NIPS 2016 Symposium on People and machines: Public views on machine learning, and what this means for machine learning researchers

Hover over the left for a table of contents. These hastily-captured notes are not accurate; they are paraphrase and may contain inaccuracies, typos, and falsehoods. Importantly, they may not reflect the true intent of the speaker. I apologize for any inaccuracy.

Symposium organizers: Susannah Odell · Peter Donnelly · Jessica Montgomery · Sabine Hauert · Zoubin Ghahramani · Katherine Gorman

Susannah Odell: Public perception surveys and storytelling for professional communication

(missed the first part of this talk and its title!)
  • 9% of people in the UK have even heard of the term "machine learning"
  • …but 70% have heard of computers that could understand human language and could answer human questions
  • …and 75% have heard of driverless vehicles.
  • 75% get information from mainstream media (tabloids?),
  • 21% get info from entertainment (Westworld? Black Mirror?)

But they do see applications.

Where do people think machine learning could be helpful?
  • Recommendation
  • For doctors
  • Improving education
  • Tackle global challenges
  • Aging population

People have questions and concerns about machine learning
  • Algorithms should be safe and trustworthy
  • People want to know that they won't be replaced by machines
  • They don't want these algorithms to be making decisions that could have a real impact on their life, or label them in a way that prevents them from doing something down the life

How can we address these concerns?
  • Connecting directly with the public (see what the Royal Society is doing)
  • WE are the ones who need to be helping! Expert view, demystifying technology. Talk to people: policy makers, users, future generation, investors, taxpayers, funding agencies.

If you engage with the public in a human-readable format (such as a blog post), it makes your research more visible.
  • Blog posts have high visibility because they become the main point of communication
  • Research is more accurately portrayed
  • Helps with citations
  • Increases opportunities (speaking, …)
  • Growing your network
  • Helps you understand your field, shape your own ideas, practice communication skills
  • Allows you to raise funding
  • Have your opinion heard

There are a tremendous number of reasons to be communicators.
And there are a tremendous number of reasons why we don't do it.
  • We don't know how to use modern communication tools;
  • Communicating effectively is a time sink;
  • Don't have a platform (tools like Twitter help with this)

What can you do to make your own document more human-readable?
  • Interviews
  • Video conferences
  • Increase diversity