Practical and constructive solutions as to what to do about global warming are much more welcome than demands to stop using electricity, cars, jets, plastic, or transported food.
It irritates me that this article from Yahoo! News describes these ideas as crazy. If it can be done, and if it works, and if it doesn't cause enormous horrible side-effects then it is worth considering.
My favourite idea in this world-engineering vein has been the "gigantic space mirror". It seems wonderfully hubris, but more importantly would give us an excuse to get into space.
Once we have a toehold in space we can begin real space development, construction of a space elevator, self-sufficient space habitat, start mining the asteroids, dismantle Mercury to build a Dyson swarm, and build floating cities in Venus' atmosphere - all the usual things.
Because if it's not the nuclear bombs, or the global warming, or the plagues, or the asteroid impacts, or the earthquakes, or the gamma-ray bursts, it will be something else that gets us. Existential threats surround us, and their prevalence may account for the Fermi paradox.
We need to use the vast resource of oil we are fortunate enough to have to bootstrap ourselves to the next level of technology. Molecular nanotechnology offers enormous potential because it involves manipulating matter at the most fundamental level. If developed to its fullest extend, along with genetic-algorithm-based design software and other things we haven't even thought of yet, it would also create a self-sustaining, self-operating, self-designing, and highly durable "technosphere" independent of the need for human intervention or maintenance.
Why would such a thing be desirable? There is a sort of paranoia about handing too much control over to artificial machines. I believe this is rooted mostly in our knowledge that machines are unreliable in all but the most routine of circumstances, and sometimes they break down. This is why jet-aircraft still have pilots, and trams still have drivers.
But if artificial machinery becomes more like naturally-evolved machinery it would become at least as durable as we and our biosphere are (...if not more so, because it would be less restricted in terms of its use of materials, and it could apply sentience to the problem of design, eliminating flaws like the combining of the breathing-hole and the eating-hole in land dwelling vertebrates...) and these objections to handing control to artificial machines would be irrelevant.
The argument that suggests that "the evil computer will take over the world" is an interesting one. I believe that if we are ever to create an AI that will equal or exceed human "intelligence" (however we quantify such a thing) its mind will need a model of the world at least as good as ours, a model of all the most complex things in that world (including humans) as good as ours, and it would need a model of itself at least as good as ours.
Such a thing would undoubtedly be sentient, as it would be able to view and map its internal processes as well as (and probably much better than) we can. In fact, it would be very close to being a human mind.
Such a mind would be in pretty much the same position as the rest of us as regards taking over the world, but it is worth pointing out that if we imagine this future to be as democratic as the present. Given space, and full human rights, and citizenship of a state, there is no reason why an AI/upload/virtual person couldn't create enough independent copies of itself to affect the outcome of elections.
But such an AI would go beyond human. Part of the power of software programs is they can rapidly modify themselves to suit the job they are doing. Imagine if you could increase your level of curiosity over the tedious report you have to write for work, or become more logical for a maths exam, or develop hand-eye coordination for a badminton match.
So I imagine that the first "true AI" will simply be incredibly accurate models of human beings running on a computational substrate. In this case it won't be a matter of "handing over control to the machines" as it will be simply giving control to those in the best position to use it (I'm assuming human beings running on this substrate will experience time at the same speed or greater than human beings - but I suspect that we will be able to develop computers powerful enough).
Once we have a durable technosphere then, for the first time ever, everything that really matters in this world will not be stuck within three pounds of goo, protected by a thin layer of bone from all the nastiness of the universe.
Or so the transhumanists would have us believe...
Anyway. Realistically, we need to conserve our oil resources, keep our industrial infrastructure, gain a substantial hold in space (including Earth orbit, the Lagrangian points, the Moon, the asteroids, Mars, and Jupiter), develop the third world, and conserve the beauty that can be seen in our only example of a functioning biosphere. That means (for the time being) nuclear power, unless someone comes up with a decent fusion-power-generator.
And a giant-space-mirror would also be pretty cool.