Can AI be Conscious? Insights from Three Diverse Perspectives
Existing perspectives and not definitions provide a good starting point
The public imagination has been captured by a the possibility of AI being conscious. This may take the form of AI enabled robots or the dismembered AI agents based on large language models such as ChatGPT. But what is consciousness ?
Starting with definitions of consciousness usually leads to a dry discussion or ties up the conversation in knots.
This article draws on three well-known perspectives of consciousness to develop a working understanding and explores how conscious AI may be based on each perspective.
The medical perspective
The first perspective of consciousness is the medical one. The image that comes to mind is a person in a hospital bed (possibly after an accident or stroke) being assessed by a doctor to see if they are conscious. In this perspective the doctor is assessing whether the person is aware of their environment and can respond to events in the environment. This is usually measured through scales such as the Glasgow Coma Scale that evaluates eye, verbal and motor responses. In more usual settings outside of emergency scales such as the Level of Cognitive Functioning Scale are used to assess various aspects of cognitive functioning like memory, orientation, conceptual thinking and calculations. It is likely that AI will pass many of these tests, which is remarkable but at the same time this is not what is getting the world excited about AI and consciousness.
The perspective of subjective experience
The second perspective of consciousness is the one of subjective experience (the why and how of subjective experience is known as the hard problem of consciousness).
Amongst a group of friends looking at the setting sun on a beach together the hue of colours and the associated feelings, emotions and reflections is unique to each person in the group. This then affects moods, thoughts and emotions of each person in different ways.
If our subjective experience were mostly the same we would lose a lot of our individuality and humanity. This individuality and humanity shines through in our interactions through care, concern, empathy in our interactions with other human beings, animals and nature. Our individual subjective experience also influences much of the unique style of non verbal communication through body language, variation of vocal tones and pace of speech, use of space and physical distance that we bring to all our interactions. While AI probably surpasses humans in many different kinds of intelligence, problem solving and even adapting its behaviour to a specific situation it seems that it will be a long time before AI is conscious from the perspective of having and being able to draw upon its own subjective experience.
The perspective of what it means ‘to be’ human
The third perspective of consciousness is one that of ‘being’ as described by the German philosopher Martin Heidegger in his work Being and Time (Heidegger is a difficult read; this is a good starting point for a fuller discussion of ‘being’). ‘Being’ is what it means ‘to be’ a conscious human through the continuous and varied set of experiences as we go through life. There are many ways of ‘being’ explored by Heidegger but the everyday mode and the identity mode are quite relevant to this discussion.
In the everyday mode of being we move around with ease, almost on autopilot, going about tasks such as getting up, brushing our teeth, catching public transport and going to work. Here ‘being’ or consciousness seems to be fairly routine, predictable, low effort and intuitive. AI will likely be competent and proficient in this everyday mode of being and take the drudgery out of our lives, although it must be said there is a sense of enjoyment and satisfaction that comes from being in the moment and enjoying the mundane things in our lives.
But this is not the only mode of being. Over time in life we start to care more about certain things rather than everything in general and build up a role or identity for ourselves that becomes central to our being. Some examples of these roles or identity are being an accomplished dancer or striving to be a caring parent. To do well in these roles we go through a learning curve and proactively and ‘consciously’ develop our expertise and take risks to become more accomplished. Although AI can be taught to do some professional roles (e.g. a data analyst or customer service agent) reasonably competently it is unlikely to stick through the trials and tribulations of being a dancer from the age of five of its own volition.
AI and consciousness will evolve and change over a long period of time. Existing perspectives drawn from diverse fields of knowledge can help us navigate the journey ahead.
