Having done a lot of work with large data sets myself it's interesting the enormous amount of people hours still required in Silicon Valley at firms like Google to get data prepped into a format which the computer can process.
AI programs are not well equipped to say this data must be corrupted or incomplete because it makes no sense.
It's like the Tesla car the owner had in self driving mode when and his Tesla were crushed by a semi-trailer merging into his lane. The color of the semi truck looked the same as the sky to the Tesla cameras, so the conflicting information from the LIDAR of a nearby object was ignored as corrupted data or interference.
AI programs are completely lacking in what we call "common sense" which is actually very sophisticated thinking on our part.
Another example when Apple replaced Google Maps with their own system bought in from other firms. The data had not been cleaned by humans so routes for your car would include driving up and over 60 story buildings rather than driving around them - their mapping program being incapable of not being able to assess the fact that cars could not scale the tower.
But once data is "cleaned" by humans AI applications are remarkably efficient, inexpensive and accurate in helping us.
You may have heard of the new CRISPR genetic engineering "libraries" which are able to accurately/precisely find a gene and replace it with something else or turn the gene on or off. Pretty cool, but bear in mind this CAS mechanism is something we stole from a bacteria which had developed it to be able to rid itself of viruses and bacteriophages which had spliced their way into the bacteria's DNA. scientists working with CAS realized we could use this as a tool.
We've run out of other people's Social Security taxes needed to subsidize our low income tax rates.
