Will program managers be replaced by AI? Not in any way that matters to the people who rely on them. AI will take over a large share of the routine work: collecting updates, drafting reports, chasing actions and spotting patterns in the data. It will not take over judgement, trust or accountability for decisions. Those are the core of the job, and they stay human.
That does not mean nothing changes. A programme manager who spends most of the week assembling status packs should be worried, because that part of the job is going. One who spends the week making decisions, managing people and holding vendors to account will find AI makes them faster, not redundant.
What the research says about AI in project management
- 80%of today's project management work predicted to be eliminated by 2030 as AI takes on data collection, tracking and reporting
- 91%of project professionals surveyed believe AI will impact the profession
- 21%of respondents said they were already using AI in their projects
Source: Gartner press release, March 2019; PMI, Shaping the Future of Project Management With AI, 2023
The first figure is a Gartner prediction from 2019, often quoted as proof that project managers are finished. Read the Gartner press release carefully and it says something narrower: 80 percent of today's project management work will be eliminated as AI takes on functions such as data collection, tracking and reporting. That is the admin, not the role.
The other two figures come from PMI's 2023 report Shaping the Future of Project Management With AI. Nearly everyone expects AI to change the profession, but only about a fifth were using it at the time. The same report says 82 percent of senior leaders expect AI to have at least some impact on how projects are run in their organisation over the next five years. Change is expected, but most teams are early.
What AI does well for program managers
On the programmes I have run, a surprising share of my week went on work a machine could do. Collecting updates from twelve workstream leads. Turning meeting notes into actions. Rewriting the same status in three formats for three audiences. This is where AI earns its place.
- Drafting. First drafts of status reports, risk descriptions, meeting minutes and change request summaries.
- Summarising. Turning a long RAID log or a steering pack into the five things a sponsor must know.
- Spotting patterns. Noticing that the same risk appears in three workstreams, or that actions from one team are always late.
- Checking consistency. Flagging a green status on a programme whose forecast has gone over budget.
A simple example: after a design workshop, AI can read the notes and propose six new actions and two risks, each with an owner and a due date. Writing those up used to take me most of an hour. Checking the proposals takes ten minutes. The judgement about which ones matter is still mine.
Can a project manager be replaced by AI? What stays human
Programmes do not fail for lack of reports. They fail because of choices people make, or avoid making. That is the part AI cannot carry. Trust is built over months: a sponsor learns that your amber really means amber, and a vendor learns that your deadlines are real. No model earns that for you.
AI does well
- Drafts reports and minutes in seconds
- Summarises long logs and documents
- Spots patterns across many records
- Checks figures against each other
- Works at any hour without tiring
Stays human
- Weighs trade-offs between cost, scope and time
- Builds trust with sponsors, vendors and users
- Reads the politics in the room
- Takes accountability for a decision
- Tells a sponsor something they do not want to hear
Accountability is the line that matters most. When a go-live slips, the sponsor needs a person who owns the recovery plan and will stand behind it. An AI model can suggest a plan. It cannot be held to it.
Why AI should propose and humans should approve
The safest way to use AI on a programme is simple: AI proposes, a person approves. Nothing the AI drafts becomes a record, a report or a decision until someone with the authority has read it and said yes.
Propose and approve, in practice
- AI drafts
It prepares a status update, a new risk or a set of actions from the data it can see.
- AI shows its basis
The proposal points to the records it used, so you can check it.
- A person reviews
The owner reads it, edits anything wrong and adds context the AI could not know.
- A person approves or rejects
Only an approved proposal changes the programme's records.
- The record shows who approved
Accountability stays with a named person, not a model.
This matters for three reasons. AI makes mistakes, and on a programme a wrong figure in a steering pack can drive a wrong decision. Sponsors need to know a human stands behind every number. And many clients have rules about what AI may touch, so the organisation must be able to control it, including switching it off.
How to stay ahead as a program manager
Make yourself hard to replace
- Use AI for drafting and summarising now, so you learn its limits
- Move your time from assembling reports to making decisions
- Get closer to the business case and the money
- Invest in sponsor and vendor relationships
- Learn to check AI output quickly and spot when it is wrong
The programme managers I would hire today are not the ones who avoid AI, and not the ones who trust it blindly. They use it every day, know where it goes wrong, and put their own name only on work they have checked.
The time AI gives back is best spent on the conversations that decide a programme. For one example, see how to report project status to executives.
How ProjectCAPEX uses AI
In ProjectCAPEX, the AI agents propose and you approve. They can draft a status report, pull risks, issues and scope items out of your documents, and triage test defects. Each one returns a proposal you can read, edit, accept or reject, and nothing changes in your registers until a person accepts it. Each person uses their own AI key, so you choose the provider and your data goes where you decide. A company admin can switch AI off for the whole company, for example when a client does not allow AI on its data. See it in the demo or sign up free.