Our recent public dialogues on machine learning have spurred our thinking on the importance of communicating machine learning research. Whilst the full results of our dialogues are set to be published this year, Dr Sabine Hauert, a member of our machine learning working group, recently highlighted the preliminary results and how they reflect the importance of science communication at the Neural Information Systems Processing Conference. Sabine was joined by machine learning professor Zoubin Ghahramani FRS, and Katherine Gorman, Executive Producer of machine learning podcast Talking Machines.
Why communicate machine learning research?
Sabine opened our session by discussing where the public are likely to hear about machine learning from. For instance, a YouTube video about artificial intelligence and robotics with six million views is likely to be more significant in shaping the way the public perceives machine learning than the number of citations an academic paper has.
With this in mind, we’ve been contributing to the public discussion on machine learning through infographics and public debates. Individual researchers also have a role to play here as they are experts in the field and understand best how the technology works.
The benefits for machine learning researchers
Our panellists explored the benefits for researchers in making their research more visible.
For example, publishing a blogpost could inform, and be shared by, the journals a researcher publishes in, their institute, the mainstream media and their funding agency. This could increase the profile of their research and the accuracy of subsequent articles written by journalists about the research. This visibility could also offer further opportunities such as Ted Talks and help researchers to build their network. For younger researchers, it can help shape their ideas, gain better understandings of the field by explaining it to the wider community, and hone their communication skills.
Research translation can also help raise funding, as seen by Jibo the robot, whose developers raised $2 million in crowdfunding.
The practicalities: how to communicate your research
As a first step, writing an informative profile and creating a strategy for communication, such as retweeting relevant content on Twitter, can be an important way to start building networks and profile in the research community. Another technique – described by Katherine Gorman – uses an ‘algorithm toolkit’ to communicate machine learning research. This involves crafting research like a story, using a clear and open structure to communicate complex concepts effectively. Katherine gave us some tips for crafting such a research story:
- Beginning: paint a picture of your world and set up the characters; in this case these are the researchers, their personal motivation for pursuing certain research questions and the current state of the field. These are stories worth telling!
- Middle: explain the research problem to be solved and what questions are being asked.
- End: describe the future world once the problem has been resolved, or possible applications of the research.
We’ll be continuing to engage with the public in this area through a debate series on artificial intelligence, robotics and society, an event on machines and creativity at the Northern Ireland Science Festival, and we’ll be publishing our report on machine learning in Spring 2017.
The Royal Society is currently carrying out a project looking at the potential of machine learning over the next 5-10 years, and barriers to realising that potential. As part of this project, overseen by an expert Working Group, we’ve been holding public events, industry roundtables and expert panels. For further information, take a look at our website.