What is this "random" of which you speak?

OF course RNA would not "assemble itself" as in the individual atoms that would make up RNA just kind of "coming together" by swirling them in a petri dish.  I don't think anyone who is studying this would ever imagine that to happen. 

What they are imagining is that an RNA-like molecule would evolve from smaller molecules under certain conditions over a period of time. 

The example of a tornado assembling an auto is a not a good example as presented.  Imagine NOT that the goal is for an entire auto or jet or whatever assembling itself, but that all steel components aare attracted to magnets.  It would not be surprising that swirling steel objects about in an area with magnets would result in the steel eventually connecting and....and this is the important part...sticking TO the magnets as the swirling continues.  I imagine atoms and molecules as being selectively attractive to other atoms/molecules in a manner that might be simulated or represented by selectively magnetic metals.  It is that regular and predictable way in which they tend to bind with each other in certain ways but not in other ways that leads to more complexity.  Add to the "tornado in a junk yard" analogy other selective processes like various sized and shaped openings in a matrix that would only allow certain kinds of items through, eventually you'd have small round objects in one place, larger square object elsewhere and other oddly shaped objects in other places.  Then swirl them again and depend on the regular ways in which they combine and you'd eventually get standard and repeatable chains of components.  THEN those repeatable chains of components undergo other selective processes and eventually you'd have complex structures that happened to start looking surprisingly regular and would enjoy a preferential benefit due to their shapes and composition.  Of course they'd still not look like autos, because humans needed autos to look the way they did for an intended purpose.  But the objects forming naturally don't have to serve anyone's purpose, they just have to be able to replicate themselves.

And the example of the light bulb was to show that it took 3000 attempts to achieve something as simple as on material that would burn long enough to work as the glowing element in an electric light bulb.  And yes, they've found other ways, but each with a long series of attempts and failures.  In each case, before they solved the problem...the problem had not been solved.  If they had stopped at 100 or 1000 or even 2900 attempts they may have left assuming it was impossible. And again, it is important to emphasize that this was just to get some material to glow long enough to create a dependable light source.  IE a very simple goal. 

With simulating life, it would be a much more complex goal and would require either a lucky breakthrough or a proportionately longer time to accomplish, if it could be accomplished.  And it's not certain that it could because I doubt we'll ever know the conditions that existed where life might have formed in early earth.  But the work at Scripps institute and at Harvard is likely to be the most promising.  (And, I present these sites NOT to convince you Arlon, but for other's looking on.)

Another example would be if you took a flat horizontal surface and dumped billions of randomly shaped marbles on it with a hole on one side.  If you simply swirled the flat surface about (tornado) you might expect some of the spheres to eventually, essentially by chance to find their way into that hole.  And you could assume that only spheres that were small enough to get through would end up on the other side of the partition.  Not too selective, right?  But add to that the thought that the partition had lots of varying sized holes.  Still, swirling the marbles about would separate them out somewhat, in that only small ones would go through the small holes, but alas they would also go through the larger holes.  And even then, only by "chance" would the marbles ever reach the side that had the holes.  But add to that a slightly sloping floor.  Still not much difference except that all the spheres would, on their own and quite naturally roll toward the wall with the openings.  Eventually most of the marbles would be against the perforated side and many many more would move through the holes which would selectively allow smaller objects, but not larger ones.  Still...smaller ones could get through the larger holes.  But ever so slightly tip the surface such that the spheres roll first to one side then roll along the wall reaching the smallest hole first, then larger ones in a sequence.  That would do an excellent job of sorting out the smaller from the larger marbles.  Still no particular intelligence needed, just gravity and different size openings.  And that represents how different conditions would sort out different atoms and molecules naturally.  Add to that a further selective process of adding the concept that each sphere is selectively attracted to certain other spheres.  Still no particular intelligent design, just spheres behaving naturally. This could lead to interesting compositions that would never have formed just by the tornado style swirling...but would still occur with no intelligent design needed.

Consider that the presumption is that not only are we assuming there was hundreds of millions of years, but that molecular combinations and recombinations occurred in untold trillions of various compartments under trillions of conditions at probably millions of times per second and all following standard and regular ways in which atoms and molecules combine.  So the fact that they've been studying this, what?  A few decades, maybe a hundred years in a handful of labs around the world and still haven't hit on a way is meaningless.  What is interesting is that they seem to be making progress.  This leads me to believe it is very much possible they'll hit on some way in which these long chained molecules could form and replicate and eventually produce something akin to RNA. 

But again, for the lurker, even if they manage to create bona fide complex entities that qualify in every way as life, it still wouldn't prove that God didn't create the universe in just the right way that it COULD produce life naturally and it still wouldn't mean they found the way life actually did form, IF it formed naturally.  Also, they might manage to put together such "living" entities, but it still could be that a God of some sort did actually start the life on earth as we know it. 

IOW, successful experiments wouldn't prove anything OTHER THAN it might be possible for life to have formed naturlally...NO STRIKE THAT.  Given that God could still manipulate experiments, even if they created something that qualified in every way as life, the theist could still claim God "did it again."

By the same token, they may study this for centuries...millennia and never find a way to get something that qualifies as life started.  That would not prove it could not have happened. 

But and this is a big but, whatever they discover, even if it is bona fide living entities that look and behave exactly like primitive cells/bacteria and do it from scratch...ie putting the right components in the beaker at the right time to  simulate possible selective environmental conditions, you can bet there'll be someone or some group moving the goal posts to make their story still come out right.