How Technology Affects Our Ideas And Interesting New Book ‘Everyday Chaos’ With David Weinberger

1 min pitch for what you are doing now?

I’ve been writing short pieces explaining concepts in AI Tech and trying to show why they matter. I’m also very interested in how AI may be affecting our ideas about morality and ethics, including policy responses to these challenges. (Spoiler: I’m concerned about the calls for transparency as a universal solution to AI’s ethical issues.)

In more concrete terms, most days I go into Google in Cambridge Massachusetts where I’m a part-time writer-in-residence in a machine learning research group called PAIR: People and AI Research. I also am involved in Harvard’s Berkman Klein Center as a senior researcher. I also edit an open access book series for MIT Press. I’ve been giving talks and workshops about the book. And, best of all, we spend time with our grandchildren several times a week.

What inspired you to write the book?

I started on it before I got involved in AI. I’d been taken with the way in which so much of what was most exciting about the Net went against our age-old strategy of relying on anticipating the future and narrowing it down to the possibility we wanted.

The Net is chaotic by its nature – it’s many “small pieces loosely joined”, if I may quote the title of a book I wrote in 2002 – but many of our most successful online techniques work by making it even more chaotic and unpredictable. Then machine learning came along, creating chaotic models of the world that can surpass our understanding. We are undergoing a deep inflection point in our understanding of understanding. (My academic background is in philosophy, so this last point is especially compelling for me.)

What does the future look like in terms of better dealing with chaos and complexity?

Our success with both technologies, the Net and AI, are enabling us to acknowledge the chaos and complexity that’s always characterized our lives and world. We will continue in our multi-decade, global, collaborative efforts to make the world more unpredictable at the same time that AI is making more domains more predictable … although not always understandable. I think this will continue to make us more and more comfortable with acknowledging the chaos in which we live, especially as we create more strategies and tactics for dealing with that chaos.

In addition to making us more successful in our businesses and our lives, I think we are undergoing what I can only call a metaphysical change on a scale brought about by Copernicus and Darwin. We’ve had so much invested, for millennia, in our being the species that can uniquely understand how the universe works. Now we’re learning that our way of thinking about things, while relatively effective, may not actually reflect how the world works. That’s a wrenching change that it will take us a generation to work through.

Any tips on how we can leverage this for good, faster than others will aim to use this for less positive ends?

Tough question. I think at best it’s going to be like fighting computer viruses, with each side always trying to get the upper hand. We obviously need to invest in research to keep ourselves one step ahead as best we can. But there are also risks coming from well-intentioned people who may look to AI for easy solutions without considering AI’s inherent tendency towards repeating and even amplifying human biases. Machine learning after all learns from data, and that data often reflects existing prejudices.

It can be very hard to remove these biases entirely from the data. Further, if we don’t involve entire communities — making sure they are representative of the full diversity of opinion and situation — in deciding and planning the ways in which these systems are trained and deployed, machine learning can have very negative systemic effects even while technically achieving its stated goals. The standard example is a municipal bus system that AI successfully optimizes for speed and capacity, but does so by under-serving the poor parts of town.

Anything you wish you went into more detail on?

I’ve continued to learn and think about the ethics of AI, and AI’s effect on ethics. If I were writing it now instead of a year ago, I’d say more about that.

What are you going to write about next & why?

This book is part of a lifelong interest in how our technology affects our ideas. I’m going to keep on writing about that, mainly about AI. I worry that regulators are going to want AI always to be simple enough for humans to understand, which seems to me to be a mistake perhaps based on the metaphysical fears I mentioned earlier. Sometimes, yes, of course, we will want our AI to be understandable. But transparency is a tool, not an end in itself, and there are other ways of managing and controlling the AI that is helping us achieve important human goals.

How can people find out more about you & your work?

Here are some links:

Book site: http://everydaychaosbook.com

Twitter: https://twitter.com/dweinberger

My blog: https://JohoTheBlog.com

My speaker’s page: https://hyperorg.com/speaker/

My posts about AI as a writer-in-residence at Google: https://accelerate.withgoogle.com/ai-outside-in

Excerpt from the book: https://onezero.medium.com/machine-learning-might-render-the-human-quest-for-knowledge-pointless-5425f8b00a45

If you would like to have your company featured in the Irish Tech News Business Showcase, get in contact with us at Simon@IrishTechNews.ie or on Twitter: @SimonCocking