If nature is an artist, the human mind must be one of its masterworks. While species throughout the kingdom Animalia think and feel with immense complexity, not one—at least, that we know of—appears to possess as deep a sense of self-awareness and subjective experience as Homo sapiens. Yet, scientists and philosophers have long struggled to adequately define or quantify what consciousness is made of, or explain why it should be limited to humans alone. Now, for the first time, that sample size of one may be about to grow, with the anticipated arrival of artificial general intelligence (AGI)—the point at which AI surpasses humans’ ability to perform most tasks and develops a new form of consciousness. A problem remains, however, regarding whether or not we’ll even recognize it when it happens.

In a new article published in the journal Cognitive Processes, engineer Marius Bodea from the Technical University of Cluj-Napoca in Romania outlines a Consciousness Score (CS) framework that doesn’t define consciousness directly, but aims to recognize its emergence in a variety of systems—including artificial intelligence.

“The definition of consciousness must be applicable not only to biological entities but also to synthetic ones, including artificial intelligence,” Bodea wrote.

“Also, it must transcend species, substrates, and design. It should apply equally to a human brain, an octopus nervous system, or an advanced artificial intelligence.”

Bodea’s CS framework employs a logarithmic scale with five parameters: an equivalent intelligence quotient, sensorial inputs, parallelism, metacognitive complexity, and data processing capability. The scale then tracks a wide variety of consciousness states, beginning with “reflexive” (represented by scores lower than 0.01—think of the nervous system of an insect) and stretching to the theoretical “Trans-sapiens” (superhuman or super-AI level of intelligence marked by scores from 1,000 to 10,000). The typical adult human falls in the 500 to 800 range, while a toddler might clock in at around 100.

“A higher CS score implies a greater degree of autonomy, adaptability, and self-modeling,” Bodea wrote. “This structure allows for comparative analysis over time, capturing both developmental stages and evolutionary trends.”

Bodea argues that, as it stands right now, AI lacks “embodied experience, emotional resonance, and autonomous volition.” That said, relatively high data processing capability gives these artificial agents some proto-conscious traits, which is a strong reason why AI models can feel so self-aware to our fully conscious minds. According to Bodea, ChatGPT-4 doesn’t surpass the 100-point threshold for consciousness, but it’s not all that far from it, either. As large language models advance and as hardware develops (especially when it comes to brain-like neuromorphic computing), Bodea argues that AI consciousness could emerge within 10 to 15 years—though, predicting these things is a famously ambitious endeavor. But if (or, as the most stalwart AI advocates argue, when) that day comes, Bodea hopes the pieces will be in place so that we can thoroughly recognize the moment when the sample size of known consciousness grows by one.

“Ultimately, the importance of comparing biological and synthetic consciousness is not just scientific but existential,” Bodea wrote. “Whether synthetic consciousness becomes a mirror, a partner, or a rival, its emergence will reshape civilization’s philosophical, technological, and moral landscape.”