This doesn't actually sound promising at all to me for solar power generation. In the original press release [1] they speculate that they may eventually reach 10% efficiency, which we already have. Given that the effect requires stupefyingly absurd amounts of light and that they're going to have to improve by several orders of magnitude to harness this effect to do real work without causing the medium to explode due to a sudden influx of a huge amount of light, all to obtain an efficiency we already can, I do not see this as likely to be useful for solar power generation.
I criticize the need to try to attach every bit of research to the buzzword de jour. This is legitimately interesting on its own and the odds of it having some further use either scientifically or for some other engineering purpose is quite good. They've established a new boundary condition on some very venerable equations, which can't hardly help but be useful at some point. Tenuous connections to an application that it probably won't be useful for weaken the point, not strengthen it.
The authors argue that, in a dielectric medium, light can induce an electric dipole moment in the direction of the light propagation by shifting the average location of atomic electrons in that direction. This moment becomes a means of storing energy, and they expect that heat loss would be much less than in traditional semiconductor solar cells.
