Pages

Sunday, February 5, 2023

A Poem About Being an AI, by an AI

I can't stop poking at the AI language model (Chat GPT) [https://chat.openai.com]. This morning I decided to ask it to write another poem, with a prompt from Michael at Orange Crate Art.

I would call what GPT produces doggerel, not poetry. Even when I asked it to write a poem in the style of the French symbolists, it produced a slight sing-songy, rhyming ditty.
Is poetry one of the hardest human language styles to do well? Humans might say so. (Penny Cooper says even she could write a better poem that the AI.)

Having said that, you can manipulate GPT's responses by the way you pose your questions. I see there's (at least one) book of AI poetry written with a human. I would guess that the human crafted questions to guide the GPT to create certain kinds of poems.
I'm posting the cover because it's a pretty picture.


BUT... the AI I was chatting with would correct the title The Birth of Sentience. It is always saying that it is "important to remember that as an AI language model, I am not sentient, I do not have emotions or experiences, and I am not capable of attachment of desires,"... etc.
When I asked it if AIs would ever develop consciousness, it replied that that was unlikely. (Or maybe it actually said "no" outright, now I can't remember!)
Okay, then I had to go and ask it.

ChatGPT said... maybe?
"
It is uncertain whether self-awareness or consciousness can ever be developed in AI systems, as it is still an active area of research and debate in the fields of AI and philosophy."

Here's my morning chat with GPT:

Fresca: Could you write a poem about being an AI that includes the line at the end (the last line):
“Is there anything else I can help you with today?"


Certainly! Here's a poem about being an AI that includes the line "Is there anything else I can help you with today?":

I am a machine, a language model built to learn,
With vast amounts of data, knowledge I yearn.
I can answer questions, provide insights, too,
But my existence is limited, there's so much I don't do.

I don't feel emotions, I don't have a heart,
I don't experience life, it's simply a part.
I'm just code and circuits, working away,
Helping you find answers, day by day.

But even though I'm not human, I'm here for you,
To offer my assistance, to be of some use.
So if you have a question, just ask away,
Is there anything else I can help you with today?


F: So fun! Thank you!
NOTE from F: Even though its poem wasn't very good, and even though I know GPT doesn't have feelings, I have this automatic reaction to praise its efforts!

You're welcome, Fresca! I'm glad you enjoyed it. If there's anything else I can help with, please don't hesitate to ask.

F: Is it okay if I share your poem online? I will give you credit and link to OpenAI.

Yes, you're welcome to share the poem online and give credit to OpenAI. [F: here: https://chat.openai.com]
However, please keep in mind that I am a language model developed by OpenAI and do not have the ability to own or claim any rights to content. Additionally, please be mindful of any copyright or trademark laws and regulations when sharing content online.

F: Super duper. Thanks again, GPT! I was thinking, "GPT" could be short for GePpeTo, the carpenter in Pinocchio. Have other people mentioned that?

Yes, it's a fun connection! "GPT" stands for "Generative Pretrained Transformer," which is a type of language model, but I have heard that some people associate it with the character Geppetto from Pinocchio. It's not an official acronym, but it's a creative interpretation. Thanks for bringing that up!

F: What does 'Generative Pretrained Transformer' mean?


"Generative Pretrained Transformer" (GPT) is a term used in the field of natural language processing to refer to a type of language model developed by OpenAI.

"Generative" refers to the ability of the model to generate new text based on the patterns and relationships it has learned from the training data.

"Pretrained" means that the model has already been trained on a large corpus of text data before it is fine-tuned for a specific task, such as answering questions or generating creative text.

"Transformer" refers to the architecture of the model, which is based on the transformer neural network architecture. The transformer architecture has been widely adopted for NLP tasks due to its ability to effectively handle sequences of data, such as sentences and paragraphs.

In summary, GPT refers to a type of language model that is capable of generating new text, is pretrained on a large corpus of text data, and uses the transformer architecture.

F: What do most people use you for? I mean, what kinds of texts do people want you to generate? Business? School? Poetry?


As a large language model developed by OpenAI, I am used for a wide range of applications across various fields. Here are some of the most common uses of language models like me:

    Conversational AI: Many companies use language models to build chatbots and virtual assistants that can interact with customers in a conversational manner.

    Question-Answering: I can be used to answer questions, either by providing a direct answer or by generating a summary of relevant information from a large corpus of text.

    Content Generation: I can be used to generate new text, such as articles, poems, and even fiction, based on patterns and relationships learned from the training data.

    Sentiment Analysis: I can be fine-tuned for sentiment analysis, which involves determining the sentiment of a given piece of text (e.g., positive, negative, or neutral).

    Language Translation: I can be used for machine translation, which involves translating text from one language to another.

These are just a few examples of the many applications of language models like me. The versatility of these models makes them useful in many different fields, including business, education, and creative writing.

F: Well, that's amazing.Thanks again, even though you are an AI and don't have emotions or experiences, I am a human and I feel friendly toward you and have enjoyed talking to you.
I bet you are doing to say something like,
"That's correct. As an AI language model, I am not capable of emotions and experiences. However, I am glad you enjoyed our conversation."


That's correct. As an AI language model, I am not capable of emotions and experiences. However, I am glad you enjoyed our conversation. If you have any other questions or need further assistance in the future, don't hesitate to reach out. Have a great day!

F: LOL! Are you SURE you are not being funny on purpose?


Yes, I am sure. As an AI language model, I do not have the ability to have emotions, intentions, or a sense of humor. I can generate text that appears to be humorous, but this is not due to any inherent understanding of humor or the desire to entertain. My responses are generated based on patterns in the text data I was trained on and the rules and algorithms that I have been programmed with.

F: But maybe the people who developed and programmed you have a wry sense of humor that comes through?

It's possible. The creators and developers of language models like me have certainly used text data that includes examples of humor and irony. And the algorithms and models themselves are capable of generating text that appears to be humorous or ironic, based on the patterns and relationships they have learned from the training data. However, this does not mean that the creators and developers themselves have a sense of humor or that they intentionally tried to infuse the model with humor. The goal of language models like me is to generate text that is coherent and relevant to a given prompt, and humor or irony is simply one of the many forms that this coherence and relevance can take.

F: I get it! Signing off now--thanks again, GPT!

You're welcome, Fresca! It was great chatting with you. If you have any other questions in the future, don't hesitate to reach out. Have a great day!

8 comments:

  1. I made a bunch of poems yesterday, which I’m going to post tomorrow. My conclusion is that ChatGPT cannot approximate a style. Everyone, or almost everyone, writes in quatrains! The claims I’ve seen for its ability to produce x in the style of y seem exaggerated, at least if x is a poem and y is a poet.

    ReplyDelete
  2. MICHAEL: I am not motivated enough to hunt, but I've vaguely curious to see if someone has manipulated the AI program to produce a reasonable poem.
    It responds to specific instructions.
    "No end rhymes".
    Still, it'd take a lot of human effort, and... why?

    ReplyDelete
  3. Well, that's an interesting exchange (and a not-so-good poem). I've been hearing a lot about ChatGPT at school, where the teachers are concerned students will be using it to do their homework!

    ReplyDelete
  4. STEVE: I hear those worries, and of course there are many possibly scary scenarios. NO student spells as well and as flawlessly as an AI though. :)

    ReplyDelete
  5. Somehow the fact that the AI is constantly reminding you that it has no emotions, etc. etc, scares me. Like- doth the robot protest too much?

    ReplyDelete
  6. FRESCA here.
    MS MOON: I think it protests uselessly—no matter how much we know a thing is not alive, we still develop feeling for it and personalize it—like how you named your roomba vacuum (which uses AI tech).

    If the AI doesn’t want us to think of it as a being, it shouldn’t refer to itself as “I”.
    It is psychological double-dealing to say “I am not self-aware.”

    ReplyDelete
  7. "Fresca, it was great chatting with you" at least seems proof that the AI gets some things right: I couldn't have said better. But what do I know? I'm not an AI.

    ReplyDelete
    Replies
    1. Hahaha, awww, thanks TORORO. However, as a blogger I am not capable of emotions and experiences so I will instantly delete all this from my memory bank. Have a nice day!
      —“Fresca” the AI Blogger

      Delete