Gold, being an element, needs nuclear processes for manufacture instead of chemical processes (like diamond). Before modern particle accelerators, this was flatly impossible, and even today, it is very, very hard and expensive. The cost to produce an amount of gold like this is millions of times more than it's value.
Somewhat interestingly, gold is actually being actively destroyed in particle accelerators, as it has several properties that makes it a desirable target. As my professor once put it, we finally have the secrets the alchemists only dreamed of, and we use them to destroy gold.
I have a rather vague recollection that gold has been transmuted from other elements in experimental (fission, I presume) reactors as well, although the best reference I can find right now is [1]. Accelerators do seem to be the modern/"practical" approach though.
Which properties in particular make it a good target?
Edit: more details from wikipedia[2] suggest starting from either mercury or platinum, although it's all a bit fiddly and isotope-specific. That page has some other interesting facts as well - I'd never really thought of element synthesis as economically practical (and with capital equipment costs, probably still isn't), but tungsten ($30/kg) -> rhenium ($6k/kg) -> osmium ($12k/kg) sounds like a nice business to be in if you can solve the practical problems.
When the balloons all run out, we might want to start making our own helium as well, which has the bonus of being really quite easy to do.
Gold has a large cross section, which basically means it presents a big target (amusingly such measurents are made in units of 'barns'). It's also stable, so you don't have to worry about natural decay events polluting your data.
Somewhat interestingly, gold is actually being actively destroyed in particle accelerators, as it has several properties that makes it a desirable target. As my professor once put it, we finally have the secrets the alchemists only dreamed of, and we use them to destroy gold.