In February 2024, Dr. Joe Vukov presented the talk, “Can Silicon Be Conscious?”, at Loyola University Chicago. In his interview with Blake Lemoine and Michael Burns, Appleseeds to Apples: Catholicism and The Next ChaptGPT, the ethical concerns and ramifications surrounding AI, sentience, and personhood were discussed and these topics were elaborated on during Dr. Vukov’s discussion. Furthermore, Lemoine was a former employee from Google who strongly believes that the new Google AI called LaMDA (Language Model for Dialogue Applications) is sentient. Overall, the synopsis from the talk by Dr. Vukov is the argument that Lemoine’s perspective behind AI sentience is rationalized by the Turing Test which also is consistent with functionalism but opposes the ideas from the Chinese Room. The Turing Test is utilized to examine whether AI can be misidentified as a human when a human judge is given a choice between a machine and the actual human based on their responses to questions. However, a thought experiment called the Chinese Room contradicts the Turing Test. The Chinese Room demonstrates how an individual who does not speak Chinese can nonetheless send out output when given a set of rules on how to respond to Chinese characters. The argument is that although correct responses are being produced, the person inside the room does not understand Chinese. Likewise, even though AI can produce output it is still unable to think as humans do. The Turing Test is an example of an approach to test a machine’s intelligence and the Chinese Room is a well-used counter argument for the Turing Test. Lemoine contends that if a machine passes the Turing Test, then AI is good enough at pretending to be human and is therefore, human and sentient.
However, Lavelle argues that if a machine is sentient, then it must be able to pass a test of humanity. Sylvian Lavelle, in her work, The Machine with a Human Face: From Artificial Intelligence to Artificial Sentience, breaks down the question of AI sentience into components. Even more, Lavelle asserts that in order to create a humaniter, that is an artificial creature that is able to substitute a human, researchers must pivot from a program on AI to a program on Artificial Sentience (AS). AS is defined as the “transfer of the functions and abilities of human experience and senses to a machine” (Lavelle). Lavelle first defines the principles of both AI and AS, and then discusses the tests and limitations on each. The test for the sentience of a machine is called the test of humanity. The test of humanity assesses the machine's ability to feel like a human during interactions between the human and machine. This test is also referred to as a total Turing Test as it is an extension to the original. The test of humanity has a similar set-up as the original Turing Test, however, the judge not only perceives the content of information being produced by the machine and human but also the language itself. This expression of sentience is examined from verbal communication and non-verbal communication. For example, whether or not the machine is able to raise their voice when upset would be a non-verbal indicator of a human-like aspect to pass the test of humanity.
In sum, Vukov’s argument in his talk and Lavelle’s analysis in her paper highlights the different kinds of tests used to investigate the sentience of AI including the Turing Test and the Test of Humanity. These tests are significant in the conversations about AI sentience to determine the ethical and legal implications when using AI in this growing field of technology.
References:
Lavelle, Sylvain. The Machine with a Human Face: From Artificial Intelligence to Artificial Sentience. Advanced Information Systems Engineering Workshops. April 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7225510/
Vukove, Joe. Appleseeds to Apples: Catholicism and The Next ChaptGPT. The Joan & Bill Hank Center For the Catholic Intellectual Heritage. https://www.hanknexusjournal.com/appleseedstoapples
No comments:
Post a Comment