I'm somewhat intrigued by this study https://www.eurekalert.org/pub_releases/2019-11/bcom-dnn110419.php
One innovation is to use a "mesoscope" to measure the activity of a large number of neurons simultaneously. This is the kind of development I've been waiting for.
First, measure the activity of neuron populations in response to thousands of images.
Then develop an artificial neural net to mimic that neural activity.
Then, give the artificial neural net new images, and see how its simulated neurons respond.
Compare the simulated output to how brains respond to those same images.
"To test whether the network had indeed learned to predict neural responses to visual images like a living mouse brain would do, we showed the network images it had not seen during learning and saw that it predicted the biological neuronal responses with high accuracy,"
If the method continues to work with other brain data, this could accelerate both AI and brain modeling.
(Whether this leads to good or evil is another question)