Opinion: Deep learning: A misrepresented phenomena

Posted on Sep 11 2017 - 7:59am by Woody Dobson

Machines are evolving – quickly. Remember the “old days” when you could directly tell a computer what to do, right down to lines of code?

Well, that time is over. Humanity is getting one step closer each day to perfecting artificial intelligence. However, this powerful future transition could present a possible cost. Recent evidence on “deep learning” shows it could be beneficial worldwide and ultimately could be used productively.

Interpreting “deep learning” and its global use is imperative to healthy continuation of humankind, as I will explain from my own understanding.

Deep learning is a specific technique used instead of machine learning. Rather than a “trained,” direct program that’s normally instructed for one, specific task or type of tasks, deep learning uses combined “layers” of information that the machine learns how to complete tasks naturally and progressively. This method alleviates the problem of giving a machine the capability of becoming a super-high intelligence right from the very start.

As a matter of fact, this enables machines to learn, much like human infants, only when instructed. This presents a failsafe mechanism, so advanced intelligence won’t adapt instantly, thus, preventing aggressive artificial intelligence (AI) or the infamous “killer robot” state of mind causing animosity around the globe.

However, some view this claim from a different angle. Due to presence of new technologies that look like the science fiction fantasies of previous generations, many global inhabitants won’t willingly agree with deep learning, simply because they feel a sense of insecurity in technology and the implications it has when thinking about religious beliefs on free will and where human intelligence comes from.

Recent advancements of medicine, like synthetic skin and 3-D printing, have publicized previous technological concerns.

This concern begs a serious question: How manageable is deep learning as a force in new technology, and can it be implemented without mass dissatisfaction?

Deep learning is undoubtedly a matter for global security and international stability due to widespread rises of other technologies within its midst. However, it is a very healthy option when handled correctly, and it could preemptively save the human race a headache by reducing time for human instruction.

The recent resurgence of science fiction-esque material and news has’t helped deep learning move forward to more practical stages of AI industrialization. This places conceptual knowledge of deep learning AI in “crisis-like” state of mind around the globe. The differing stances on how to approach deep learning are debatable; however, let’s analyze why deep learning is the best step forward.

Would you rather stay in a pessimistic mindset where past failures have proven machines are helpful when controlled specifically and carefully? Or would you rather pioneer the next generation by adding controlled robot learning to society and ultimately improving humanity?

Becoming a pioneer sounds legendary.

Understanding implications surrounding artificial intelligence is crucial for the next step of machine learning and its societal implementation. Although many obstacles must be overcome, the future regarding deep learning could prove promising from this deeply held and misrepresented phenomena.

Woody Dobson is a senior political science major from Tupelo.