What would need to happen for Bitcoin to fork to a ASIC resistant mining algorithm like RandomX used by Monero for CPU mining only?
The bitcoin community would have to decide that they don't particularly care about having significant security anymore. ASIC-resistance weakens the security of proof of work chains because it eliminates the need to invest in a chain in order to attack it.
Someone who owns lots of expensive ASICs will not want to let them be used to attack the chain they are mining because if they harm that chain, they will significantly reduce the value of their own ASICs. By contrast, if you can mine with commodity hardware, attacking the chain does not significantly reduce the value of your hardware, so you have no real incentive to prevent your hardware being used to attack the chain. Worse, commodity hardware is easy to rent or steal (with malware) to attack a chain.
So the entire bitcoin community would have to change their opinion on how to make a chain secure to a worse opinion.
It's also important to understand the security is a huge limiting factor on price. If the benefit from causing a double spend exceeds the cost of doing a double spend, there will be double spends. Nobody wants that. So market forces will act to keep the benefit of a double spend below the cost of a double spend.
ASIC-resistant algorithms have much lower cost to do double spends (because you don't need to invest in ASICs but can instead rent or hack commodity hardware) so the profit from a double spend must necessarily be much lower. That will mean either waiting for many more confirmations, dropping the price of bitcoin, or some of each. Who wants that?
Agree with David's answer, but the security detriments of ASIC resistance (of which there are many) is somewhat besides the point.
The entire value proposition of bitcoin, and what makes it so valuable above all other coins, is that it cannot be controlled by humans. If such a large change were to be made, the chain would not be considered "bitcoin" by current miners (who mine with ASICS and thus have everything to lose) nor the community (who recognize the importance of protocol stability).
I essentially disagree with David Schwartz here. The aim of bitcoin was decentralisation, and never high prices. And to say commodity hardware promotes the idea of attacking the chain is simply ridiculous. Why will anyone who has the ability to bring so much malicious hardware together, will ever use it to make all his bitcoin useless? Since compound interest is obviously better than a one time cash out. So do we want decentralisation or higher prices is essentially the question. Do we give power to centralisation of ASICs and forget decentralisation, or actually go back to the roots of the problem. Can hear anyone's suggestion on this one.