For a long time, the question of artificial intelligence consciousness remained a mystery to scientists, philosophers and technologists. The researchers’ report, published on the preprint server arXiv, presents some key indicators that can be used to determine whether software is conscious.
One of the fundamental questions scientists face is: “Is artificial intelligence capable of consciousness?” There is no single theory of consciousness, which makes this research all the more challenging and fascinating.
The team of researchers considered various approaches and hypotheses in an attempt to find indicators of consciousness in AI. They focused on global workspace theory, which integrates the ability to think and perform tasks in parallel.
They also examined indicators from higher-order theory, including mindfulness and predictive processing.
In their report, the team concluded that currently none of the existing artificial intelligence systems are conscious. However, they emphasized that the lack of awareness in current AI does not mean there are technical limitations to creating similar systems that can meet these metrics.
“This is a big question,” says co-author Liad Mudrik. “The search is exciting… One of the most important aspects of projects aimed at defining consciousness is its ability to bring us closer to developing clear criteria for consciousness.”
Consciousness in AI may be an accidental side effect of its complexity. Given that consciousness in animals is associated with high levels of cognitive skills, more advanced artificial intelligence systems could suddenly become conscious without prior planning.
Although scientists have not yet drawn definitive conclusions, this step brings us closer to understanding the nature of consciousness and its possible manifestations in machines.
There is a clear trajectory in the search for consciousness in machines, but fundamental questions remain open, highlighting the importance of further research in this area.
