1. 4


I propose to add a tag machinelearning or data-science or something of the like.

I know there is ai, however, for most submissions in the field I don’t consider it to be a good fit, as the term is very vague and for that reason is avoided like the plague by many machine learning engineers and data scientists.


  2. 7

    I think ai is fine. One of the respected conference is named AAAI after all. On Lobsters, ai tag’s current description is “Artificial Intelligence, Machine Learning”.

    If we had to rename, I support machinelearning.

    1. 6

      I like ai rather than machinelearning, as it’s usually considered the broader term, and I’m not sure we need different tags for all the different AI subfields. There are indeed some machine learning researchers who adamantly view their work as not situated within AI, but I believe it’s a minority; most see themselves as working broadly within AI and publish in both general AI and ML-specific conferences and journals.

      Data science is more of a separate field though, and I could see a datascience tag. Or alternately a stats or statistics tag.

    2. 3

      I agree. AI shouldn’t even belong to an environment full of tech-savvy people.

      1. 5

        It’s the pretty normal term used by researchers. I mean it’s not ideal, but we need some umbrella name. It’s especially useful to have an umbrella term for people who work on applications, where machine learning is usually just one of several tools in the toolbox. If you want to make a robot behave in certain ways, for example, you can use tools from many different subareas of AI: automated planning, reinforcement learning, “deep” reinforcement learning, classical estimation techniques like SLAM and Kalman filtering, subsumption architectures, reactive planners, etc., etc. Some of these techniques are “machine learning”, some are “deep learning” and some aren’t either, but they mostly fall within AI. You even see that on toy problems: AlphaGo used two AI techniques (deep RL and MCTS), only one of which is a machine-learning technique.

        That said, I do kind of prefer the term my own degree has on it, “intelligent systems”. That name is supposed to deemphasize the philosophical and sci-fi angle and emphasize the systems-engineering aspect. Other places have experimented with other names. Oxford recently floated “autonomous intelligent machines and systems” (a name that sounds like it was written by a committee) or “autonomous intelligent systems” for short. We’ll see if any of these catch on…