The first AI problem, the effects of which can be felt today is the disruption of jobs due to automation. The automation penetration levels are only slated to rise, and thus jobs in every field, and of every skill level shall be impacted adversely. The two main methods to tackle this problem include a) an automation tax, which calculates the automation level of each company, and accordingly taxes them, thus incentivizing the hiring of human workers, and b) the establishment of a universal basic income, whereby free money is distributed to everyone, with no strings attached.
The second AI problem, deals with who controls AI. It is pretty obvious that evil AI mustn’t fall into evil hands. This leads to the pressing need of an all-encompassing AI charter- forbidding companies and government agencies from not sharing substantial progress in AI technology, and forbidding countries from using such AI for military purposes, ushering in the demilitarization and democratization of AI.
The third problem is the one with the greatest consequences- the actions of a superintelligent robot. Given how difficult it is to program an end goal into a computer, the difficulty is only amplified when you realize this must occur within a moral, and legal framework. You, then have the additional task of validating those defined end goals using those moral and legal frameworks to prevent any evil use. If you were an AI researcher with such technology at your hands, it would be much more tempting to simply unleash it into the world. To tackle this problem, the scale of ongoing AI safety research must be greatly amplified, and expeditious solutions found.