I think Turing was right to limit his test to machine intelligence rather than consciousness. Consciousness is deeply subjective and cannot be reliably measured. The Turing Test works by comparing linguistic performance, but the outcome depends heavily on the interrogator’s own background and experience. For example, someone with advanced coding skills might know how to ask questions that expose “bot-like” behavior, while someone else could be more easily convinced by the same responses. This variability makes consciousness impossible to standardize in a test like this.

I actually have a personal example of this. Recently, my dad was admitted to the ER for a medical condition that could have been easily taken care of by his primary care physician, as he correctly diagnosed his condition, but essentially refused to treat him until the situation worsened with painful blisters and rashes. In other words, he relied strictly on medical protocol, which is, basically, medical knowledge in the narrowest sense, while ignoring the broader context of my dad’s pain, his age, and other medical factors, despite having his entire medical chart in front of him. That might pass for “knowledge,” but it fails what we might call the “consciousness test,” since he showed no awareness of the overall situation. By contrast, when I consulted ChatGPT about the same diagnosis, its responses were not only more sympathetic, but it was also way more medically sound, pratically the same early treatment protocols that the ER doctors has told us that our PCP should have taken, that could have prevented my dad’s suffering and the eventual urgent care and ER visits where stronger interventions were needed due to the lack of earlier and easier interventions.

If we were to imagine this scenario in a Turing Test setting, the outcome would depend entirely on the interrogator, where a medical student, trained to adhere strictly to textbook protocol, might have given my dad’s PCP a pass. But a seasoned doctor, specifically a doctor with more experience, as was confirmed by the disbelief of the ER nurses and physicians when they learned of the PCP’s approach, would have immediately judged him as failing. In this case, since ChatGPT would have matched the PCP in terms of medical knowledge, it shows that the Turing Test cannot serve as a measure of consciousness, but only of the level of knowledge or intelligence displayed in a given context.