AGI may not be exactly the same as a machine
Posted: Thu Dec 26, 2024 9:36 am
An experiment with ChatGPT published in April 2023 makes this clear. As part of this experiment, researchers confronted healthcare professionals, on the one hand, and ChatGPT, on the other, with almost 200 medical questions from an online patient forum.
The answers were then evaluated by healthcare professionals. And on average, ChatGPT 's answers were rated as more correct and more precise than those put forward by real healthcare jiangsu mobile phone numbers database professionals. However, this is not the most disturbing data revealed by the experiment, which also evaluated the empathy of the answers and concluded that ChatGPT's responses were significantly more empathetic than those of healthcare professionals.
It can be inferred from this that machines can sometimes imitate or generate empathy better than humans , even though empathy is probably the most eminently human emotion.
Along the same lines, the "consciousness perceived in machines" could be governed by a similar pattern, so that once a certain threshold is exceeded, it would be totally impossible to decipher whether a machine really owns its own consciousness or simply pretends to have consciousness thanks to the superintelligence housed in its entrails. And perhaps in three months, in a year or in five years the superintelligence will know more about us than we would like, concludes Lobo.
The answers were then evaluated by healthcare professionals. And on average, ChatGPT 's answers were rated as more correct and more precise than those put forward by real healthcare jiangsu mobile phone numbers database professionals. However, this is not the most disturbing data revealed by the experiment, which also evaluated the empathy of the answers and concluded that ChatGPT's responses were significantly more empathetic than those of healthcare professionals.
It can be inferred from this that machines can sometimes imitate or generate empathy better than humans , even though empathy is probably the most eminently human emotion.
Along the same lines, the "consciousness perceived in machines" could be governed by a similar pattern, so that once a certain threshold is exceeded, it would be totally impossible to decipher whether a machine really owns its own consciousness or simply pretends to have consciousness thanks to the superintelligence housed in its entrails. And perhaps in three months, in a year or in five years the superintelligence will know more about us than we would like, concludes Lobo.