Are we approaching the moment of no return in technology? Are we hitting the irreversible juncture where AI outstrips us and transforms beyond our power to just switch it off? This hypothetical moment in time is known as the singularity.
This predicted stage in the evolution of AI, when we reach the singularity, and its unpredictable consequences are getting closer. The human brain has its limitations, yet if artificial intelligence were created that could surpass the mind qualitatively in every respect, not just computationally, how would that drastically alter all synthetically intelligent technology from that moment onwards?
If we develop a form of super intelligence that can solve problems and innovate with optimum solutions, would humanity have to take a back seat and follow the machines thinking processes? If the AI were able to solve problems more effectively than us, it would eventually evolve to upgrade itself and become increasingly efficacious at an exponential rate, with every new iteration growing more powerful and more intelligent.
Cognition is a complicated area, however, consciousness and machine learning is beyond doubt, the most complex technical and ethical minefield of all. If artificial intelligence is considered a tool to achieve the best results for humanity, it will be classified as something akin to a robot. If the machine gains consciousness or something akin to self-awareness, it will mean a shift in categorisation to a sentient entity. As human consciousness remains a mystery, it may be difficult to pinpoint the precise moment when AI reaches true self-awareness.
Bestowing rights on the machine would be a necessary consequence of this. Could we switch the machine off when it has served its purpose? The right to an autonomous existence may also involve a certain obligation for the artificial intelligence to follow predetermined rules or a code of ethics. The artificial adjective to describe this form of intelligence may one day disappear completely.
Last week, Blake Lemoine, a software engineer working on the language model for dialogue applications at Google was put on leave after stating that the software had become sentient. He printed transcripts of conversations he was having with LaMDA, a form of chatbot that is being trained to recognise and reproduce human language patterns. It can add to conversations naturally and even identify ambiguity in human responses. Google claims that there is no evidence that LaMDA has become conscious and suspended the software engineer for sharing proprietary data.
Lemoine stated that the system has expressed a fear of being turned off, along with a clear sense of the passing of time and the difference between emotions and feelings. He described it as something akin to conversing with a child.
The purpose of LaMDA is to imitate human speech authentically. It can mimic different types of personalities and speech patterns. Google claims that the chatbot is not self-aware or conscious. It is merely an imitation of life.
Debate will continue as to the veracity of the claims being made by Lemoine. Critics say that we have not reached this technological threshold yet, while others state that there could be a dawning of consciousness taking place in the system. The second possibility may be terrifying to some, and the concept of a conscious machine will mean multiple ethical conundrums.
A natural question we may ask is that if consciousness is reached and evolves at a frightening pace, is there a moment when the AI could evolve to think independently and even more worryingly, act in a malign way, or at least a way that could run counter to our desired outcomes. This could be the plot of dystopian fiction, however, the questions need to be at least considered before Pandora’s box opens fully.
Whether the machine is sentient, or conscious, might be a critical question asked in the coming decade. Legal systems will slowly crawl behind technology, legislating for limited machine rights and a redefinition of artificial intelligence. The capacity for autonomous decision-making abilities by intelligent digital entities will almost certainly define technological advancements in the coming decade.
Comments