A few suggestions:
1) IVF will be better when it more closely recreates how the body functions normally. We've done a pretty good job of decreasing the number of embryos replaced per transfer and dropping the multiple pregnancy rate. Now we need to focus on minimally stimulated and unstimulated IVF. Let's be ambitious: one embryo, one baby. Supra-physiologic stimulation will seem really primitive one day.
2) Advances in genetics and genome sequencing will make genetic analysis of some sort a part of every IVF cycle. We will find genes that correlate with implantation and normal development and we will find genes that correlate with early or late implantation failure. Screen, select, and improve outcomes. There has been steady progress on this front but the more impressive results are still to come.
3) The advances in genetic screening will go far beyond getting people pregnant, however, as we better define genetic diseases and learn how to screen for a greater number of serious conditions. Yes I know this is complex, both medically and ethically, but the future convergence of well-defined genomes and very effective embryo testing makes this application inevitable -- at some point.
4) Most importantly -- we need to find a way to make IVF cheaper. IVF will eventually follow the cost curve of most technology-intensive endeavors; that is, over time they become cheaper. A half-decent computer in 1988 cost at least $2000 and a copy of Lotus 1-2-3 or WordPerfect cost $600. Today a Dell Inspiron One 20 starts at $449 and the entire Microsoft Office suite is $219. What will it take to bring down the costs of IVF? The biggest step will be automating the IVF lab. We use the same embryologist opens the incubator door, takes out the petri dish, does something to the embryo or the media, opens the incubator door, put the dish back, close the incubator door series of steps for every IVF case that we have used since the 1980's. The technology -- robotics, microfluidics, incubators with cameras etc -- already exists. There are probably smart biomedical engineering students at work on this now.
More to come.