AI for Project Management
We recently saw a clip where the Japanese Prime Minister was asked a question that had been generated by ChatGPT, and was later asked to compare his answer to the answer also generated by ChatGPT. He concluded that he was able to add more specifics (in his case, people) so he concluded that his answer was better. This might be considered a little frivolous, but it does show the ubiquity of AI conversations currently.
Whilst we at Gowan Projects have a general interest in the rapid development of AI (and who can consume all the content currently being generated on the subject?), our specific focus remains on the development and application of AI for Project Management.
We have recently used both ChatGPT and Bard to generate first drafts of a Project Execution Plan and of a Project Programme (Schedule). The results were useful, if generic, and were most likely limited by the lack of detail and specificity of the initial questions we asked. There is a term for the writing of (better) questions - “prompt engineering” and we would urge people to read this to help improve their experiences with AI and LLMs. Having said all this, the sort of AI-generated results we were able to produce would help project teams to prompt & fill gaps, to provide standardisation of approach & terminology, and to cross-check their own documents and schedules.
It must then be noted that these aforementioned generative AI models are only a part of the AI (hate to use the term) ecosystem. Indeed, this Forbes article cites the generative properties of AI for Project Management, but then the article appears to tail off just as it gets to what would be the meat of the subject for many project teams:
Surely the limitation and the opportunity is an organisation’s ability to open up its own proprietary data to AI and then to train the deployed AI on that data. This would be necessary to create catalogues, summaries, gap-analyses, and useful data for forecasting and managing future projects, learning lessons from the past.
We have heard and read of many areas where AI is helping industries, but less about how to mobilise AI within an organisation and comply with data security concerns, for example. Many client organisations I have worked with are very risk-averse and would be sceptical about opening up company data to any form of AI if the company data is not adequately ring-fenced and protected.
The development of AI is also beyond the financial and technical capabilities of most organisations, so development will be left to others. You wouldn’t write your own BIM software, but you might have your own standards for employer’s information exchange requirements. Will AI for PM be able to be deployed in the same way?
Would you want to deploy a novice AI assistant learning on the job, or would you want a trained assistant brought inside your organisation and then allowed to query, summarise, catalogue, learn, predict and assist based on your organisation's private data and your own prompt engineering? This perhaps is a commercial opportunity in AI for Project Management, but will it be a sufficiently lucrative opportunity for the AI industry? If not, the development of AI for Project Management is likely to lag behind other areas. Collaboration is possibly the best development option, with the tech companies providing the basic platform / tools and the customisation / advice being provided by the "industry". Who the industry is and how this happens is far from clear to us at this time. Perhaps the APM could have a AI for PM SIG formed and contributing quickly?
So how do we deploy AI within our organisations and projects? Perhaps for those of us that use SharePoint or Microsoft 365, the recently released Microsoft Dynamics 365 Copilot could be developed to carry out those laborious and time-consuming tasks on our stored data? To do so, Microsoft (and other providers) would need to know what project teams and project people want from AI for Project Management. Microsoft is asking for feedback, so give it in the Dynamics 365 Community Forums and help shape the future of AI in Project Management.
Finally, and echoing a balanced view from commentators that AI will not “take our jobs”, rather PM expertise will be freed up from the mundane to focus on more valuable PM activities, we end with a quote from the short story The Lifecycle of Software Objects by Ted Chiang: “if you want to create the common sense that comes from twenty years of being in the world, you need to devote twenty years to the task … experience is algorithmically incompressible.”
