Which do you think would be more helpful to humanity, AI or Automation, and why? how can we differentiate between them?
It's not an either/or question, but a question of utility (usefulness).
Many automated processes do not involve decision-making, and would not qualify as AI, but automation in general has been a boon--allowing for massive increases in productivity, although with the negative side effect that automation often replaces human jobs. (See the Luddites)
- The distinction is that Artificial Intelligence allows automation of decision-making processes
Note also that applications are not restricted to any specific medium (physical vs. virtual), so AI is an element in many areas of robotics.
Both automation and AI have been great boons to humanity, but, as the Luddites suggest, this does not come without potential downside. (Technology, in general, is a double-edged sword.)
Humanity ideally possesses the quality of being humane; benevolence, with compassion, kind loving and many humanly figures, such as kindness, kind-heartedness, consideration, understanding, sympathy, tolerance, goodness, mercy, pity, tenderness, benevolence, charity, generosity, magnanimity.
Artificial Intelligence work could seek to simulate this idea of humanity, although it would be faking it. I think automation is just a process, to automate things. Strong AI could involve artificial feelings but automation, that I refers as process doesn't.
Hope this answer the question.
These are two completely different disciplines and they both can be (are) very beneficial to humanity. And they are both very broad terms. You can speak about automation in a sense of dynamic systems control (like auto-pilots) or in a sense of simple conditional control: if A do B. Same with AI, it can be AI in a field of scheduling, planning, deep learning, state space exploration etc.
Nevertheless, I think that they both complement each other. I think that the goal of AI is to make a decision while the goal of automation is to perform actions. Let me make an example.
Let's say you want a drone to follow a person without the interference of a human actor. What you will need? Firstly you somehow need to tell a drone what is a person and where it is, then you need to make a drone to follow that person. So you will need a person detection algorithm, for example, convolutional neural networks. The output of your CNN will be a bounding box which contains x and y coordinates relative to the image (not relative to the global 3D coordinates). So this is apparently an AI part of your task. Here AI made a decision that it sees a person that it should follow and determined its coordinates. Then you need a drone to follow that person, the simplest solution is to ensure that x,y coordinates of the human will always stay in the center of the image, so basically you want to make, let's say, a PID controller that will ensure that drone performs correct actions. The beautiful thing about pid controller is that it completely relies on the physical description of your system (drone) and it is relatively simple to make one. Of course, you can say that we can make some AI that will also tell a drone how to control the power of its engine in order to turn. But the approach of the control theory is way more deterministic, hence we have the ability to predict when it will fail.
So as you can see, if we will be able to create a strong AI we will also need to enable it to interact with the physical world and in my opinion, this will be achieved by means of robotics and control theory (read automation). Will this be beneficial to humanity? Well, that's an opened question :D.