Think like a data scientist
I don’t pretend to be a data scientist, which makes being the co-founder of a data science consultancy interesting - but that’s a story for another day!
What can I say?
Frightened. Appalled. Inspired. And I’m only ~80 pages in!
The opening section covers representation, focussing on the bias in ML systems - making the point that algorithms are only as good as their data.
Citing the Google Photos classification example, amongst others, the author drums home ‘if a certain type of data is underrepresented or absent from the training data but present in the real world, then all bets are off’.
But this isn’t intended to be a book review.
Rather, a call to action.
Everyone should be reading ‘all things data’, not just data scientists. Algorithms are already woven into our worlds. We all need to #ThinkLikeADataScientist. Business (especially) needs to better understand AI’s inherent limitations to benefit from its expansive solutions.
Originally posted on LinkedIn: