So what's your guess as to when Tom?
1 year, 7 months and 2 days
(Just kidding
)
I'm not current on the fields needed to make it so but there is a bit of reasoning to any consideration for a time frame.
If you separate it into mind and body, each has many factors currently beyond mankind's capacity to create.
Looking at the body...
Its like having a Lego man with the movement of a CGI man.
The artificial bodies we create are mechanical and void of sensors and actuators compared to the human body. If human sensitivity and balance is a 10, our current creations are way down below a one (by comparison).
Constructs have sensors and actuators. Humans have a nervous system and a muscular system.
People often misunderstand the mechanisms which cause human locomotion. They have the tendency to over-simplify human function.
They think...
Humans have muscles which move a bony skeleton.
Robots have actuators which move its frame.
Humans have nerves which give the brain feedback information.
Robots have sensors which give the processor feedback information.
While it is true it is a gross simplification. Like rounding off the number 9 to the next 100,000.
When you move your finger it is a complex series of signals which controls its precision. It takes more than finger nerves and finger muscles to move it and even more to move it precisely as you intend. Your brain doesn't just send a signal for your finger to strike a 'q' key on your keyboard.
Your finger is not the only thing moving and its nerves are not the only thing controlling it.
Your hand is working, your wrist is working, your arm is working, your shoulder is working and so on. All these components are finely compensating in harmony to press that key. They are also constantly adjusting for changes.
By comparison, a construct performing the same task must be programmed with a set of instructions to move the arm in position, move the hand into position, move the finger a predetermined distance down then up again. Move the keyboard, the same movement will occur with an unsatisfactory outcome. Its also entirely possible the finger would smash thru the keyboard or not press hard enough for the letter to register.
For a construct body to have the movement and precision of a human being it would need to be covered in multi-functional, multi-capacity, multi-sensitivity sensors. Our brains sense our muscle movement in all directions, We feel the muscles in our fingers moving under our skin and across our bones. This is because the cells in our bodies connect thru a nervous system.
We are not aware of it because it is not an input/output command.
With a construct, it needs to be an input/output command because the construct is not governed by a complex nervous system of sensors and actuators. The sensors and actuators must be built and installed then aligned to the function the perform.
All hope is not lost tho.
Nanotechnology has the potential to open technology that could make an artificial nervous system a reality. The construct would be covered by a complex saturation of atom sized sensors and actuators coupled with nano-brains feeding/taking orders from the CPU based on the information gathered from gross movement.
I doubt that technology will unlock in this century, maybe the next century, if we even care to have android technology.
Looking at the mind...
The three laws of robotics is flawed. Need to get that out of your head right away. Those three laws would lock a construct into a loop then destroy itself from overload.
All actions have the potential to harm a human being, including no action.
The AI controlling the construct would need to have the capacity to understand causality.
It would need the capacity to understand human emotion, motivation and intent.
Hell, human being don't even have that capacity.
The biggest roadblock to creating a human-like AI is for humans to fully understand humans.
This may never happen?
But, for the sake of conversation lets just say we do.
Once again there is a gross over-simplification in the comparison of human movement/intent to construct movement/intent.
Someone says they're thirsty (just speaking out loud). You hear them and decide to fetch them a nice cool glass of water. A simple thing for us but could be kinda complex for a child. But still doable.
You get up, go to the cupboard and get a glass, go to the fridge and remove the pitcher of water, pour the water in the glass, pit the pitcher back in the fridge, pick up the glass and carry it to them and hand it to them.
Simple.
A robot must be programmed to do each of these tasks in order. All components in the sequence must be where the program directs including orientation of each component.
Distance, speed, force, pressure, etc... must all be programmed into this simple task.
It will not hear the proclamation and start the sequence on its own.
An android with a basic AI might be able to recognize the request and start the sequence. It may be able to compensate for variables as it is programmed to recognize. It might know what a fridge is, it might know what a pitcher is, it might recognize the handle, it might calculate the water level and flow into the glass, it might be able to put the pitcher back and carry the glass to the person and hand it to them. As long as there are no unforeseen occurrences in its sequence.
A human-like construct would be able to recognize the request and decide whether it was an actual request or just an intent that person tells themselves before they themselves get up and fetch their own drink. It will decide to go get them water instead of coffee or anything else available. It will decide which glass to use. It might move a glass or two so it can pick the glass which best fits the purpose or select that person's favorite glass. It will be able to determine if the glass is upside down and carefully reorient the glass if it is not. It might do this by inverting its hand before it picks up the glass so when upright its hand will be in normal position.
It will move towards the fridge but compensate to account for the family dog which just walked in its path. It might stop or walk around the dog.
It will know which container holds the drink it wants to present to the person.
It will know the orientation of the handle or spout, how much liquid the container contains, the rate of pour for the best control and when the glass is filled to the desired level.
It will know whether to put the container back into the fridge or throw it away or sit it on the counter to be refilled when the person has their drink.
It will compensate how it walks so as to not spill the liquid.
It will know how the person is oriented and how they reach out to take the glass. It will feel the tugging pressure as they take the glass and slowly let go so it doesn't spill.
When the person says thank you, they know to respond with you're welcome.
They then move away from that person to respect their 'bubble'.
All the actions, responses, compensation, adjustment, intent, etc... must be governed by the AI.
Its still programming but it is real-time and finely regulated. It must also be multi-functional because environmental changes are also happening including sound and light intensity. It must account for and compensate for pressure, balance, temperature as well as determine the importance of all the peripheral activity going on around it.
The weird thing is we do this stuff everyday, sometime to some extent, every moment of every day. It would be easier to teach a chimpanzee to be a house butler than to build one.
Cue Planet of the Apes scenario.
The thing is, if we eventually build an AI able to cause a construct to function normally, that AI would quickly surpass us. So, even success would be failure. Who is gunna clean the house?
Even the Bicentennial Man stopped being a servant.
Basically you have;
A Robot limited to its construction and programming.
An Android limited to its construction and reasoning.
and
An AI Construct which surpasses its purpose and its creator.
Realistically, we will never have
Bipedal Robots as Human Sidekicks
