I’ve read a number of books on similar topics this year around artificial intelligence, machine learning, algorithms, etc. Coming to this topic with little in the way of prior knowledge, I feel like I’ve learned a great deal.

Our increasing reliance on decisions made my machines instead of humans is having significant – and sometimes truly frightening – consequences. Despite the supposed objectivity of algorithmic decision making, there is plenty of evidence of human biases encoded into these algorithms and the proprietary nature of some of these systems means that many are left powerless in their search for explanations about the decisions made by these algorithms.

Each of these books tackles the subject from a different perspective and I recommend them all:

It feels like “AI in testing” is becoming a thing, with my feeds populated with articles, blog posts and ads about the increasingly large role AI is playing or will play in software testing. It strikes me that we would be wise to learn from the mistakes discussed in these books in terms of trying to fully replace human decision making in testing with those made by machines. The biases encoded into these algorithms should also be acknowledged – it seems likely that confirmatory biases will be present in terms of testing and we neglect the power of human ingenuity and exploration at our peril when it comes to delivering software that both solves problems for and makes sense to (dare I say “delights”) our customers.

6 thoughts on “On AI

  1. Michael Bolton

    Thank you for this. I one-clicked Automating Inequality based on your recommendation.

    I’d like to offer two more suggestions for your list: Meredith Broussard’s Artificial Unintelligence, and Harry Collins’ Artifictional Intelligence. Both are first-rate.

    1. therockertester Post author

      Thanks for the suggestions, Michael. Meredith’s book is awaiting my pickup from the Melbourne Library and I’ve suggested they purchase a copy of the Harry Collins one.

      1. therockertester Post author

        Thanks to the Melbourne Library Service for agreeing to acquire a copy of Collins’ “Artifictional Intelligence”. This is a really interesting book that requires a patient read to fully absorb the great content. It’s well written and researched and adds a different perspective on where we are – and where we are likely to ever get to – when it comes to AI. The key point is that machines would need to be embedded in society to become truly intelligent and Harry borrows heavily from his sociology background as well as interactional expertise experiments with the gravitational wave community to make his points. Recommended reading.

  2. robertday154

    Here’s a slightly off-the-wall suggestion, because it’s for a book that isn’t about IT. But it is about human biases and the messes they can get us into. It’s a critical account of British political decisions from the 1970s to about 2010. The book is called ‘The Blunders of our Governments’ by Anthony King and Ivor Crewe; it is available from all the usual sources, but here’s a link to a free PDF – http://online.anyflip.com/lcdq/koyk/mobile/index.html

    It talks a lot about confirmation bias, groupthink and the ways those things can influence real-world decisions. For understanding the way biases creep into high-level decisions that then go into drawing up specifications and definitions – say, for AI systems – it makes pretty sobering reading.

  3. Pingback: 2019 in review | Rockin' and Testing All Over The World – therockertester

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s