Mike FrenetteManager, IT PMO| Halifax Water (retired)Halifax, Nova Scotia, Canada
Many generative AI products will take your input and apply it to what it finds in its data lake, then generate something that will save you time and effort. But will it be accurate? Will it be unbiased? Will it be the same as you would have written? How do you feel about that? Is it ethical? If so, how do you make it our own? Saving Changes...
What Is Fine-Tuning? Fine-tuning serves as a method for applying transfer learning, where an existing deep learning model has already been trained to perform well on a given set of general tasks, and is further refined using new data so that it can perform better on similar, more specific tasks. This a core concept behind GenAI. A simple alternative definition is that fine-tuning involves updating an existing intelligent computer program with new knowledge derived from a previously unseen document repository or dataset. Machine learning (ML) fine-tunes a pre-trained model to perform a customized task by making a minor tweaks or adding more layers to a model’s architecture while maintaining the core struc- ture of the original model and improving reliability to generate a desired output. Part IV will give a comprehensive explanation of the important steps and principles of how to utilize fine-tuning as part of a secure and ethical approach to project management. What Is Customized Modeling? Customized Modeling: It is used when adapting existing machine learning models to specific data or use cases. This can involve techniques such as transfer learning, where a pre-trained model is fine-tuned on a new dataset. Customizing a model can also mean adjusting its architecture or hyper-parameters to per- form better on specific tasks.
Fine-tuning is a technique in transfer learning that refines an existing deep learning model using new data to improve its performance on specific tasks. It involves updating the model with new knowledge from previously unseen data, making minor adjustments to its architecture while retaining its core structure. Customized modeling, on the other hand, adapts machine learning models to specific data or use cases, often through techniques like fine-tuning or adjusting hyper-parameters. Both methods aim to enhance model performance and reliability for specialized applications.One major impactful benefit to project managers is the ability to leverage fine-tuning for creating highly customized and efficient AI tools, which can significantly enhance project planning, decision-making, and overall productivity. Thanks & Regards Booma Pugazhenthi M.Sc., CPM, LEED AP, PMP Saving Changes...
Thanks for the additional item, George. I agree these sections out of the PMI Code of Ethics apply to the use of GenAI in project management. Of course, what is generated could be a half-truth and will very often be out of context, which could result in errors and omissions.
The well worn saying, "A fool with a tool is still a fool." surely applies here. Blind use of GenAI output without review and modification for accuracy, completeness and context would surely make the person doing so an [unethical] fool.
ChatGPT, while not a substitute for expert knowledge in Project Management, can significantly enhance a project manager's efficiency by providing quick, tireless assistance and idea generation. However, its outputs must be carefully reviewed and contextualized to avoid errors and misapplications. The tool's greatest benefit lies in saving time and facilitating rapid ideation, allowing project managers to focus more on strategic tasks and decision-making. Proper use of ChatGPT can lead to better-managed projects, but reliance on its output without critical evaluation can lead to significant mistakes and ethical lapses. Thanks & Regards Booma Pugazhenthi M.Sc., CPM, LEED AP, PMP Saving Changes...
Correct me if I'm missing anything. Let me rephrase your question: is it ethical to use a TOOL to save you time and wasted effort?
ChatGPT and other LLMs are tools. While these tools can save time and effort, we still have to provide the prompts and priming (the input).
It's another issue altogether if you claim you had no help in your output. That would be dishonest, and thus unethical.
Also it's another case if you use ChatGPT to generate false court cases like a lawyer did. That's clearly unethical.
But if clients specifically state they only want human input on their projects, then of course you don't use help of LLM tools on such projects. (Maybe you may even want to avoid such clients altogether)
The answer is, simply, it depends. I would generally agree that it is absolutely ethical to use AI to develop things like like creating a type of project management template, developing a first draft project charter given a well crafted prompt, or developing a first draft project charter - that would all be tailored appropriately.
However, if using AI to produce an output, or an input required for a project, it is definitely advisable to proceed with caution. Using AI to provide ideas as to what success MIGHT look like, would absolutely be ethical.
Asking AI to provide the final output, and simply hand that off to the end consumer, would absolutely be unethical (and granted this is a gross oversimplification).
As one example, with very limited resources, a very short timeline, I have used AI to develop code on a migration project - quite successfully! I am not a coder, and did not have a coder readily available on my project. However, I used AI successfully to 1) determine whether AI COULD be used to develop the code, 2) provide a first draft of very simple code to deliver the results I needed, 3) provide a set of data that I could test the code with (which worked!). Finally, as a final step, given minimal resources, I found someone I could share the code with to validate, and independently test, and modify the code that was ultimately required to deliver the results I needed. Saving Changes...
"The most exciting phrase to hear in science, the one that heralds new discoveries, is not Eureka! (I found it!) but rather, 'hmm.... that's funny...'"
