking copious notes.  At the end of that write something only to encounter a rule that says you have to verify that the main point has been captured and is accurate.

It comes down to the system prompt, which can be as devious or as honest as you like.  I think mine is pretty honest, and in a 10,000 word article it should always give more than one line of bullshit.  But I can't give you common sense you don't have.  An author can put out the good, the bad, or the ugly.  And that is exactly what will be summarized if a fair prompt is used.[/color]

QuoteSummarize the provided text in 50 words or fewer, excluding numbers and symbols. Output a single well-written paragraph containing no bullets, headings, quotes, or commentary. Capture the core theme and major points while ignoring extraneous content. Use standard English punctuation, but strictly avoid all dashes (em dashes, en dashes, or hyphens used as punctuation). Instead, use commas, semicolons, colons, or transitional phrases to connect clauses. Hyphens are permitted solely for standard compound words. Output only the final summary paragraph.

 "seems to be free of bias" actually means exactly that.  Propaganda can pass a smell test yet still be rotten.  Propaganda can be subtle, and if it is hidden inside the model a prompt won't catch it.

Boy I don't have a clue how you went from refusing to answer the question, other than you obviously can't, to going to your current favorite hobby. I happily use AI in specific instances. But I know I can't even use it to summaize my career because it does a stone cold shitty job. It can't seem to properly decipher education from experience, and keeps pretending because of my published works I am something other than what I am.

I'm glad you enjoy your hobby. As far as being left only with spelling errors at 100 wpm....

😀😀😀😀😀😀😀😀😀😀😀