see url: https://www.bbc.co.uk/programmes/b0bk19xh
see url: https://en.wikipedia.org/wiki/Bayes%27_theorem
see url: https://www.bbc.co.uk/news/world-asia-china-34592186
see url: https://en.m.wikipedia.org/wiki/Social_Credit_System
I have just been listening to a series of 5, 15 minute podcasts from
"Book of the Week". Technically, they are not the best of BBC
productions, but at least they are free, and fascinating too. All about
algorithms and how they affect human beings and their relationship with
decision making processes.
The algorithm is mostly used at the moment to predict the movement of
share prices and markets, and to predict how consumers can be targetted
from the vast amount of information which is gathered about us from our
use of the internet, which interests, foibles, hobbies and little
obsessions which we have which can be turned into marketing, advertising
and selling us stuff, as well as for the state catching terrorists.
She deals with everything from subconscious bias and judgements based on
algorithms, and how the results can affect particular sections of the
community, from cultural biases to scientific, genetic ones affecting
race and gender. She also deals with the great benefits which are
coming from algorithms and how they can help predict diseases like
cancer, particularly when used by pathologists, doctors etc. One
experiment was where the students were trained for a short while on
discerning patterns and the colours of red, blue and purple were used
and the students were asked to discern between cancerous cells, and
benign ones, after the cells had been put through a process of sorting
and sifting and selection. The students did remarkably well on their
own, achieving as much as 66% accuracy, but when the worked together,
and the results calculated, this climbed to an astounding 96.6%. And
who were these students...pigeons...yes, you have read right, pigeons!
With trained human beings it was even better.
She also looks at how algorithms are being used on determining who gets
bail in the US and who doesn't, who gets put on a "no fly" list by
accident, error or subconscious design, and whether driverless cars is
mere hype to sell more technology or whether it can actually happen.
The perfect algorithm cannot exist of course, whether it be targetted at
false positives or false negatives, there will always be some kind of
error, because the quest for truth, or scientific proof, is never an
absolute, but more a relative process, and very difficult define when
one really goes into it in depth, mother nature is so vague at micro and
macro level, it seems. What happens at the nano level is little
understood, and one can never predict human fallibility with complete
accuracy.
She makes reference too, to the Chinese state using an algorithm, called
the Social Credit System Algorithm, based on awarding points between 300
and 900 to each of its citizens, based upon their income, location,
education, class, whether they are a good citizen, etc. The higher the
points, the easier it is to get credit, fly abroad and all sorts of
other perks.
Hello World: She discusses the origin of that phrase too...:-)
Enjoy.
ATB
Dougie.
P.S. If the programme is unavailable in your country, you can purchase
the book on the likes of Amazon or Google, or other good
algorithms...:-). I haven't found a free edition yet, my algorithms
aren't good enough, I suppose...:-)