Community Specialist| Project Management InstituteNewtown Square, PA, United States
As community managers and moderators, PMI's Online Community Engagement Team is tackling a new frontier as community members here on our discussion forums have been experimenting with composing responses using AI tools like ChatGPT.
We've gotten feedback from the community that you'd like to know if and when a community member is using an AI tool to compose a discussion post or comment.
How can we support your experience as you engage here on the discussion forums on ProjectManagement.com? Would you like to know if someone is using ChatGPT to craft their post?
Should it become a requirement of our Online Community Terms and Conditions to disclose the use of an AI tool to compose a discussion post or comment?
I simply ignore the posts that appear obviously AI-generated and then tweaked (or not), especially when I see several posted in a short timeframe. That does reduce user engagement however as there could be valuable posts before those that are missed.
As for imposing a requirement, PMI would need to think carefully about how to define that requirement. AI "use" is very much on a spectrum and good requirements are concise and not ambiguous. For researching subjects, I find AI useful in performing an initial literature study providing me with subjects to research in more detail. Newer generations of writing aids are very good at language checking and stylistic suggestions which are especially useful for those writing in a non-native language.
Does that require disclosure? A common sense answer might be that those types of AI usage are not the intent and using grammar-checking tools does not require disclosure. Defining the boundaries as a requirement however is not easy, and the technology is still in a rapid development phase. Grammarly will not only perform basic word processor-type suggestions but also generate outlines and more and is not prone to being caught by AI detection. Where is that line drawn?
(Disclosure: I used some AI editing this post.) Saving Changes...
I agree with Keith that AI uses exists on a continuum. If AI is used to help edit or tune wording, or if the member used it to help draft their contribution but then made significant manual updates to it themselves, that is fine. On the other hand if it is a verbatim (or close to) copy & paste, that should be flagged as "AI generated".
However, having got used to posts of the ilk of "I agree with <XYZ>", I might prefer AI generated posts to those :-)
Kiron
...
2 replies by Mike Frenette and Rami Kaibni
Dec 15, 2023 10:15 AM
Rami Kaibni
...
Kiron, you cracked me up with the “I agree with” answer which we’ve been suffering from for a few years now. I too, would take AI generated responses over those answers!
Mar 19, 2024 5:41 PM
Mike Frenette
...
Or posts that say nothing but "Thanks" with no reference to the content of the post to which they are replying.
Saving Changes...
George FreemanThought Leader | Author | Architect| Florida, United States
Hi Laura,
Although not a commonly known fact, “OpenAI” requires (per their sharing and publication policy) consumers/users of their content to “make a disclosure” when they publish/post that content to other platforms. Specifically, they require the disclosure to state that the content is AI-generated and then further ask that the disclosure statement be written so that no reader could reasonably misunderstand this fact.
In addition, they ask that when publications or postings are “co-authored,” meaning that AI-generated content was used “in part,” that the author states AI’s role in formulating the content, again in a way that no reader could reasonably misunderstand this fact.
That said, I know some may interpret their policy differently.
The value proposition of this platform draws its strength from human-expert opinions. These opinions are not just facts but represent knowledge from the whole continuum of the “human experience.”
In my opinion, AI-generated content, through its construct, dilutes the human-expert opinion that makes this site what it is.
With its professionally polished grammar and style, Generative-AI content differs from the “I agree with…” or the “Thanks for sharing” noise-based posts, as it has a “perception of validity” through its presentation. In other words, a contextually wrong listing of generalized bullets has a real chance of being accepted as fact by someone newly “starting out” in the project management field.
Why not enhance the terms and conditions and mitigate these risks through policy? Saving Changes...
Senior Projects Manager | Field & Marten AssociatesNew Westminster, British Columbia, Canada
Dec 14, 2023 6:41 PM
Replying to Kiron Bondale
...
Laura -
I agree with Keith that AI uses exists on a continuum. If AI is used to help edit or tune wording, or if the member used it to help draft their contribution but then made significant manual updates to it themselves, that is fine. On the other hand if it is a verbatim (or close to) copy & paste, that should be flagged as "AI generated".
However, having got used to posts of the ilk of "I agree with <XYZ>", I might prefer AI generated posts to those :-)
Kiron
Kiron, you cracked me up with the “I agree with” answer which we’ve been suffering from for a few years now. I too, would take AI generated responses over those answers!
Senior Projects Manager | Field & Marten AssociatesNew Westminster, British Columbia, Canada
Laura, my thoughts are in line with my fellow colleagues here and George brings very good point as well. If someone is using AI to improve upon their own manually drafted answer then this is great but if they are just copying and pasting the answer in bulk as is or with minor edits then, while maybe informative to the community, yet, it doesn’t reflect human experience which like George said, goes against the value proposition of this platform. I used AI few times and did find some of the answer very interesting so decided to share them, however, I clearly disclosed that I did ask ChatGBT and I think everyone should disclose this as I believe it is an ethical and professional obligation.
Will you be able to regulate this on this platform? Not sure to be honest. The question we need to ask is:
How is this different than someone checking Google and Wikipedia then copying and pasting the answer? I don’t think it’s much different so in both cases I believe we as professionals, do have the ethical obligation to disclose that we posted something not of our own creation and not get credit for the works of a third party.
Great to read the above, and wanted to share one of my recent experiences.
Not as far as yesterday, I have prompted Bard and used the text for a post on company Viva Engage site. The Text felt so well structured and informative (and complete). I would have never been able to deliver it in my own words btw!
Nevertheless, I have added few important (in my view) things:
1/ I tailored the text and changed some sentences that I felt not very convenient for me..; it took me 10 minutes, better than 30 min writing the post.
2/ I added some cosmetics and links to some internal resources.. things that I would not ask Bard to do on my behalf
and most importantly,
3/ I acknowledged to colleagues that most of the text was AI generated: this is key from Ethics and professional obligation PoV.
I did the above in order to make use of available GenAI tools, with human in the core...
As Professionals, I am not sure we are here to impress others by posting ready texts whether it comes from GEN AI, a book, a blog... a minimum of honesty and transparency by indicating the source of any content is required. I see it more a personal initiative that is required, and a best practice to put in place, rather than a hard policy to implement...
Any way Human intelligence is able to make the difference between generated Text and authentic text in a time of few seconds in my opinion.
...
1 reply by George Freeman
Dec 15, 2023 2:32 PM
George Freeman
...
Hi Ali,
I generally agree that “human intelligence” can recognize the patterns in content rendered from generative AI; however, I would add the proviso that this is true based on “current rendered patterns” and may not be accurate in future near-term versions. Additionally, the reader needs to be “aware of and watching” for these patterns for this statement to be true.
I’ve seen postings where users respond to generative AI content with the sincere belief that it was thought out and tailored to their question/inquiry. Hence, we have an issue of awareness regarding this issue.
On a different note, A policy states “what is the standard,” with one’s “personal initiative” being whether or not they abide by that standard. The issue of how you regulate the standard is a different question altogether.
Saving Changes...
George FreemanThought Leader | Author | Architect| Florida, United States
Dec 15, 2023 11:51 AM
Replying to Ali CHAIEB
...
Great to read the above, and wanted to share one of my recent experiences.
Not as far as yesterday, I have prompted Bard and used the text for a post on company Viva Engage site. The Text felt so well structured and informative (and complete). I would have never been able to deliver it in my own words btw!
Nevertheless, I have added few important (in my view) things:
1/ I tailored the text and changed some sentences that I felt not very convenient for me..; it took me 10 minutes, better than 30 min writing the post.
2/ I added some cosmetics and links to some internal resources.. things that I would not ask Bard to do on my behalf
and most importantly,
3/ I acknowledged to colleagues that most of the text was AI generated: this is key from Ethics and professional obligation PoV.
I did the above in order to make use of available GenAI tools, with human in the core...
As Professionals, I am not sure we are here to impress others by posting ready texts whether it comes from GEN AI, a book, a blog... a minimum of honesty and transparency by indicating the source of any content is required. I see it more a personal initiative that is required, and a best practice to put in place, rather than a hard policy to implement...
Any way Human intelligence is able to make the difference between generated Text and authentic text in a time of few seconds in my opinion.
Hi Ali,
I generally agree that “human intelligence” can recognize the patterns in content rendered from generative AI; however, I would add the proviso that this is true based on “current rendered patterns” and may not be accurate in future near-term versions. Additionally, the reader needs to be “aware of and watching” for these patterns for this statement to be true.
I’ve seen postings where users respond to generative AI content with the sincere belief that it was thought out and tailored to their question/inquiry. Hence, we have an issue of awareness regarding this issue.
On a different note, A policy states “what is the standard,” with one’s “personal initiative” being whether or not they abide by that standard. The issue of how you regulate the standard is a different question altogether. Saving Changes...
Community Specialist| Project Management InstituteNewtown Square, PA, United States
Thank you, everyone, for your thoughtful feedback and insights! This will be extremely helpful as we start to refine our policies and possibly develop new ones. Saving Changes...
Markus KopkoAI Enabler for Project & Program Mgmt | Founder PMotion.ai / The PM
AI Coach| PMotion.aiHamburg, Hamburg, Germany
Dec 15, 2023 10:15 AM
Replying to Rami Kaibni
...
Kiron, you cracked me up with the “I agree with” answer which we’ve been suffering from for a few years now. I too, would take AI generated responses over those answers!
PMI Team Member
Community Champion
