SPONSORS

SPONSORS

On the Subject of AI, ChatGPT and Risk

 

LETTER TO THE EDITOR

By Martin Hopkinson

United Kingdom

19 May 2023

Ref:  Minelle, F., Stolfi, F. (2023). AI to support PM: a ChatGPT quality assessment (ß test); PM World Journal, Vol. XII, Issue V, May. Available online at https://pmworldlibrary.net/wp-content/uploads/2023/05/pmwj129-May2023-Minelle-Stolfi-AI-to-support-PM-a-ChatGPT-quality-assessment-3-1.pdf

Dear David

I enjoyed reading Federico Minelle and Franco Stolfi’s paper, AI to support PM: a ChatGPT quality assessment (ß test). They presented the results of an interesting test in which they described a particular project and its status to ChatGPT and then asked it to provide advice on how the project should be managed. I agree that this could be a useful way of testing an AI tool’s capability as an administrative assistant.

I suspect that an even more popular use of Chat GPT will be to use it as an educational assistant. I also suspect that many users will provide the tool with less information or with information that is not so concise and well ordered. With this in mind I carried out my own experiment by devising a simple question that might be used in either an educational context or that of using Chat GPT as a project assistant. For reasons I have identified later, I have a significant concern about its response. But first, here is my question and ChatGPT’s response.

My Question: How should I perform a project risk analysis on my project?

Chat GPT’s response: Performing a project risk analysis involves several steps. Here’s a general outline to help you get started:

  1. Identify potential risks: Make a list of all possible risks that could impact your project. Consider internal factors (e.g., resource constraints, skill gaps) and external factors (e.g., market changes, regulatory issues).
  2. Assess the likelihood: Evaluate the probability of each risk occurring. Assign a numerical rating or use a qualitative scale (e.g., low, medium, high) to gauge the likelihood. Consider historical data, expert opinions, and available information.
  3. Evaluate the impact: Determine the potential consequences of each risk if it were to occur. Assess the impact on project objectives, timelines, costs, quality, and stakeholders. Again, you can use numerical ratings or qualitative scales to assess the severity.
  4. Prioritize risks: Prioritize the identified risks based on their likelihood and impact. You can use techniques like the Risk Probability and Impact Matrix or the Risk Scoring Model to prioritize them effectively.
  5. Develop response strategies: For high-priority risks, create appropriate response strategies. These can include risk mitigation (taking actions to reduce the likelihood or impact), risk transfer (such as insurance), risk acceptance (acknowledging the risk without taking specific actions), or risk avoidance (changing the project approach to eliminate the risk).
  6. Monitor and review: Continuously monitor and review the identified risks throughout the project lifecycle. Update the risk analysis periodically and adapt your response strategies as necessary.

Remember, risk analysis is an ongoing process. Regularly reassess and adjust your approach as your project progresses and new risks emerge. It can be helpful to involve stakeholders and subject matter experts to gain diverse perspectives and insights during the risk analysis process.

A first point to note is that ChatGPT’s response to my short question is much longer than any of its responses provided to Minelle and Stolfi. This is despite their having provided it with more information. Chat GPT describes its response as being a general outline, implying that more detailed advice can be sought about each of its six steps. I confirmed this by asking it some follow up questions. However, if we step back from thinking about such detail, I would suggest that Chat GPT has made a serious error by failing to prompt any enquiry about the nature of “my project” in the first place.

More…

To read entire Letter to the Editor, click here

How to cite this work: Hopkinson, M. (2023). On the Subject of AI, ChatGPT and Risk, Letter to the Editor, PM World Journal, Vol. XII, Issue VI, June. Available online at https://pmworldlibrary.net/wp-content/uploads/2023/05/pmwj130-Jun2023-Hopkinson-AI-ChatGPT-and-risk-Letter-to-Editor.pdf