Over the last few decades, we have observed tremendous improvements in the ability of machines to carry out human tasks. Thanks to progressive studies, series of algorithms are consistently improved to ensure that these functions are carried out efficiently.
Regardless of daunting improvements in recent times, scientists continue to improve the functionality of these machines. This brings to the question, will machines ever attain consciousness or be self-aware like the human mind?
Self-consciousness cannot be understood by mere behavioral observance because it’s an act of the mind. If we are to determine the probability of machine self-awareness, a deeper study is required.
A long while ago, a host of philosophers and scientists tried to give appropriate answers to the question of understanding consciousnesses. There are three significant tiers in the scientific strata concerned with proper investigation and evaluation of consciousness. These tiers are:
- Computer science/Artificial Intelligence(AI).
We’ll be focusing on the mutual gaps between these three tiers and understanding the ability of a machine to attain consciousness or be self-aware.
From a Philosophical Standpoint
In philosophy, over two thousand years ago, Aristotle, the father of logic, made it clear succinctly based on a personal conviction that only humans possess a soul rational in thoughts and behaviors. This attribute explicitly differentiates humans from animals that lack core consciousness, which is present in humans.
Some philosophers corroborate with the idea that “only humans can be self-aware,” and this has primarily created the bedrock for any philosophical approach in this regard. Renowned philosophers have also highlighted and proposed various theories that further eliminate the possibility of self-aware machines. Some of these postulations are; The Theories of Eliminativism, Strong Reductionism, Mysterianism, Dualism, and Epiphenomenalism.
All these theories explain in detail the difficulty in defining the consciousness of human beings. Thus, this allows us to infer that philosophically, it is almost impossible to understand or explain the state of self-awareness fully. This implies little or no room for a world where machines are self-aware as humans from a philosophical standpoint. This is a significant setback in the exploration of this topic.
On the other hand, neuroscience offers a host of approaches towards measuring levels of consciousness, allowing us the luxury of probing this topic further.
Electroencephalography (EEG) was the first method used to detect electrical brain activity. EEG was introduced by Han Berger, a neurologist, in the year 1924. This was a great discovery that broke the grounds for exploring different mental states (self-awareness) over the years.
This became a stepping stone for proper investigations cutting across various cognitive behaviors, actions, memory evaluation, biological basis for perception, and a better understanding of the brain’s neural networks. Without a doubt, neuroscience holds vast scientific information about human consciousness as researchers over the years have tackled issues such as: how consciousness correlates with neural knowledge, the computational phenomenon achieved through consciousness, the theory of a global workspace, and the model of consciousness postulated by Damasio. His theory, in a nutshell, states that feelings and emotions are fundamental concepts, not necessarily an innate propensity in humans opposing the beliefs of philosophers. Damasio defines emotions as direct signals indicating the level of positivity or negativity in any individual. At the same time, feelings, on the other hand, occur due to an individual relationship with images of the universe. This theory is interesting because it provides a biologist and machinist point of view.
Biologically he created a plausible model on consciousnesses as he had assigned all stages of consciousness to specific structures in the brain and associated them with respective functions. He also made a machinist model that, at least by principle, can be infused into a computer program. However, with all this wealth of knowledge neuroscience offers, neither electroencephalography (EEG) or functional magnetic resonance imaging (fMRI) can come close enough to exposing what exactly happens in the complex human brain. This leads us to seek our answers from the progress in the recent massive influx of machine and artificial intelligence studies.
Tapping into the bedrock of discoveries made by neuroscience, artificial intelligence hosts many theories on consciousness, obviously from Damasio’s Machinist point of view. However, it’s a hard nut to crack or quite tricky as most of these theories are based on the idea of a global workspace with limited learning capabilities. Most of the consciousness is both difficult to train and hard-coded. Therefore, most of the assertions and studies during the study and algorithm creations are based on symbolic representations than the real human consciousness.
Artificial Inteligence (AI)
From an Artificial intelligence point of view, there are a plethora of theories of consciousness. However, virtually all algorithms investigated to create self-conscious machines have toed the line of a global workspace model of consciousness, which may be likened to a mechanical model.
The challenge with such mechanistic models is that the focus is predominantly set on representation than self-awareness. And suppose such mechanical models may be applied to achieve higher forms of consciousness, such as Damasio’s extended consciousness. In that case, complete models of a human brain with all known connections are necessary.
While representation is an essential component in the actualization of self-conscious machines, self-awareness remains an integral part of this process. And so, there remains the daunting challenge of an incomplete knowledge-base of implementable models of consciousness, which is paramount to develop human-like consciousness.
Machine learning provides a basis for solving this problem, given its ability to form and train complex models. Unfortunately, due to widespread belief among the scientific community that human consciousness will never be simulated on a computer due to the infancy of AI ideas, there’s a lackadaisical attitude towards implementing theories in this space.
An exciting line of thought is Solomonoff’s Universal Theory of Inductive Inference, which gives a relatively comprehensive insight into inductive reasoning by combining results and theories resulting in simpler models.
As a result, Hutter’s Universal Artificial Intelligence theories have applied Solomonoff’s Universal Theory of Inductive Inference as a basis for more research.
Also, based on the works of Solomonoff and Hutter, Maguire coined that consciousness is complex and is a resultant effect of bound information. This implies that a deconstructed consciousness is impossible as the existent human data compression will be made implausible. This means that consciousness cannot be logically separated into mechanical parts.
These representations, however groundbreaking, are but finely tuned attempts as an actual model for creating machines capable of mimicking or possessing a human mind is not the real deal and not in any way close to the functionality of the human labyrinth mind.
This brings us to a paradoxical conclusion. Suppose further extensive research is carried out over the coming years with these existing landmarks in these varying branches of sciences. Would we achieve a breakthrough in machine consciousness leading to a new generation of machines, or rather keep hitting a brick wall signaling the peculiarity of self-awareness to only the human species?