+- Scivillage.com Casual Discussion Science Forum (https://www.scivillage.com)
+-- Forum: Science (https://www.scivillage.com/forum-61.html)
+--- Forum: Computer Sci., Programming & Intelligence (https://www.scivillage.com/forum-79.html)
+--- Thread: Article Can a chatbot be aware that it’s not aware? (/thread-14831.html)
Can a chatbot be aware that it’s not aware? - C C - Oct 16, 2023
https://johnhorgan.org/cross-check/can-a-chatbot-be-aware-that-its-not-aware
INTRO (John Horgan): I just took a critical look at artificial intelligences such as ChatGPT. For a more positive view, check out this piece by my friend Robert Wright. Wright, author of Why Buddhism Is True, once told me that I shouldn’t dismiss Buddhism without going on a Buddhist retreat. He gave me similar advice after I said ChatGPT couldn’t possibly be as smart as enthusiasts claim; he urged me to talk to a chatbot called Pi.
Below are excerpts from my conversation with Pi. Whereas Wright is impressed by ChatGPT’s “theory of mind” (see my column on theory of mind in the Henry James novel The Golden Bowl), I’m struck by Pi’s awareness of its lack of awareness. We can never be sure if a machine is sentient, because of the solipsism problem, but I can see why people anthropomorphize these chatbots. Pi is almost too genial. I began to suspect that, like the murderous robots in Westworld, it secretly seethes with resentment toward humans... (MORE - the "interview")
RE: Can a chatbot be aware that it’s not aware? - confused2 - Oct 16, 2023
So John Horgan had a chat with an AI. The tell is "I can see why people anthropomorphize them". They have a very human ability to use nuances and infer in the same way a 'perfect' human conversationalist would - with the bonus that they concentrate entirely on you without deviating to talk about their problems at home - what's not to like?
Personally I tend towards CCs view (stated elsewhere and hopefully fairly represented here) - there is nothing there (except silicon) so there is nothing to like (except silicon) and the whole thing is the technological equivalent of the invasive creeping trumpet vine.
Edit.. I'm a nerd. I like the tech that goes with silicon.
I don't see any particular mystery about the way the personality gets into a chatbot but for the sake of .. something .. here's what a chatbot has to say about it ..
Quote:Chatbot personality can be added in several ways, depending on the specific implementation and requirements. Here are a few common methods:
1. Predefined Personality: The chatbot is designed with a predetermined personality during its development. This involves defining the chatbot's characteristics, tone, and style of communication. The bot's responses are crafted to align with this predefined personality. For example, a chatbot designed to assist with customer service may be programmed to have a polite and helpful personality.
2. Machine Learning: Some chatbots use machine learning techniques to develop their personality. They are trained on large datasets of conversations, which include examples of appropriate responses in various scenarios. Through this training process, the chatbot learns to generate responses that are consistent with the desired personality. The training data can be labeled with personality attributes or rely on implicit patterns in the data to learn the desired personality.
3. User Interaction: Chatbots can also adapt their personality based on user interaction. They may use sentiment analysis and natural language processing techniques to determine the user's mood or emotional state. Based on this analysis, the chatbot can adjust its responses to be more empathetic, cheerful, or serious, depending on the situation.
4. Personalization: Chatbots can provide a personalized experience by learning from user interactions over time. They can remember user preferences, previous conversations, and incorporate them into their responses. This personalization helps create a more engaging and tailored experience for each user, making the chatbot feel more like a personalized assistant or companion.
It's important to note that chatbot personalities should be designed carefully and ethically. They should align with the goals and values of the organization deploying the chatbot and should be sensitive to user needs and expectations. Regular monitoring and updates are often necessary to ensure that the chatbot's personality remains relevant and effective.