Producing new content is exciting. Maintaining content, on the other hand, is kind of a drag and can feel overwhelming. So, reviews gets put off. Or we just assume they’re happening in the background…somehow.
AI just compounds the maintenance problem. These new technologies enable us to produce content quicker and more easily than ever, which means we have that much more to keep up with.
Every new piece of content creates a future obligation
Remember that classic scene from I Love Lucy, where Lucy and Ethel are wrapping chocolates on a conveyer belt? The belt starts moving faster and faster, until there’s no way they can handle all those chocolates. Ultimately the whole situation escalates into (hilarious) chaos.
In the world of content creation, AI is like that conveyer belt. You’re creating more and more content to maintain. If you’re not paying attention, you might find yourself unable to keep up, just like Lucy and Ethel.
This would be easy if you had unlimited time and resources
I’ve never been in a position where my team could review everything as often as we should. If anything kept me up at night, that was it.
Reviewing and updating content is constrained by the time available, the size of the team, and the volume of the collection. Because resources are always limited, that means making some hard decisions. And the larger your catalog becomes, the harder those decisions get.
In many cases, there’s no clear signal for what needs review, and when. You may have a mix of urgent updates and scheduled reviews. Limited time plus a growing catalog means you may only be updating materials when you think of it, or when it becomes a crisis.
The risks of letting content maintenance slide
Unmanaged content goes out of date, becomes redundant, or no longer addresses user needs. When users access that content, they wonder if they can trust what they’re reading. And in health, policy, or technical content, outdated info isn’t neutral. It can mislead or confuse people at a critical moment when they’re trying to do something or make a decision.
Outdated content is a reputational risk, signaling that no one is keeping things current. People may begin questioning your ability to be a trusted information source. And believe me, it’s uncomfortable to have to explain to your leadership or the public why something is out of date.
What do we do about it?
You might be thinking: if AI is part of the problem, can’t AI help fix it? Yes, but it has to be part of a larger governance system. You can’t just paste your old content into ChatGPT and hope for the best.
Here are some of the fundamental elements of content governance that AI can support right now:
1. AI can show you what needs review.
How do you know what’s out of date or how often content needs to be reviewed? You may have time-based signals, like the date of last review. There are also risk-based flags, like high-traffic pages or topics that change quickly. You can also define event-based triggers like new research, policy changes, or product updates.
Where AI can help:
- Highlight sections that are likely to be outdated.
- Flag references or claims that may no longer align with current guidance.
- Surface content that conflicts or is redundant with newer materials.
2. AI can help you prioritize instead of trying to cover everything.
If you have a large collection, it isn’t realistic to think that everything can be reviewed equally. But that’s okay, because not all content needs the same review cadence. Here’s the tradeoff: Focus on high-risk, high-visibility, or frequently used content, and accept that some lower-impact content will be reviewed less often.
Where AI can help:
- Identify high-traffic or high-risk content.
- Cluster related content so you can review strategically, not one page at a time.
3. AI can support you in being more intentional about what you create.
Every new piece of content is a future maintenance commitment. So set a higher bar for new content development. That may look like a priority list or a set of criteria for when new content is justified. Or not creating one-off materials that are hard to maintain. Also, consider retiring or consolidating content that no longer adds value. There’s low ROI to update something no one uses.
Where AI can help:
- Create priorities or criteria for new content creation (beyond “wouldn’t it be nice if we had…”).
- Identify overlapping or duplicative topics during planning.
- Suggest candidates for retirement (e.g., low traffic, redundancy, superseded content).
4. AI can reduce the effort required to maintain content.
Realistically, you probably can’t add much capacity in terms of people or time, so the maintenance process has to get lighter. That’s where standardizing review processes and criteria can help. Use checklists to focus reviews on the things that are most likely to change. And streamline your collaboration with SMEs and contributors to spend less effort on coordination.
Where AI can help:
- Pre-review content using a checklist to highlight likely issues.
- Generate consistent review summaries or requests.
- Create templates and SOPs to reduce repetitive work.
AI isn’t a substitute for governance
I wish I’d had AI to create content in the programs I led at NIH. I also wish I’d had it to help with review and updating—that was an ongoing challenge for my teams. We were always looking for ways to make updates easier and faster. But AI isn’t a magic bullet here. It needs to be incorporated thoughtfully into a larger governance plan. The real value comes when it helps you manage the full lifecycle of your content, not just produce more of it.