Anyware from next week to a couple thousand years away. What is needed is a software breakthrough, we already have the hardware. Sure a robot that can match humans in terms of memory and processing power would still need to be quite large, but it could be done. The software however is a totally different story.
I’m no master programmer, but when i was in my teens and where a little dumber and more gullible than i am now i tried to create AI, needless to say i failed big time.
What i did, is i basically started with a hello world script, when someone says hi, respond saying hi. I then modified this and gave it some random variation, so when someone says hi or hello or howdie or any other greeting it would access the “greetings()” function and respond at random with either hello or hi or howdie.
As i developed it more and more I gave it more options and skills, and even the ability to remember by using read and write functions that where connected to txt documents. If you told it your name it would remember it, and if you then asked it “what is my name” or “do you remember me” you would get a response.
Towards the end, i tried what i thought was my smartest move yet, and made it learn based on the majority and minority. In other words, if when someone says hi to someone that someone says hi 60% of the time, hello 20% of the time and howdie the other 20% of the time, it would learn to respond to “hi” by saying “hi”.
After a few weeks of working on it, i put it in chat rooms to try and get it to learn, and had friends talk to it. however, no matter how realistic it looked and how well it imitaded other people, it was still only ever doing what i had told it to do. sure, it could remember, and it could make choices, but it would only remember what i told it to remember, and it would only make choices that i told it to make. no matter how much i worked on it and how advanced i made it, it was still just a glorified hello world script.
Eventually, i put some of its functions in to txt documents, and gave it permission to reprogram its self. but still, it would only modify its self according to how i had told it to modify its self. Eventually, i modified it so much that it just became a piece of garbage that said nonsense and gibberish, and was more or less good for nothing.
To create AI, we basically need to create software that will allow software to think for its self, and not thing based on how we have told it to think. we then need to put it in a robot, and give that robot the ability to balance and walk and do several other things.
Until we have a software breakthrough, no matter how advanced or intelligent some software may look, it will still always basically be nothing more than an imitation of the real thing.
So, as soon as we understand how we think for our selves and make choices we will be able to create AI, but until we have a breakthrough in the software department, we wont make anything capable of thinking for its self. much less be self aware and conscious. it could happen tomorrow or it may take centuries.