Talk:Collective intelligence

AI-generated content tag removed

Have checked for hallucinated content etc. and only found one shaky reference, which I removed. Happy to reinstate if other editors can give specific reasons and examples. Chalk giant (talk) 08:51, 7 December 2025 (UTC)

Revised discussion of 2022 correction to Riedl et al. - now more neutral

I’ve revised the paragraph discussing the 2022 correction to Riedl et al. (PNAS 2021) to align the article more closely with WP:NPOV and WP:V. The previous text correctly stated that the authors’ originally reported AVE value was corrected from 44% to 19.6%, but it inferred as a factual conclusion that this alone demonstrates that the evidence for a collective-intelligence factor is “quite weak or nonexistent.” That interpretation exceeds what the cited sources explicitly support. The PNAS correction confirms the AVE change but also states that the corrections “do not change the major conclusions of the paper.” In standard SEM practice an AVE below 0.50 raises concerns about convergent validity for the specific measurement model, but AVE is only one diagnostic; Riedl et al. report other relevant evidence, and independent critiques (e.g. Credé & Howardson 2017) indicate that the robustness of a single CI factor remains an active methodological debate. So the revised wording retains all corrected factual information, explains the meaning of AVE and common thresholds using a standard SEM source (Fornell & Larcker 1981), presents the implications neutrally, and acknowledges both the authors’ statement and the existence of critical analyses. This reflects the state of the literature more accurately without overstating conclusions, consistent with Wikipedia’s content neutrality standards. Chalk giant (talk) 07:58, 9 December 2025 (UTC)

Writing process for section 'Collective intelligence and leadership' with judicious use of a LLM

WP:LLMDISCLOSE:

Writing process for section ‘Collective intelligence and leadership'

Phase 1: A couple of weeks collecting articles on collective intelligence and leadership mainly via Google Scholar and reading them.

Phase 2: Selecting the articles I wanted to base the section on.

Phase 3: Crafting the prompt to use with an LLM, this time ChatGPT, free version.

Prompt: Research the academic, business and management literature for information on how collective intelligence relates to teams and leadership. You can use some or all of the papers with the links below but do not limit your research to these. Write a new section for the collective intelligence entry in Wikipedia titled ‘collective intelligence and leadership’. Use an encyclopedic tone and limit your verbosity to the absolute minimum. Do not hallucinate. Double check all references. Do not use DOIs. Think ultra hard. Do not be sycophantic.

A Multi-Level Leadership Spectrum for Collective Good https://scholarlypublications.universiteitleiden.nl/access/item%3A4247162/download

Understanding collective forms of leadership through text mining-based review of literature https://www.emerald.com/joepp/article/doi/10.1108/JOEPP-01-2024-0004/1254338/Understanding-collective-forms-of-leadership

Distributed leadership: taking a retrospective and contemporary view of the evidence base https://durham-repository.worktribe.com/OutputFile/1178513

‘Contestation, negotiation, and resolution’: The relationship between power and collective leadership https://www.researchgate.net/profile/Erica-Foldy/publication/366148586_%27Contestation_negotiation_and_resolution%27_The_relationship_between_power_and_collective_leadership/links/690cbb7da2b691617b6a27fb/Contestation-negotiation-and-resolution-The-relationship-between-power-and-collective-leadership.pdf

Systems Leadership: a qualitative systematic review of advice for policymakers https://pmc.ncbi.nlm.nih.gov/articles/PMC11862379/

Phase 4: Initial review of output. Accepted suggestion that one or two more scholarly references be provided. Review of new output.

Phase 5: Four to five hours rewriting and re-crafting output and checking references.

Phase 6: Adding the the new section and its paragraphs in Wikipedia one by one, each time checking and sometimes correcting references.

Phase 7: Clicking on each in-line reference, going to the paper and checking again whether the contents were well represented in the text. Doing rewrites based on that. Chalk giant (talk) 16:19, 12 December 2025 (UTC)

Phase 8: Researched and wrote extra paragraph on a more applied approach to CI and leadership based on a Harvard Business Review article in similar cycle as described above.

Phase 9: Refined, made the section less complex and easier to read on the basis of feedback from a non-expert test reader.

Incorrect flagging of LLM content by bot

See my conversation with another editor here before adding LLM flag again. Bot is incorrectly identifying content which has been worked on thoroughly and responsibly outside of Wikipedia in e.g. Google docs, and then pasted in, as LLM-generated. Bot cannot distinguish between content written responsibly and edited outside Wiki, and content uncritically pasted in from LLM without checks or edits. Bot needs refining. Is causing unnecessary work for editors. See: User talk:Chalk giant Chalk giant (talk) 06:41, 14 December 2025 (UTC)