The data round-up #3 - June/July 2023
The EU AI Act - a major step
On the 14th of June, 2023, the EU Parliament has approved its “negotiating position” over the EU AI Act, the legislative proposal setting regulations on the use and development of AI systems within the union. This means that the next steps towards an actual law will entail fleshing out and negotiating the details with all member countries. The EU claims they expect to close the whole process by the end of 2023.
The act will establish that AI systems either developed or used within the EU will respect the union’s values in terms of human rights, freedom and safety. This will mean that the use of software that, for instance, detects and identifies people in real-time in public spaces or classifies people for potential criminal attitude, will be banned. Also, there will be regulations in place for the data that feeds into the training of automated systems, to ensure ethical practices have been followed.
It’s the first big step, at a global level, towards regulating AI. A good article by the Guardian here.
A free book on Statistics
Statistics is the core of data science, and, I think, still the most mistreated of its components. You will hear a lot from me on this topic but for now, let me recommend you a book, it’s a jewel.
The book is “An Introduction to Statistical Learning“ by G James, D Witten, T Hastie, R Tibshirani, J Taylor and it’s not really new - it first came out in 2013 and there’s been a revision a couple of years ago, but for the first time now it has been published with material in Python (the original was in R). You can buy it in print from Springer or you can download the PDF for free from the link above - the authors have made it free and I think that’s just wonderful. There is also a free companion e-course, all at the link.
A subset of the same authors (T Hastie, R Tibshirani, J Friedman) has also published “The Elements of Statistical Learning”, still free and still available in print by Springer. It is a much more maths-heavy book aimed at people with quite advanced mathematical training. I would say the “introduction” is more accessible and high-level, and it also has practical code.
A challenge/grant in tech for social good
I saw this via the excellent newsletter by N Martin:
This is a grant for the development of generative AI systems aimed at social good, participants can be charities, non-profits and academic institutions. The grant is organised by Data.org in partnership with Microsoft and I think it is a wonderful thing. The application deadline is in August, please share the news!
A course on algorithms
Even if AI dominates the news now and the word “algorithm” has become synonym with it, there is much more behind the curtain. An algorithm is a general term.
I’ve been (slowly but surely) following a course on the MIT OpenCourseWare (videos and material free to peruse) called “Introduction to Algorithms” - I like to keep studying, learning and I often use MOOCs. This one is really recommended if you are a coder and want to brush up on/learn some of the foundations of computer science.