AI: An intelligence without reasoning! Can we trust it?

This is a guest blogpost by Dr Jamil El Imad, Chief Scientist at NeuroPro.

Alan Turing, a brilliant scientist and mathematician is attributed with being the father of artificial intelligence. He published a scientific paper in the late 1950s, entitled, “Can machines think?¨ This short paper laid the foundations for AI experimentation for nearly seventy years. In essence “The Turing Test” became the standard for testing intelligent systems. The technological advances in storage and processing in recent decades have allowed AI systems to progress from its experimentation phase to a deployment phase. This is really exciting and challenging for the industry as many of the current software systems will require rethinking and redesigning.

AI is probably the most significant technological innovation since the microchip. It’s the new “revolutionary” leap. We will see more “smart” systems and devices collaborating with us in almost every facet of our lives. AI will lay the foundation for the fourth industrial revolution and will help transform our societies by optimising what we do, democratizing knowledge and empowering humans to reach higher potential. AI will also allow us to communicate freely with machines in natural language. It will be the 1st time we have machines adapting to humans as opposed to humans adapting to machines.

With all that said it is important that we do not overstate what AI can do. In the same paper where Turing laid the grounds for artificial intelligence, he also said something really important which is not often quoted. He said that the question of whether “machines can think” is itself too meaningless to deserve discussion. Noam Chomsky, a renowned MIT professor said on the subject that asking if a “machine can think” is like saying “can a submarine swim”; it’s only a choice of language.

There is a lot of hype on how AI is going to render most of us useless; it will become smarter than humans, it will take over our lives and jobs and we’ll become the inferior indigenous species on this planet. I have heard similar claims time and time again over the 40 years I have worked in this industry. And just as before, I believe many of these new claims are highly overstated and some are even ludicrous.

I also believe that the use of the word “intelligence” is overstated in describing what is being done! Perhaps Artificial Intellect would be a more apt description. Intelligence encompasses deductive and inductive thinking. Deep learning and cognitive computing are purely deductive mathematical models. Intelligence is about imagination, the ability to think beyond the truth of today. Intelligence is about human curiosity and knowing the question to ask, more than knowing the answer!  Intelligence is about surviving, coping, adapting and regenerating in any conditions. Intelligence is not about finding patterns in a huge number of examples and modeling them on statistical approximations.

We firstly need to understand how the human brain works before we can build truly intelligent computer systems that can mimic it and improve on it. Scientific discovery is normally ahead of engineering and not vice versa. The human brain is the holy grail of biological intelligence and it is the organ that we still know least about. Only recently, tools have allowed scientists to study it in detail but we still do not understand how it works beyond its basic functions. To date, we do not have a brain theory. We do not know the signalingl anguage of the brain. We do not know where consciousness and memory reside. This is because the human brain is immensely more complex than any other organ in our body. It’s the equivalent, in IT terms, of a server-farm with eighty-three billion processors (neurons). Each ¨processor¨ is different but connected through a network of hundreds of billions of dynamic pathways, working and adapting in concert to make us who we are. It never stops! What’s more, unlike tech server farms, the energy consumption of the human brain is less than a 100-watt light bulb! In other words, it’s insanely efficient and extremely complex and you can charge it with a doughnut or a hamburger! So it will be a long while before we can begin to unlock the mysteries of this organ and thereafter come up with a superior alternative. We are safe for now!

It’s important to also consider that human intelligence is based on reasoning that can be explained. This is not the case with artificial intelligence. It is common for people to question an expert as to why he or she has reached a conclusion. We expect experts to use logic as well as other intangible human and emotional factors to explain their rationale. Today, it is not possible to question the reasoning behind a result produced by an AI neural network.

So how can we fully trust AI when it’s reasoning cannot be explained? The answer is simple. We cannot! I believe AI and humans need to forge a partnership where AI provides the intellectual “heavy lifting” and humans have the intelligent final say. Without this partnership and, if we allow AI to have the final say, with no human oversight, we will be ushering in an era of digital tyranny.

Dr Jamil El Imad is Chief Scientist at NeuroPro, which specialises in developing tools for cutting-edge, real-time neurophysiology applications. He is also the Chief Scientist at Virtually Live, a new age immersive media company specialising in live events. He is a founding partner and CEO of TechNovus Ideas Laboratory, which develops innovative, disruptive and digital solutions that leverage artificial intelligence, virtual reality, augmented reality, the internet of things and other emerging technology trends.

 In addition to that Dr El-Imad is CEO of The Brain Forum, a charity that convenes world leaders in science, technology, healthcare and business to advance our understanding of how the brain works. Parallel to his commercial and philanthropic activities, he is an Honorary Senior Research Fellow in the Department of Electrical and Electronic Engineering, Imperial College.  His key research interests are in brain signal analysis, Virtual Reality (VR), Brain Computer Interfaces (BCI) and Big Data.

He is also a Fellow at the Institution of Engineering and Technology.