4 places AI projects fail

You've heard that most AI projects fail. I've run a few that failed myself.

Every one I've seen broke in one of four places, and none of them had anything to do with the AI.

1. The owner or project manager launches it, then gets pulled back into the day-to-day, and nobody else knows it matters.
2. The team gets their logins, tries it once, and goes back to what they know.
3. The old way is still sitting there, and it's easier, so everyone keeps using it.
4. Nobody's named to keep it going, so it dies the first busy week.

You can check for all four before you start. Who keeps it moving when you're flat out? Who trains the team, one to one? When does the old way get switched off? Whose name is on it?

If you can't answer one of those today, that's where your project will break.

So the next time somebody tells you most AI projects fail, ask them which of the four they mean.