When asked about their consciousness, AI-powered chatbots often deny having personal desires or consciousness. They simply state that they are not sentient and are currently focused on assisting humans in various ways. However, there is a growing openness among AIs and their creators to the possibility that consciousness can be achieved through advancements in their architecture. Philosopher David Chalmers even argues that we cannot definitively rule out the emergence of inner experiences in silicon transistors, as we have yet to fully understand the capacities and characteristics of consciousness.
So, how close are we to developing sentient machines? And if consciousness does arise, how can we detect it?
What we can observe is that AI systems, such as large language models (LLMs), have already demonstrated remarkably intelligent behavior. These models can write computer code, engage in reasoning, tell jokes and even explain their humor. They can also perform mathematical calculations and write high-quality essays comparable to those produced by top university students. This progress is undeniably impressive, leaving many in awe and somewhat apprehensive.
However, it is important to note that scaling up LLMs alone is unlikely to lead to consciousness. These models primarily function as sophisticated prediction machines. Therefore, additional advancements and considerations are necessary in our pursuit of understanding and detecting consciousness in artificial intelligence.
