“There are no AI-shaped holes lying around”
— Matt Clifford (@matthewclifford) September 12, 2024
-> this is how I reconcile the facts that (a) AI is already powerful and (b) it’s having relatively little impact so far
Making AI work today requires ripping up workflows and rebuilding *for* AI. This is hard and painful to do…
* * * * *
The thing about chess is that the structure of the search space is well-understood. The space may be unimaginably huge, but it's well-structured, and we know a great deal about searching such structures.
What's the search space for practical fusion power? Or the cure for pancreatic cancer? For the matter, what's the search space for the meaning of Shakespeare's Hamlet?
Search is only one aspect of a task, there are others – see my posts on chess and language. In general, what are the computational requirements of a task? If we want to find AI-shaped holes in organizations, we need to look at the tasks being performed and how those tasks are assigned to roles. In order to find AI-shaped holes we may have to refactor the way the computational requirements of tasks are assigned to roles so as to create AI-shaped roles.
