Think Like a Human Too

The rate at which today’s world is advancing is unbelievable. We see new technologies and machinery all around us. Machines have been programmed to work in the absence of human supervision, doing the same job that we humans do, but at a much faster rate. Does this really mean that Artificial Intelligence (AI) is always the best pick?

Machines integrated with AI, perform certain actions, observe the outcome, and accordingly take the next step. These agents keep learning from their previous decisions. AI always avoids human interferences and creates situations where humans cannot stop it. AI engineers must prevent these AI machines from finding a way around the commands given by humans. This gives us much more confidence that AI will always take the right step and have very less room for error. Yes, this is true to a certain extent. The scope for error is reduced by using the machine learning technique.

Artificial Intelligence users reinforce learning as part of their machine learning technique. It follows a reward based system where machines will earn rewards for every right action they take. We can analyze this with an example if a robot is programmed to carry items from the main gate to the warehouse and then stack them, every time it completes this job it earns a point. If it was interrupted by a human operator when it goes to the main gate, the robot will learn that it’s better to stay in the warehouse and stack the items. Because every time the items are stacked correctly it gets a point. It’s more interested in making as many points as possible. The human interruption was actually done in order to train the robot to avoid these interruptions. Hello human clones, think like a human too.

Artificial Intelligence engineers try to resolve this problem through a method called safe interruptibility. This method allows humans to interrupt AI during its learning process. These human interruptions don’t change the way the machines are supposed to learn. The method just deletes certain data from the machine’s memory hence the learning process is not affected.

The chase between humans and machines will never end. So at this point, it is hard to decide if AI can work without any humans to supervise it.

“I fear the day that technology will surpass our human interaction. The world will have a generation of idiots.” – Albert Einstein

Alisther Barreto