We are already experiencing scenarios where software has decision power over us (think about the early warning systems in our cars, think about the autonomous driving in cars, think about the landing systems in aircraft - no way a pilot can make the adjustments in adverse conditions, consider the dynamic adjustments made to video that is fed back to us so that our retinas don't get overloaded, think about the auto-translation provided by Google, Siri, and FB).
i think we have to look at the glass half full since we have the power to democratize manufacturing, knowledge, and intellectual property to unleash a whole new era of value creation and more widely distribute affordable services and products (which require less individual wealth). This last point is crucial as the market will shift the resources from the centralized scale to the individuals and so to the redistribution of wealth and (perhaps power...tbd).
As for the biasing of AI - very real threat, which is why it's important for us to vote for the right representatives (in each of our countries) to ensure they allow the democratization of AI (no one country or company should own this).
Yes, a universal income will need to be established to address the uneven gaps that are occurring now and will come in waves.
As for politicians abusing AI - yes, many will abuse it for their own purposes. We're already seeing some of that play out by FB and Cambridge Analytica (which is all we know about today....ahem). The number one issue that needs to be addressed by the Internet community (not states) will be ensuring TRUST. I can foresee a time when AI will supplant and/or augment the decision making by representatives in states so that the people can see (real time ) the impact of a decision BEFORE they vote on it ...don't look now, but we're already seeing an increasing use of AI to augment decision making in legal matters (to improve decision making and hopefully, reduce bias and manipulation)