LETTER TO THE EDITOR
By Dr Ken Smith
9 March 2025
Dear David,
Almost a year ago, we were apprised by Dr. Dimitri Antoniadis that several AI platforms could not correctly calculate a sample network of five activities.[1] At the time, I double-checked the computations on my Copilot and confirmed his findings. However, although I agreed with him a successor activity without ‘waiting time‘[2] and starting immediately after its predecessor — like a relay runner passing a baton – is an unrealistic scenario,
it was indeed feasible, and projects are typically scheduled in that manner — as illustrated below:
But I recently learned his inaccurate findings with AI were apparently not isolated. Listening to a webinar presentation[3] by Ms. Kathleen Walch of PMI’s HQs. last month, she reported AI has a shocking 80% failure rate! While highlighting causes, one aspect that really struck home with me was when she emphasized AI data bases are dynamic rather than static. Consequently — echoing Dr. Antoniadis — Ms. Walch advised us not to use AI for situations where tried-and-true tools – such as dedicated scheduling software — already exist. On a more encouraging note, she also outlined some corrective measures for utilizing AI information.
To put her assertion about AI’s data base dynamicism and learning curve to the test, I reran Dr. A’s ‘five-activity’ exercise. Sure enough, this time my Copilot came up with correct answers on all counts; so like a good student, in this instance my Copilot had somehow ‘learned’ its CPM computational lessons. Nevertheless, on opening, Copilot does caution “Copilot may make mistakes;” so forewarned is forearmed. Caveat Emptor. [“let the buyer beware.”]
More…
To read entire Letter to the Editor, click here
How to cite this work: Smith, K. F. (2025). AI Can’t Calculate CPA Indicators Correctly! Letter to the Editor, PM World Journal, Vol. XIV, Issue IV, April. Available online at http://pmworldlibrary.net/wp-content/uploads/2025/03/pmwj151-Apr2025-Smith-AI-and-CPM-Letter-to-Editor.pdf
[1] Antoniadis, D. N. (2024). A critique of AI tool errors in Critical Path Analysis theory – a project control case study; case study, PM World Journal, Vol. XIII, Issue VI, June.
[2] Which I subsequently highlighted last month; Smith, K. F. (2025). Slipped Schedules, Touch-Time, and Black Elephants! Advisory article, PM World Journal, Vol. XIV, Issue II, February.
[3] Kathleen Walch ‘Successful Approaches to Running AI Projects – Avoiding the Top Reasons Why AI Projects Fail.’
