Timeline for GPT on the platform: Data, actions, and outcomes
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 9, 2023 at 22:18 | comment | added | anx | @Trilarion raw score is heavily confounded by answer speed, which I expect is abnormal for tool-assisted speedrunners. It probably is measurable by score though, when applying a suitable filter on whose upvotes to look at. | |
| Jun 9, 2023 at 11:44 | comment | added | NoDataDumpNoContribution | To check the ideas in this answer, one could for example look at the score of the answers. Maybe the fraction of positively scoring answers did increase. That would backup the idea that low quality answerers took refuge in using GPT. | |
| Jun 8, 2023 at 19:46 | comment | added | anx | Maybe someone speaking the language better than I do can add the words "representative sample" and "confounder" in here, because that seems to be the core of the arguments: GPT offenders being compared to totals as if they were just any other user. Yet we should absolutely expect them to stand out in statistics for reasons impacting both their human and their entirely made-up contributions, making this likely measurable. | |
| Jun 7, 2023 at 23:26 | history | edited | gnat | CC BY-SA 4.0 |
minor wordsmithing
|
| Jun 7, 2023 at 23:03 | history | edited | gnat | CC BY-SA 4.0 |
minor typo corrected
|
| Jun 7, 2023 at 22:46 | history | answered | gnat | CC BY-SA 4.0 |