This is your best film yet, and that says something.
For automatization for clinical decisions, it has been known for decades that simple algorithms are better than humans. This has so far not been put to much practice, but it will eventually. See review article: Grove, W. M., Zald, D. H., Lebow, B. S., Snitz, B. E., & Nelson, C. (2000). Clinical versus mechanical prediction: a meta-analysis. Psychological assessment, 12(1), 19.
There is only one temporary solution for this problem. It is to make humans smarter. I say temporary because these new smarter humans will quickly make robots even smarter and so they can replace even the new smarter humans.
How to make humans more intelligent? The only effective way to do that is to use applied human genetics aka. eugenics. This is because general intelligence (g-factor) is about 80% heritable in adults (and pretty much everything else is also moderately to highly heritable). There are two things we must do: 1) Find the genes for g. This effort is underway and we have found a few SNPs so far.[1-2] It is estimated that there are about 1k-10k genes for g. 2) Find out how to apply this genetic knowledge in practice to make both existing humans and the new ones smarter. The first effective technology for this is embryo selection . Perhaps CRISPR can work for existing humans.
- Rietveld, C.A., Medland, S.E., Derringer, J., Yang, K., Esko, T., et al. (2013). GWAS of 126,559 individuals identifies genetic variants associated with educational attainment. Science 340: 1467-1471.
- Ward, M.E., McMahon, G., St Pourcain, B., Evans, D.M., Rietveld, C.A., et al. (2014) Genetic Variation Associated with Differential Educational Attainment in Adults Has Anticipated Associations with School Performance in Children. PLoS ONE 9(7): e100248. doi:10.1371/journal.pone.0100248