What Is The Difference Between Robots and Humans, Really?

‘Give Me A Hand.’ What does that statement mean to you? Usually, it would mean helping somebody get up, right? Now, what if a robot heard this statement - What would the robot make of it? It would be the literal meaning of unscrewing the hand and giving it to the requestor.

And now, can you imagine how difficult it would be to ‘program’ the context of this simple statement for a robot? Let’s try to explain it anyway - it would be a list of IF situations.
As in - IF a human is sitting down, lying or incapable of performing normal functions in a normal fashion AND IF he asks you to give him a hand, THEN help him perform the function he wants to accomplish.
Another could be - IF a human is fixing something AND IF he is unable to do something due to the position he is in THEN provide help necessary for him to complete this task.

Are you beginning to see the difficulty in explaining what ‘Give me a hand’ means to a machine? The key word here is ‘Meaning’ - Machines are logical; they will follow instructions, but do not understand the why, context, or meaning behind them. They are excellent at being told what to do, especially when a task is routine. And this is what computers do - as a programmer, I wrote code so that it could add two numbers or display a web page, but to the computer, it is just 1s and 0s (electric signals). But coding a computer to understand love, art, and empathy is impossible. Forget emotions - one of the most considerable difficulties roboticists have is programming a robot to break an egg, which any child can learn easily.

Computers are dumb machines - they have to be told what to do. Garbage in, Garbage out - sums it up. With Generative AI or LLMs (Large Language Models) or ChatGPT, there is a big difference - they can learn on their own. Garbage in and Garbage out still stand true, except they can learn from the Garbage and spew more independently. And we do not know how they learn. For example, you train a model to figure out how many jobs failed in a program you wrote. However, by mistake, you coded the word ‘fail’ in the program for specific successful outcomes, and the model will count the success as a failure. It has learned and learned really well, but it cannot reason. It is logical but not reasonable, as Isaac Asimov states in his Robot series.

As humans, we have not built a muscle that can discern when a machine is spewing nonsense and making sense, especially when the words it uses trigger an emotional response. Humans are emotional creatures, first and foremost. In the above example - you catch the error and instruct the model to correct the mistake, and the model says - ‘I am sorry.’ Whether you want it or not, you will feel an emotional tug when you see or hear the word sorry. But, the machine has no such qualms - it could call you swear words or say sorry - for it, there is no meaning. The recent version of ChatGPT was rolled back because it was overly sycophant-y.

So, what does this mean? Personally, I am coming to terms with the fact that I do not know whether what I am stating is true or not. And also that the world will continue its progress, however it wants. But, I do believe that we are running before we are crawling - we need to do controlled experiments with AI and constrain the data on which it is being trained. Then again, who knows, maybe Robots or AI will also evolve to the level of consciousness. I will end by saying that what makes us humans is our ability to question - our curiosity, which unfortunately is not a skill many of us possess anymore, having given into a life of automaton to get relief from ‘living’.

Are you living by rote, or truly living?