Frank
@frankjeweetwelwie.bsky.social
That's worth another discussion, but if it only spits out citations, does that fall under copyright infringement?
November 19, 2025 at 6:02 PM
That's worth another discussion, but if it only spits out citations, does that fall under copyright infringement?
I liked Google more 3+ years ago than today.
November 19, 2025 at 5:57 PM
I liked Google more 3+ years ago than today.
Is this all about word choosing? Should I have said, copied, indexed, or something? In computer language, it all boils down to funtions like fread() to get the data out of a file, probably an ebook found online. And yes, I know nothing about LLMs.
November 19, 2025 at 5:52 PM
Is this all about word choosing? Should I have said, copied, indexed, or something? In computer language, it all boils down to funtions like fread() to get the data out of a file, probably an ebook found online. And yes, I know nothing about LLMs.
Semantics. In that sense I'm happy it didn't "read" and just ate the bytes from the ebook it found. If a real person writes an article about is, with enlighting personal thoughts, that will be read/scraped too and become useful when someone prompts for "any thoughts on the attitude of..."
November 19, 2025 at 5:44 PM
Semantics. In that sense I'm happy it didn't "read" and just ate the bytes from the ebook it found. If a real person writes an article about is, with enlighting personal thoughts, that will be read/scraped too and become useful when someone prompts for "any thoughts on the attitude of..."
Fact or fiction 😉 The first version hadn't read the book completely. The second had found a timeline and gave a correct outcome. As with everything, it will never be perfect but as long as efforts are made to improve it..
November 19, 2025 at 5:31 PM
Fact or fiction 😉 The first version hadn't read the book completely. The second had found a timeline and gave a correct outcome. As with everything, it will never be perfect but as long as efforts are made to improve it..
Hopefully it will improve further, until that random becomes (very/extremely) rarely.
November 19, 2025 at 5:26 PM
Hopefully it will improve further, until that random becomes (very/extremely) rarely.
The first one, with 104 years, would have fooled me, as I have not read that book. It's good that it is improving, and the second example does mention the underlying reason too. As long as it keeps improving..
Other sources can be wrong too. Some colleagues I value more than others.
Other sources can be wrong too. Some colleagues I value more than others.
November 19, 2025 at 5:17 PM
The first one, with 104 years, would have fooled me, as I have not read that book. It's good that it is improving, and the second example does mention the underlying reason too. As long as it keeps improving..
Other sources can be wrong too. Some colleagues I value more than others.
Other sources can be wrong too. Some colleagues I value more than others.
Statistically, it will scrape mostly statements that a diode conducts in only one direction. Which is correct, for most diodes anyway. Was there a reason to prevent repeatedly same output? Or is that only about the format of the message rather than its content?
November 19, 2025 at 5:01 PM
Statistically, it will scrape mostly statements that a diode conducts in only one direction. Which is correct, for most diodes anyway. Was there a reason to prevent repeatedly same output? Or is that only about the format of the message rather than its content?
So the value judgement is made on highest frequency of a particular statement. I would assume.
November 19, 2025 at 4:53 PM
So the value judgement is made on highest frequency of a particular statement. I would assume.
If AI scrapes technical stuff from reddits or blogs from companies selling technical stuff, github, or whatever, it will contain some noise and false statements but not all is crap. So if the majority says a diode conducts in only one direction, what would AI make of that?
November 19, 2025 at 4:51 PM
If AI scrapes technical stuff from reddits or blogs from companies selling technical stuff, github, or whatever, it will contain some noise and false statements but not all is crap. So if the majority says a diode conducts in only one direction, what would AI make of that?
That's a funny example! Chatgpt produces this:
November 19, 2025 at 4:38 PM
That's a funny example! Chatgpt produces this:
She's not on the internet.
November 19, 2025 at 4:29 PM
She's not on the internet.
On a technical topic, my aunt Nancy doesn't give much input of value. So that filters itself. I tend to agree on social issues, seeing the crap on social media - which I believe should be forbidden or at least limited to sharing funny cat videos only - but that's a more political discussion.
November 19, 2025 at 4:04 PM
On a technical topic, my aunt Nancy doesn't give much input of value. So that filters itself. I tend to agree on social issues, seeing the crap on social media - which I believe should be forbidden or at least limited to sharing funny cat videos only - but that's a more political discussion.
I'm not doing anything deliberately here. AI spitting out the current concensus and doing a fairly job, I said. And using data it didn't dream up but scraped from all over the place. There you have your community. The whole world.
November 19, 2025 at 3:58 PM
I'm not doing anything deliberately here. AI spitting out the current concensus and doing a fairly job, I said. And using data it didn't dream up but scraped from all over the place. There you have your community. The whole world.
For sure I am not perfect at checking the output, so I'll give you 5% extra. Leaving 75% of satisfying output for those situations where my other tools failed 100%
Didn't know me overlooking those grid faults had such impact. Had you asked me to produce such grid, I would have used excel.
Didn't know me overlooking those grid faults had such impact. Had you asked me to produce such grid, I would have used excel.
November 19, 2025 at 3:49 PM
For sure I am not perfect at checking the output, so I'll give you 5% extra. Leaving 75% of satisfying output for those situations where my other tools failed 100%
Didn't know me overlooking those grid faults had such impact. Had you asked me to produce such grid, I would have used excel.
Didn't know me overlooking those grid faults had such impact. Had you asked me to produce such grid, I would have used excel.
Yes, I am aware of that. By adding noise or plain malicious crafted data, or censoring, it all falls apart. Made a comment somewhere, expressing my hope that AI doesn't scrape social media for data, just to name one. It's a problem with all media too, or even some of our leaders around the world.
November 19, 2025 at 3:41 PM
Yes, I am aware of that. By adding noise or plain malicious crafted data, or censoring, it all falls apart. Made a comment somewhere, expressing my hope that AI doesn't scrape social media for data, just to name one. It's a problem with all media too, or even some of our leaders around the world.
Consensus does not mean it's only between experts. Any group, community, even in a family, there can be consensus on a topic. You tried to question me with your comment on 'expert consensus' which I never said. Maybe look up the latin origin.
November 19, 2025 at 3:35 PM
Consensus does not mean it's only between experts. Any group, community, even in a family, there can be consensus on a topic. You tried to question me with your comment on 'expert consensus' which I never said. Maybe look up the latin origin.
Fairly decent job -> 80%
I already admitted I overlooked the missing column and some other faults, did check only a few squares. And added that my lame excuse for that was being more or less conditioned by the first grid that had all columns and rows. I should be sent to prison for that.
I already admitted I overlooked the missing column and some other faults, did check only a few squares. And added that my lame excuse for that was being more or less conditioned by the first grid that had all columns and rows. I should be sent to prison for that.
November 19, 2025 at 3:26 PM
Fairly decent job -> 80%
I already admitted I overlooked the missing column and some other faults, did check only a few squares. And added that my lame excuse for that was being more or less conditioned by the first grid that had all columns and rows. I should be sent to prison for that.
I already admitted I overlooked the missing column and some other faults, did check only a few squares. And added that my lame excuse for that was being more or less conditioned by the first grid that had all columns and rows. I should be sent to prison for that.
I only used 'current consensus'. Which I see as the averaged outcome from all data that has been scraped. Don't twist my words.
November 19, 2025 at 3:18 PM
I only used 'current consensus'. Which I see as the averaged outcome from all data that has been scraped. Don't twist my words.
I quantified it already for you, claiming it does a fairly good job (at what I ask it). Sometimes I make this joke, that on a scale of 0-10, I know a thing or two. In this group there seems to be a hidden contest to insult people who are not an inaugurated member of the same church. Bit sad.
November 19, 2025 at 3:12 PM
I quantified it already for you, claiming it does a fairly good job (at what I ask it). Sometimes I make this joke, that on a scale of 0-10, I know a thing or two. In this group there seems to be a hidden contest to insult people who are not an inaugurated member of the same church. Bit sad.
Hahaha, that's true. Point taken. Room for improvement. But in my book, still anecdotal.
November 19, 2025 at 2:48 PM
Hahaha, that's true. Point taken. Room for improvement. But in my book, still anecdotal.
That's is problematic, to put it mildly. I try to use the best available tool I have, regardless what I am doing, but at times I also use a quick and dirty method - if it works for me. Like a cutting a piece of wood with a blunt saw. Ad hoc stuff. Not all outcomes require 20 decimals of accuracy.
November 19, 2025 at 2:46 PM
That's is problematic, to put it mildly. I try to use the best available tool I have, regardless what I am doing, but at times I also use a quick and dirty method - if it works for me. Like a cutting a piece of wood with a blunt saw. Ad hoc stuff. Not all outcomes require 20 decimals of accuracy.
I'm not trying to prove anything here. It's a moment in the middle of a conversation I had. In the screen copy, on a topic of which it has no data, it doesn't hallucinate. I've never seen it hallucinate on my relatively down to earth prompts. Some folks here go ballistic about baking powder.
November 19, 2025 at 2:40 PM
I'm not trying to prove anything here. It's a moment in the middle of a conversation I had. In the screen copy, on a topic of which it has no data, it doesn't hallucinate. I've never seen it hallucinate on my relatively down to earth prompts. Some folks here go ballistic about baking powder.