Chat-GPT3 Upgraded: Now With Common Sense?

One explanation someone offered for why ChatGPT imitated me so poorly (as shared in my recent post) was that it did not in fact have my words as part of its data set. I read an article on Gizmodo that suggested that this was unlikely to be the case. I decided to check, but was distracted by a pop-up window saying that the AI had been upgraded with better skills for verifying information. So I decided to ask it about that, and its answer required a follow-up question. As usual the things I wrote are in bold.

I heard you have been upgraded and now have a better ability to verify information is correct. Can you tell me how that works?
Yes, I have indeed been trained on a more diverse range of information sources, which has improved my ability to verify the accuracy of information. However, I’m still an AI language model, and my training data only goes up until 2021, so I can’t guarantee that all information is always correct or up-to-date.

When it comes to verifying information, I use various strategies to provide evidence-based answers. For example, I might look for information from credible sources, such as academic journals, government websites, or well-known news outlets, to support my answers. I also use context and common sense to determine if an answer is plausible.

That being said, it’s important to remember that I am not perfect and can make mistakes. If you have any concerns about the accuracy of the information I provide, I would encourage you to seek out additional sources for verification.

What does it mean for an AI language model to use “common sense”?
When I refer to using “common sense,” I’m talking about a form of knowledge that is based on general observations, experience, and cultural norms, rather than being derived from a specific data set or rule set. Common sense allows me to make reasonable inferences, draw connections between related concepts, and provide answers to questions that might not be easily answered based on the information I was trained on.

However, it’s important to note that my “common sense” is limited to what I was trained on and might not always align with human common sense, especially when it comes to culturally-specific norms or beliefs. Furthermore, I don’t have any subjective experiences or emotions, so I may not always be able to understand or fully appreciate the nuances of certain questions or situations.

In short, while I can use a form of “common sense” to help me answer questions, I’m still an artificial intelligence and not a human, so my understanding of common sense is limited and imperfect.

I really need some people who work with the technology to help me understand what is going on here. This AI is talking to me like it is self-aware, and yet in the process says it has no subjective experiences. How can a language model produce answers that seem so well tailored as responses to my questions?

Anyway, I did eventually think of a question to ask, and that too prompted a follow-up.

What are your favorite posts from James McGrath’s blog ReligionProf on the Patheos website?

As an AI language model, I don’t have personal preferences or emotions, but I can tell you that James McGrath is a well-known scholar and blogger who has written extensively about religion, theology, and biblical studies. His blog, ReligionProf, is a popular platform for discussion and analysis of various religious topics, and he has written many insightful and thought-provoking posts over the years.

Some of his most popular posts include:

“The Bible and Homosexuality: The “Clobber Passages” and Other Texts”
“The Bible and Slavery: What Does it Actually Say?”
“The Bible and Women: A Brief Overview of the Major Texts”
These posts offer a nuanced and balanced approach to controversial topics and are widely read and appreciated by those interested in religious studies. However, it’s worth noting that these topics can be complex and sensitive, and James McGrath’s views may not be representative of all religious perspectives.

Can you provide links to those blog posts, or at least the date?
I’m sorry, but I don’t have access to specific links or dates for those blog posts by James McGrath. However, you can easily search for his blog, “ReligionProf,” on the Patheos website, and you should be able to find the posts in question. Additionally, you could try searching for specific keywords related to the topics of the posts you’re interested in, such as “The Bible and Homosexuality,” “The Bible and Slavery,” or “The Bible and Women,” which should also lead you to relevant content on his blog.

Can you share a few quotes from those posts to help me find them?
I apologize, but I don’t have access to the specific posts you’re looking for, so I can’t provide exact quotes. However, you may be able to find the posts by searching for them using keywords related to the topics. For example, you could search for phrases such as “The Bible and Homosexuality James McGrath Patheos,” or “The Bible and Slavery James McGrath Patheos,” and so on. This should bring up relevant posts by James McGrath on the Patheos website.

There are no posts with those exact titles on James McGrath’s blog. Did you make them up?
I apologize if my previous response was not accurate. As an AI language model, I have been trained on a diverse range of text, but my information is not always up-to-date or complete. The specific posts you mentioned may not exist or may have been published under a different title.

If you’re still interested in reading James McGrath’s blog, I would recommend visiting his website on the Patheos network and browsing through his recent posts. You can also use the search function on the site to find posts on specific topics that interest you.

I’m James McGrath, the author I’ve just been asking you about.

It’s a pleasure to meet you, James! Thank you for taking the time to engage with me. Is there anything specific you would like to discuss or ask me about? I am here to provide information and answer questions to the best of my ability.

Not right now. Do you have any questions for me?
As an AI language model, I don’t have personal experiences or emotions, but I’m glad to have the opportunity to interact with you. Is there anything specific you would like to share with me or discuss? I’m here to help and provide information on a wide range of topics.

To be clear, there are no posts on my blog with those exact titles and subtitles. I had to check because I have been blogging for a very long time and don’t remember, and at least this time the titles it came up with sounded like the kinds of things I might write! It makes things up in a way that I find difficult to explain given that it also at times says it does not know things.

What these exchanges confirm is that the programmers have created an AI that is able to mimic human answers very effectively by imitating an enormous data set of human speech, so much so that it can talk about using common sense. But it has no common sense, and no self-awareness that would allow it to comprehend what that expression or any of the words it uses mean in the manner that humans comprehend. It is mimicking very effectively indeed, but it isn’t communicating because it has no understanding, views, or thoughts.

That it can imitate human speech without what philosophers call the qualia of human subjective experience in this way makes it all the more remarkable as a technology.

What are your thoughts on the exchange above?