AI's project payoff depends on someone who can starve a project, an Info-Tech analyst argues

Prioritization is where the Info-Tech Research Group analyst puts the largest effect in a modern portfolio ^[15]. The job described is concrete: decide which project is next in line, and consciously shelve or pause existing projects when something more important has to start ^[17]. Ranking is an information task. Deciding that a funded project stops is an authority task, and a person has to hold it.

The documentation comes first. The article names five kinds of work that are not naturally interchangeable, a regulatory requirement, an infrastructure replacement, an automation opportunity, an ERP transformation and a new digital product, and says organizations behave as though they are interchangeable every time they prioritize them ^[8]. "Ambiguity is the enemy of project fungibility," the analyst wrote ^[7]. Five kinds of work described on seven required dimensions is 35 populated fields before a ranking exists ^[19]. On the tooling side, the claim is that AI can challenge assumptions, identify missing information, expose ambiguity and iterate with the requestor before an idea reaches an executive ^[10].

The article also describes what organizations actually do: start the exciting new project, continue the old one ^[18]. Its answer is software that compares projects continuously without getting tired of scanning the entire universe of ideas ^[14]. Software can handle the fatigue. Willingness is still someone's to supply. "A project can have an excellent business case and still be the wrong investment," the analyst wrote ^[12], because its value competes with every other use of the same scarce money, people, time and organizational attention ^[13].

"The inhibitor to AI project success isn't project management or project work. It's figuring out which projects to do," the analyst wrote ^[4], and the prescription is to use AI to energize the right projects and starve the wrong ones, since the principles of portfolio management have not changed and only the ability to practise them has ^[5]. It is incomplete on two counts. The column carries no figures, no dataset and no named case for the selection claim ^[20]. And the products are in the future tense: the new generation of AI-fuelled portfolio solutions "will be capable of" assessing risk, resolving ambiguity, estimating costs and scanning the market for lessons learned ^[11].

The choice this quarter is where the AI line item sits, in faster deliverables or in intake and comparison. Next quarter's consequence follows from the sequencing. If intake diligence improves and pause authority does not, the output is a better documented queue of too many projects, and IT leaders in this account are already saying yes to too many projects and creating too much demand for too few people ^[6].