That's a good reason to dislike AI images! All I'm saying is that I don't think outsourcing some tasks is a sign of lack of critical thinking. I agree with you that it's better if those tasks are outsourced to artists rather than a plagiarism machine.
November 29, 2025 at 3:34 AM
That's a good reason to dislike AI images! All I'm saying is that I don't think outsourcing some tasks is a sign of lack of critical thinking. I agree with you that it's better if those tasks are outsourced to artists rather than a plagiarism machine.
My point was specifically about the phrase "outsourcing their critical thought". I am not saying that there aren't good reasons to dislike the use of AI images. I dislike AI images.
Spending less time thinking of one thing can give you more time to think of another. That's often why we outsource.
Spending less time thinking of one thing can give you more time to think of another. That's often why we outsource.
November 29, 2025 at 2:35 AM
My point was specifically about the phrase "outsourcing their critical thought". I am not saying that there aren't good reasons to dislike the use of AI images. I dislike AI images.
Spending less time thinking of one thing can give you more time to think of another. That's often why we outsource.
Spending less time thinking of one thing can give you more time to think of another. That's often why we outsource.
I can't prove that someone who doesn't use AI for image generation doesn't use it for other things. They may just be aware of the distaste for AI images and are smart enough to avoid them.
November 29, 2025 at 12:14 AM
I can't prove that someone who doesn't use AI for image generation doesn't use it for other things. They may just be aware of the distaste for AI images and are smart enough to avoid them.
In the case of some specific people I'm thinking of, they have a body of work that predates LLMs. People who are much harder workers than I am, and I know they have a drive to make new things. I appreciate when people challenge them about their use of AI.
November 29, 2025 at 12:11 AM
In the case of some specific people I'm thinking of, they have a body of work that predates LLMs. People who are much harder workers than I am, and I know they have a drive to make new things. I appreciate when people challenge them about their use of AI.
However, using AI images may indicate that you have bad taste, or are unconcerned about models being trained on images without permission, or are unaware of the distaste many have towards AI generated content.
November 28, 2025 at 10:38 PM
However, using AI images may indicate that you have bad taste, or are unconcerned about models being trained on images without permission, or are unaware of the distaste many have towards AI generated content.
When I tested AI image generation a couple years ago, it often gave multiple options. And I could ask for changes.
Using a stock photo could save you 99% of the effort of taking your own photo. If AI saves you 99.9% instead, I don't see that as an indicator of how willing you are to put in effort.
Using a stock photo could save you 99% of the effort of taking your own photo. If AI saves you 99.9% instead, I don't see that as an indicator of how willing you are to put in effort.
November 28, 2025 at 10:38 PM
When I tested AI image generation a couple years ago, it often gave multiple options. And I could ask for changes.
Using a stock photo could save you 99% of the effort of taking your own photo. If AI saves you 99.9% instead, I don't see that as an indicator of how willing you are to put in effort.
Using a stock photo could save you 99% of the effort of taking your own photo. If AI saves you 99.9% instead, I don't see that as an indicator of how willing you are to put in effort.
I don't disagree with you on point one.
2. You type keywords in a search field and pick the image you like the best... Not sure how that's all that different from generation of an AI image, in effort required.
3. I occasionally use AI for coding, and have never used it for writing for communication.
2. You type keywords in a search field and pick the image you like the best... Not sure how that's all that different from generation of an AI image, in effort required.
3. I occasionally use AI for coding, and have never used it for writing for communication.
November 28, 2025 at 8:42 PM
I don't disagree with you on point one.
2. You type keywords in a search field and pick the image you like the best... Not sure how that's all that different from generation of an AI image, in effort required.
3. I occasionally use AI for coding, and have never used it for writing for communication.
2. You type keywords in a search field and pick the image you like the best... Not sure how that's all that different from generation of an AI image, in effort required.
3. I occasionally use AI for coding, and have never used it for writing for communication.
Would you ask the same thing if they were using clip art or stock photos?
November 28, 2025 at 8:04 PM
Would you ask the same thing if they were using clip art or stock photos?
If they used clip art or stock images instead, would you make the same complaint about them "outsourcing their critical thought"?
I dislike AI imagery, but I'm not sure that outsourcing to AI (or stock images) in one field should make me question their competency in an unrelated field.
I dislike AI imagery, but I'm not sure that outsourcing to AI (or stock images) in one field should make me question their competency in an unrelated field.
November 28, 2025 at 8:00 PM
If they used clip art or stock images instead, would you make the same complaint about them "outsourcing their critical thought"?
I dislike AI imagery, but I'm not sure that outsourcing to AI (or stock images) in one field should make me question their competency in an unrelated field.
I dislike AI imagery, but I'm not sure that outsourcing to AI (or stock images) in one field should make me question their competency in an unrelated field.
Sometimes it's to save time thinking in one area to free up more time to think in another area.
I do despise AI imagery, though.
I do despise AI imagery, though.
November 28, 2025 at 7:43 PM
Sometimes it's to save time thinking in one area to free up more time to think in another area.
I do despise AI imagery, though.
I do despise AI imagery, though.
One big issue is that AI isn't good at saying "I don't know". And it should be able to give a day of the week that matches the date it thinks it is, even if it isn't accurate about the date. I think the current value of AI is overstated, and I don't know how much the current methods can be improved
November 28, 2025 at 7:06 PM
One big issue is that AI isn't good at saying "I don't know". And it should be able to give a day of the week that matches the date it thinks it is, even if it isn't accurate about the date. I think the current value of AI is overstated, and I don't know how much the current methods can be improved
A human who can't figure out the date, or can't count the number of R's in "strawberry" is useless as an assistant for most knowledge tasks, but AI isn't analogous to human intelligence. It can be good at things that most humans are bad at while also being bad at things most humans are good at.
November 28, 2025 at 7:04 PM
A human who can't figure out the date, or can't count the number of R's in "strawberry" is useless as an assistant for most knowledge tasks, but AI isn't analogous to human intelligence. It can be good at things that most humans are bad at while also being bad at things most humans are good at.
I was responding to the implication that AI isn't very useful if it can't even tell the date, or that not being able to tell the date is evidence that it isn't very "intelligent" or accurate. For my current uses of AI—mostly coding—not knowing the day or being a year out of date is mostly irrelevant
November 28, 2025 at 7:03 PM
I was responding to the implication that AI isn't very useful if it can't even tell the date, or that not being able to tell the date is evidence that it isn't very "intelligent" or accurate. For my current uses of AI—mostly coding—not knowing the day or being a year out of date is mostly irrelevant
He may have been in college when the model was trained, so it's not that different from the date question. You can usually ask an LLM to do a web search to include up-to-date information. I'm not a heavy user of AI (mostly just used it for a few coding tasks), so I'm not sure how good it is at that.
November 27, 2025 at 5:32 PM
He may have been in college when the model was trained, so it's not that different from the date question. You can usually ask an LLM to do a web search to include up-to-date information. I'm not a heavy user of AI (mostly just used it for a few coding tasks), so I'm not sure how good it is at that.
(Some) Humans can figure out the day of the week by knowing the date, but you still rely on external stimuli to figure out the date—mostly by the changing amount of light outside.
November 27, 2025 at 4:43 PM
(Some) Humans can figure out the day of the week by knowing the date, but you still rely on external stimuli to figure out the date—mostly by the changing amount of light outside.
But humans also don't know the current date except through external stimuli, so that doesn't really seem relevant to intelligence. 3/2
November 27, 2025 at 12:36 AM
But humans also don't know the current date except through external stimuli, so that doesn't really seem relevant to intelligence. 3/2
The model doesn't "learn" and grow in knowledge when it does web searches. It can't gain new insights or knowledge by interacting with users, outside of a small context window. So in that way it's inferior to human intelligence. 2/2
November 27, 2025 at 12:23 AM
The model doesn't "learn" and grow in knowledge when it does web searches. It can't gain new insights or knowledge by interacting with users, outside of a small context window. So in that way it's inferior to human intelligence. 2/2
I just now asked ChatGPT the date and day, and it gave it to me correctly. The LLM can access information from outside itself, such as through a web search. But from my understanding, the model itself is static, and it's expensive to update it by generating a new model, much like a printed book. 1/2
November 27, 2025 at 12:22 AM
I just now asked ChatGPT the date and day, and it gave it to me correctly. The LLM can access information from outside itself, such as through a web search. But from my understanding, the model itself is static, and it's expensive to update it by generating a new model, much like a printed book. 1/2
I'm a bad coder, and have been for over 20 years. I'm OK with code that takes 1 second to do something instead of 0.01 sec (I make tools for myself, not for a server that has millions of clients).
The code I ask LLMs to generate almost always works the first time, and if not, after a follow up or 2.
The code I ask LLMs to generate almost always works the first time, and if not, after a follow up or 2.
November 26, 2025 at 8:40 PM
I'm a bad coder, and have been for over 20 years. I'm OK with code that takes 1 second to do something instead of 0.01 sec (I make tools for myself, not for a server that has millions of clients).
The code I ask LLMs to generate almost always works the first time, and if not, after a follow up or 2.
The code I ask LLMs to generate almost always works the first time, and if not, after a follow up or 2.
What an odd criticism. That's like criticizing the usefulness of books because they can't tell you the current date.
November 26, 2025 at 7:42 PM
What an odd criticism. That's like criticizing the usefulness of books because they can't tell you the current date.
🙋I've sucked at coding for over 20 years. So maybe that's why this year I've found some occasional uses for LLM generated code. I don't distribute my code to others (directly or in compiled form), so you don't have to worry about that.
November 26, 2025 at 7:21 PM
🙋I've sucked at coding for over 20 years. So maybe that's why this year I've found some occasional uses for LLM generated code. I don't distribute my code to others (directly or in compiled form), so you don't have to worry about that.
Ah.. interesting. If you click the </> icon, it has the correct result in the "Result", but it isn't able to just copy that back into the normal reply.
November 26, 2025 at 6:54 PM
Ah.. interesting. If you click the </> icon, it has the correct result in the "Result", but it isn't able to just copy that back into the normal reply.
ChatGPT has several modes/models. It used Python code to generate the reply to your query. It's possible that thasks' query went to a model not using code to generate a reply.
November 26, 2025 at 6:42 PM
ChatGPT has several modes/models. It used Python code to generate the reply to your query. It's possible that thasks' query went to a model not using code to generate a reply.
And... I'm blocked for offering an example that goes against their narrative that the only people who speak positively about AI are boosters that have ulterior motives.
November 26, 2025 at 6:22 PM
And... I'm blocked for offering an example that goes against their narrative that the only people who speak positively about AI are boosters that have ulterior motives.
LLM coding tools have been beneficial to me at my job. My job has nothing to do with AI. It's not even really a coding job. I use code to automate portions of my tasks and bring in more data so I have more context when making decisions. No AI in my bio or job description. Started using it this year.
November 26, 2025 at 5:59 PM
LLM coding tools have been beneficial to me at my job. My job has nothing to do with AI. It's not even really a coding job. I use code to automate portions of my tasks and bring in more data so I have more context when making decisions. No AI in my bio or job description. Started using it this year.