> For some subset of the problems available ML works very well indeed, for others it may be a marginal improvement and in many cases ML is just dragged in to a

> For some subset of the problems available ML works very well indeed, for others it may be a marginal improvement and in many cases ML is just dragged in to a project even though it has no place there How to recognize which problems are well-suited for ML? Are there any rules of thumb for (relative) laymen already?