AI or artificial human intelligence (AHI) is was originally conceived to develop methods to help machines exhibit human-like thinking processes. However, we have far from achieved this objective.
Instead a discipline of real machine intelligence (RMI) matured from these researches. Why deal with something that is artificial when machines have some form of real smarts to make some form of decisions, bounded by the policies set by humans. They can intelligently do things, like agents, robots, based on heuristic or statistically determined decisions boundaries. Machines are also capable of deciphering deep patterns that are difficult for humans to visualize.
The worst thing is to let the machine act smart, especially in the field of UI (user-interactive) environments. Let humans decide how they want to communicate with other humans. Outsource the menial tasks to the machine.
Bottomline machine intelligence is real; artificial intelligence is by far still artificial.
Now machine intelligence has two parts. The hardware side studies the interaction with the environment, and low level intelligence is used to manage the sensitivity of environmental signals. The software side does the high level intelligent thinking, which is then (in my view) computational intelligence.
Coming back to AHI, to do it, you need to at least provide emotional states to the machines. The machine needs to feel happy, sad etc, to evaluate its decisions and utility functions. Human intelligent behaviour is founded not only on rational thinking, but also on irrational considerations. That said, machine intelligence forms the analytical part of human intelligence, and a machine with only RMI in this sense falls short of being called an artificial human.
Also on the linguistic traits of the human. These are only small parts.
Anyway, this only means that the AHI should be a parameterless model. RMI cannot be parameter free. Afterall it doesnt decide the objective functions, because it does not do the risk-taking.
The objective function is determined by the risk-taker, or in this case, the human. And machine intelligence is only very well a decision-support tool.