In my limited world, with the blinders on, it feels like everyone is talking about AI and machine learning right now. “Can’t we just use this latest AI algorithm to solve this?” and “We must have this new AI method, everyone else has it” can be heard quite often.

I find it very exciting and interesting to test new methods so I, personally, do not mind learning about the latest developments. Companies that want to create better products should keep up with the overall development as well. When it comes to implementing and using machine learning and AI in businesses, I think it should be based on fact-based decisions, and not just silly decisions that it must be the latest, hottest, technology that should be used. But it is of course difficult to do just that, when you are bombarded by messages from both the research world and advertising from companies that want to sell their latest products in the field. For whom should one really trust? You do not want to give competitors an unnecessary edge.

I sometimes find it a bit ironic that something as data driven as machine learning has such a low replicability in studies. This is something that the news piece “Eye-catching advances in some AI fields are not real” from Science Magazine points out as well, or this article. I guess Data Science and Statistics research are not immune to publication bias either.