I imagine there are different philosophies out there on the use of an internal wiki or collaboration space like Confluence. On my team it works best to keep it as a white board that we don’t erase. Miro is the same. We might spin up a page to get some shared understanding, then a decision might be made on a call and we never return to the page to reflect the decision. The pages’ purpose was to convey an idea. We tend to leave these pages lying around because they can be a useful reference.

In this model, Confluence is a collaboration space and not a space for final documentation. For a human, it’s easy to look at a page and determine that it’s probably out of date. You can look at when it was last updated and by whom, or you might know that the information is outdated by reading it. This was never a problem before AI.

Now, the people who are diving into AI without thinking suggest that we should go and clean up all of our data so that AI has a clean data source from which to learn and suggest. If we have old or false Confluence pages lying around, then AI will point to those and get things wrong. A collaboration space that is used to make humans more efficient now turns into a hellish documentation space that is meant to make AI more correct.

If these models were really so smart, then they should be able to figure out when something is old or outdated. They should be able to reason that some information is contradictory. They can do those things, but you have to tell them to, and it takes a lot of tokens for them to scan all the data.

This is yet another way in which AI could end up making our jobs less interesting and alive and turn us into the robots. Please, let’s not. Keep the collaboration spaces human, and make the AI have to figure out our mess.