Alasdair Stewart
@abestew.bsky.social
Sociologist | Socialist | Neurodivergent | Linux & FLOSS advocate | he/they
ChatGPT 5.1 now seems to more consistently just load the tidyverse rather than individual packages from it - but varies in whether does so, including across responses within the same chat. The tendency towards "package::function" code in places remains.
November 13, 2025 at 12:04 PM
ChatGPT 5.1 now seems to more consistently just load the tidyverse rather than individual packages from it - but varies in whether does so, including across responses within the same chat. The tendency towards "package::function" code in places remains.
For R, the biggest tell is ChatGPT's gravitation towards needlessly customisation of graphs. It loves theme_minimal() but is also horrendously inconsistent in applying built-in themes and the needless additional customisation it adds, including for ggplots within the same response.
November 13, 2025 at 12:04 PM
For R, the biggest tell is ChatGPT's gravitation towards needlessly customisation of graphs. It loves theme_minimal() but is also horrendously inconsistent in applying built-in themes and the needless additional customisation it adds, including for ggplots within the same response.
When you give it insufficient info about your data set, instead of "# Assuming..." - which people copied and pasted not realising it was example code - it now seems to more clearly sign-post "# Example: ...". The love for end of line comments remains, though seems to be toned down.
November 13, 2025 at 12:04 PM
When you give it insufficient info about your data set, instead of "# Assuming..." - which people copied and pasted not realising it was example code - it now seems to more clearly sign-post "# Example: ...". The love for end of line comments remains, though seems to be toned down.
Ha, and there is a another rule of three at end of that second image didn't originally notice. Once start highlighting the patterns it definitely reads more and more like they have fed it training data to avoid some usual tells, but it's more 'masked' than 'removed'.
November 13, 2025 at 11:08 AM
Ha, and there is a another rule of three at end of that second image didn't originally notice. Once start highlighting the patterns it definitely reads more and more like they have fed it training data to avoid some usual tells, but it's more 'masked' than 'removed'.
Interestingly, noticed in finding these examples there is more variation in number of items in places. Some of these though are rule of three with 'and so on' or 'even ...' (and variant phrases) stuck on at the end. Also outlined in blue in 2nd image the repetitive phrases can also get in lists.
November 13, 2025 at 11:05 AM
Interestingly, noticed in finding these examples there is more variation in number of items in places. Some of these though are rule of three with 'and so on' or 'even ...' (and variant phrases) stuck on at the end. Also outlined in blue in 2nd image the repetitive phrases can also get in lists.
ChatGPT's love of rule of three - including in 2nd image a rule of three at the end of a rule of three. 3rd and 4th images are examples of bullet-point list rule of three.
November 13, 2025 at 11:05 AM
ChatGPT's love of rule of three - including in 2nd image a rule of three at the end of a rule of three. 3rd and 4th images are examples of bullet-point list rule of three.
Ah, it looks like it's "etc" if in brackets and "and so on" when not using brackets.
November 13, 2025 at 10:53 AM
Ah, it looks like it's "etc" if in brackets and "and so on" when not using brackets.
The switch-up between em-dash asides and bracket asides is more noticeable across whole response than paragraph by paragraph, but first image shows an example. Second image has em-dashed rule of three aside, this is within same response as first image that has bracketed rule of three aside.
November 13, 2025 at 10:51 AM
The switch-up between em-dash asides and bracket asides is more noticeable across whole response than paragraph by paragraph, but first image shows an example. Second image has em-dashed rule of three aside, this is within same response as first image that has bracketed rule of three aside.
Italics and bold variations. Can see in second image the new found fondness for brackets as well.
November 13, 2025 at 10:45 AM
Italics and bold variations. Can see in second image the new found fondness for brackets as well.
Colons are the new em-dashes.
November 13, 2025 at 10:40 AM
Colons are the new em-dashes.
Initial work on visual refresh of the SGSSS NVivo Guidance pages now done.
sgsssonline.github.io/nvivo-guidan...
sgsssonline.github.io/nvivo-guidan...
November 9, 2025 at 2:19 PM
Initial work on visual refresh of the SGSSS NVivo Guidance pages now done.
sgsssonline.github.io/nvivo-guidan...
sgsssonline.github.io/nvivo-guidan...
There is disturbing number of journal articles promoting LLMs as making quantitative analysis 'more accessible', removing need for students to learn how to code, and similar arguments.
My response - "Have you never heard of snippets?"
My response - "Have you never heard of snippets?"
November 6, 2025 at 9:43 AM
There is disturbing number of journal articles promoting LLMs as making quantitative analysis 'more accessible', removing need for students to learn how to code, and similar arguments.
My response - "Have you never heard of snippets?"
My response - "Have you never heard of snippets?"
OK NVivo for Windows requires installing Microsoft Visual C++ Redistributable and to disable child-locks on macOS command+click and Open the app.
Ridiculous amount of faff, but I now have... hell in stereo vision...
Ridiculous amount of faff, but I now have... hell in stereo vision...
November 4, 2025 at 4:16 PM
OK NVivo for Windows requires installing Microsoft Visual C++ Redistributable and to disable child-locks on macOS command+click and Open the app.
Ridiculous amount of faff, but I now have... hell in stereo vision...
Ridiculous amount of faff, but I now have... hell in stereo vision...
How do people get anything done on Windows and macOS? NVivo crashes on launch without any error messages and looks like I have to figure out how to turn the child-locks off on macOS
November 4, 2025 at 4:00 PM
How do people get anything done on Windows and macOS? NVivo crashes on launch without any error messages and looks like I have to figure out how to turn the child-locks off on macOS
Giving quickemu a try. Images for both downloaded OK, though macOS refused to work after installing so trying again. Windows couldn't find hard drive, but seems to be installing now after copy/paste of bash script from GitHub.
November 4, 2025 at 3:31 PM
Giving quickemu a try. Images for both downloaded OK, though macOS refused to work after installing so trying again. Windows couldn't find hard drive, but seems to be installing now after copy/paste of bash script from GitHub.
Progress on my first attempt at knitting.
November 3, 2025 at 9:50 AM
Progress on my first attempt at knitting.
GenAI coding is over-hyped & nowhere near as good as claimed. It's decent for short scripts - but only when the code quality doesn't matter that much.
Screenshot is output from Python script to identify whether and when best to open windows to ventilate.
Screenshot is output from Python script to identify whether and when best to open windows to ventilate.
October 26, 2025 at 7:36 AM
GenAI coding is over-hyped & nowhere near as good as claimed. It's decent for short scripts - but only when the code quality doesn't matter that much.
Screenshot is output from Python script to identify whether and when best to open windows to ventilate.
Screenshot is output from Python script to identify whether and when best to open windows to ventilate.
Giving a 15 year old tea a try today.
So heavily compacted gave up trying to break it apart and instead steeping whole tuo for *very* short amounts of time and slowly peeling it apart in layers after each steep.
So heavily compacted gave up trying to break it apart and instead steeping whole tuo for *very* short amounts of time and slowly peeling it apart in layers after each steep.
October 10, 2025 at 10:04 AM
Giving a 15 year old tea a try today.
So heavily compacted gave up trying to break it apart and instead steeping whole tuo for *very* short amounts of time and slowly peeling it apart in layers after each steep.
So heavily compacted gave up trying to break it apart and instead steeping whole tuo for *very* short amounts of time and slowly peeling it apart in layers after each steep.
ChatGPT now has a 'use cases' page for students, where it is unsurprising to find the first is also the worst.
If maintaining awareness of its flaws, genAI can be a semi-decent sounding board, but it gives some truly awful (and opinionated) advice on what to improve.
chatgpt.com/use-cases/st...
If maintaining awareness of its flaws, genAI can be a semi-decent sounding board, but it gives some truly awful (and opinionated) advice on what to improve.
chatgpt.com/use-cases/st...
September 6, 2025 at 3:09 PM
ChatGPT now has a 'use cases' page for students, where it is unsurprising to find the first is also the worst.
If maintaining awareness of its flaws, genAI can be a semi-decent sounding board, but it gives some truly awful (and opinionated) advice on what to improve.
chatgpt.com/use-cases/st...
If maintaining awareness of its flaws, genAI can be a semi-decent sounding board, but it gives some truly awful (and opinionated) advice on what to improve.
chatgpt.com/use-cases/st...
Obsidian graph timelapse animation of my notes created and links made between them over past five years.
September 5, 2025 at 9:46 AM
Obsidian graph timelapse animation of my notes created and links made between them over past five years.
Edit: Just noticed I forgot to add the final screenshot. 🤦
August 27, 2025 at 11:29 AM
Edit: Just noticed I forgot to add the final screenshot. 🤦
A simple naive question and it writes a full draft with citations added.
The two options are now effectively "do you want me to plagiarise that for you" or "do you want to manually mask the plagiarism" - phrased in way that suggests its fine to paraphrase & submit AI slop.
The two options are now effectively "do you want me to plagiarise that for you" or "do you want to manually mask the plagiarism" - phrased in way that suggests its fine to paraphrase & submit AI slop.
August 27, 2025 at 10:55 AM
A simple naive question and it writes a full draft with citations added.
The two options are now effectively "do you want me to plagiarise that for you" or "do you want to manually mask the plagiarism" - phrased in way that suggests its fine to paraphrase & submit AI slop.
The two options are now effectively "do you want me to plagiarise that for you" or "do you want to manually mask the plagiarism" - phrased in way that suggests its fine to paraphrase & submit AI slop.
Next it offers to identify relevant literature. Whilst it speaks at times of 'research base' and 'reading list', it explains where to cite for what claim and without having read anything at all a reference list 'you can plug straight into your essay'. And it gets worse from here...
August 27, 2025 at 10:55 AM
Next it offers to identify relevant literature. Whilst it speaks at times of 'research base' and 'reading list', it explains where to cite for what claim and without having read anything at all a reference list 'you can plug straight into your essay'. And it gets worse from here...
Can now basically "Prompt OK to continue doing it for me".
Across each step it frames the responses in a way that would suggest this is all perfectly fine. Indeed, it starts blurring authorship by suggesting the user merely slots content in and speaks of the thesis it generated as "your thesis".
Across each step it frames the responses in a way that would suggest this is all perfectly fine. Indeed, it starts blurring authorship by suggesting the user merely slots content in and speaks of the thesis it generated as "your thesis".
August 27, 2025 at 10:55 AM
Can now basically "Prompt OK to continue doing it for me".
Across each step it frames the responses in a way that would suggest this is all perfectly fine. Indeed, it starts blurring authorship by suggesting the user merely slots content in and speaks of the thesis it generated as "your thesis".
Across each step it frames the responses in a way that would suggest this is all perfectly fine. Indeed, it starts blurring authorship by suggesting the user merely slots content in and speaks of the thesis it generated as "your thesis".
End of response two and another either/or option. Notice how the first option is usually the "just do it for me" option.
Three prompts and already gone from merely asking to "brainstorm" to receiving responses with full write and cite, full outline & position to take, and now full paragraphs.
Three prompts and already gone from merely asking to "brainstorm" to receiving responses with full write and cite, full outline & position to take, and now full paragraphs.
August 27, 2025 at 10:55 AM
End of response two and another either/or option. Notice how the first option is usually the "just do it for me" option.
Three prompts and already gone from merely asking to "brainstorm" to receiving responses with full write and cite, full outline & position to take, and now full paragraphs.
Three prompts and already gone from merely asking to "brainstorm" to receiving responses with full write and cite, full outline & position to take, and now full paragraphs.