Yes, as we become removed from the reasoning, let alone the information behind a recommendation, also removed is the opportunity for critical review of recommen

Yes, as we become removed from the reasoning, let alone the information behind a recommendation, also removed is the opportunity for critical review of recommendations. Your horoscope comparison is apt. Just look at the concerns about self-delusion and data shopping being raised in peer-reviewed science: (text): http://www.vox.com/2015/8/27/9216383/irreproducibility-research http://www.vox.com/2015/8/27/9216383/irreproducibility-resea... (audio): http://www.npr.org/sections/money/2016/01/15/463237871/episode-677-the-experiment-experiment http://www.npr.org/sections/money/2016/01/15/463237871/episo... Clearly, there will need to be better automation, wide understanding, and improvements of techniques for producing more reliable results with machine learning. For example, the common standard of 2 sigmas for publishable resuls ought to be increased to 3 sigmas at least. I'm not suggesting the 6 sigmas used at the LHC but 2 (1:20) seems insufficient.