Do you think it is ethical to ask ChatGPT, Co-Pilot or oher GenAI products to create PM materials for your specific project?

One follow-up on the OP: GPT does not have a "data lake" in any sense that it stores data that is or resembles the original training data. All it stores are probabilities and tokens, completely disaggregated from the original content units. I don't want to be a nit-picker, but it is a fundamental aspect of the LLM architecture that is easily misunderstood, which drives misconceptions about the nature of human-AI interaction. When we cause ChatGPT to generate output, it does not retrieve a block of pre-written text (e.g., a fragment from Wikipedia) or even "assemble" the response from multiple fragments of recognisable content. It "assembles" the response at the "word" level, based on the input and vast numbers of probability parameters. This aspect of the LLM architecture removes consideration of factors such as citing the original source of the material or plagiarisation (except perhaps in the strict academic sense of the word). There is no single source from which the content is generated. In one way it's an extremely simple concept but the implementation is mind-boggling in scope, complexity and resource requirements.

This is relevant to the OP because we shouldn't interpret what ChatGPT generates in the same way as any other form or method of using someone else's work. It can be much much worse or better, depending on the use case.

Hi George, you've made some excellent points and I tend to agree with them. You're definitely right about the AI marketplace: everything is being "AI-enabled" at the moment, either in functionality or in branding. This in itself is a factor for content creators and professional workers in that the options to *not* use AI are narrowing fast. Your question: "Could the user have created the content if given the time?” is absolutely on point - it's a great question because it's a proxy for many others.

(On a side note, I've been in the IT/PM industry for over 4 decades and it's been alarming and amusing to see how corporate (or even internal) presentations get reworked to include every new fad that comes along. Global replace "Hosting" with "Cloud" for example, and many more.)

My problem with self-disclosure is that it's not an objective environment. I'm very happy to self-assess my use and decide whether i need to disclose, in the event that I did use ChatGPT in the product of my work. The questions being asked by people, platforms, educational institutions and more are far too crude to capture the nuances of these posts, and the downside is too large to engage in the questions literally.

Interestingly, I just finished some research for my Masters in PM (on the use and attitudes of ChatGPT and Project Management practitioners) - in the questions on employer awareness and attitude there was very much a "laissez-fair" majority who didn't know and/or didn't encourage or discourage its use.

This is not the same in platforms such as Amazon KDP, and more. Everyone is asking effectively "has AI touched this at all"m with the risk of being banned or rejected if the answer is yes. I understand the concerns about being swamped with "book spam"

I've developed some tools for self assessment (can't post the images here) but I'll post a link once I've finished the posts.

Well, George and Adam, I think I will refrain from the "decades in the IT industry" discussion, other than to say that have some difficulty deciding whether I am even in the IT industry or the project management industry. So counting the decades in each becomes a fuzzy exercise when trying to determine overlap and where one begins and the other ends. Indeed, some of the most enjoyable projects I managed had far more to do with policy, process and the management of change than with IT.

Thanks for your contributions to this discussion on the ethics of Gen AI. Even since I began the discussion only few months ago, things have changed.

Speaking of existing products that have been around for years claiming to suddenly be AI-based, it occurs to me that as we are comparing GenAI products to tools like Grammarly, spreadsheets, project scheduling tools and the like, we are perhaps missing a glaring difference - most of these other tools are based on rules, or bodies of accepted knowledge.

There are rules of grammar, rules of mathematics, rules of critical path calculations and so on. So it becomes far easier to recognize questionable answers, or to prove a tool is producing an inaccurate result should we feel the need to assess it.

But most of these pre-AI tools are almost always right.

Is GenAI based on rules? I think we can all agree that it is not, or at least only to the extent that some of what it was "taught" might have been rule-based. So, it is often very difficult to determine whether the result of a GenAI query is right, wrong, or somewhere in between.

No rules.

Nothing to do with Eurovision, by the way. ;)

I agree with the grey hair/wisdom metric for sure. Ignoring hubris, I did some extended calculations on my career a few years back and wrote it up on LinkedIn with a link at the bottom to a part 2 with more funky charts. I updated this last year on Medium. Links below:

I think it means simply that I've had more "Edison Moments" (found more ways not to run projects) than most other folks

Hi Mike, thanks for kicking off this thread. And, at least from where I sit, much of that past is fuzzy :-)

On the AI thing, absolutely agree that the current AI products are humungously different to previous tools and that there are no rules. Even the history of AI has limited guidance: Ethan Mollick in HBR called the release of ChatGPT the "tipping point" for AI, which changed all the rules of AI application (vast pre-training, frictionless access and generic human-AI interaction, which neither constrains nor informs the user of underlying functionality). That's the reason there's so much noise. I think most users would be better off treating ChatGPT as a "black box" beyond a few heuristics on strengths and weaknesses. Users should treat the outputs on their merits rather than investing both the tool and the results with some sort of aspirational (magical) capabilities.

When there are no defined rules, we need to look to our own rules and principles to guide us.

Leaving aside the long history of content creators using assistants, tools and muses (the acknowledgement of which seems to be at the discretion of the creator), I view my prompts in the same way a developer views code: it is my skill and knowledge that is driving a tool to perform certain actions. Not perfect but it will do me for now.

I think it was George Freeman who posed the benchmark question: "Could the author have produced the work unaided, given time (or other resources)?"

So, as a basic rule, as long as I am driving it and it's not driving me, then I don't think it's anyone's business how I produce my works. I maintain my right to acknowledge the support I've had in any way I decide. That means I take full accountability for the works, including any errors (which is the usual formula for authors crediting their assistants) and will accept the bouquets or brickbats that come as a result. There are no "the AI got it wrong" or "the AI ate my homework" excuses.

As I said earlier, if the landscape for disclosure evolves to be more nuanced and less prejudicial, then my view on disclosure will evolve too.

Hi Mike,

I'm new to the discussion boards but have been a long-time PMI member since 2008! I've become a huge fan of GenAI over the last year and it's transformed how I work. I mainly use Gemini Advanced to refine my work and get fresh ideas. My prompting skills have come a long way and the results are now impressive (though fact-checking is a must, as it's not perfect).

Here's my take on the pros and cons:

Pros:

Cons:

Ethical Considerations (It's quite simple really):

Overall, GenAI is incredibly powerful, but responsible use is key. By understanding its limitations and ethical implications, we can leverage it to boost our productivity and creativity. Any detractors of AI are merely going to have to adapt; it truly is the future that businesses are embracing.

Hello,

I don't think it's a question of ethics. We have wonderful tools that help solve complex problems and save a lot of time. We should definitely use them. However, the issue arises when we blindly follow their output. As responsible project managers, we must always verify the results and apply logical analysis.