[gerner]

Fascinating video. It's cool to see a collaboration of child development and reinforcement learning in action and to hear a research speak about both in an experimental context.

The comment in a different thread about Diamond Age comes to mind, and here we see some of those elements playing out.

[mncharity]

Nod. I wonder what might be done with richer input, gaze and pose, rather than just facial affect (and very limited interaction context).[1] AR seems likely to make those much more widely available - though it could be worked on now, but for lack of institutional context.

On the other hand... there are other pressing bottlenecks to a Primer. Status quo has people learning the Sun is yellow, from Kindergarten teachers, college textbooks, and even outreach. With only a very very few of them getting an "oops, nope, our bad" correction years later in grad school, discussing common misconceptions in astronomy education. We're collectively not able, on a timescale of decades, to even get Sun color right in the most popular college textbooks. So if a Primer is to teach science and engineering as a richly interwoven coherent tapestry... we'll need to figure out socially how to transformatively improve our creation of those stories.

The student in the video is learning the color lavender. Learning color is common preK-6. And yet, even first-tier physical-sciences graduate students are deeply steeped in misconceptions about color. So a motivational peer robot might help... at least with equity. But if we aspire to teach color successfully, a novel goal, attention seems needed elsewhere. But there aren't great incentives to ask "What might a greenfield tech-enabled preK-6 successful approach to teaching color look like?" Or at least, I've not seen it. If anyone knows of a setting for such questions, I'd greatly appreciate hearing of it.

[1] the paper: https://scholar.google.com/scholar?hl=en&as_sdt=0%2C22&q=Imp...