Raise your hand if you know of machine learning.

I sent out this tweet while attending a fantastic session at the AAAS Annual Meeting last week in Boston. This moment has been resonating with me and I want to signal boost it further.

The speaker asked the room to raise their hand if they thought 90% or more people in UK knew the term machine learning. A small, but non-trivial, number of hands went up. 70%? Even more. 40%? A vast majority of hands went up predicting that the vast majority of UK citizens knew the term. The speaker, if I recall, then jumped to if we thought only around 10% knew the term. My hand, and that of a few others, went up. The actual figure is 9%. That does not surprise me. What surprises me, and troubles me, is that most of the scientists in the room thought the number was much bigger.

That's a communication and understanding problem.

We talk about how scientists need to do a better job at communicating with lay-audiences. We need to learn how to distill our messages and tell our stories in engaging ways. I agree! But we need to know what our audiences know, and more importantly what they don't know. I have a strong feeling from talking to friends and colleagues that many scientists have a vast overestimation of what the general population knows. We're always told to know our audience when giving technical talks and this is even more true with lay-audiences. Science communication rock stars probably have a good sense of what is known, but does your average scientist?

Scientists need to not be out of touch.