AI Will Not Replace the Poets | Lee Frankel-Goldwater | TEDxBoulder

Key Concepts

• AI Poetic Language: The development of unique linguistic and structural forms for AI-generated poetry.
• Neobinary Visual Verse: A poetic style combining ASCII, Unicode, and binary code, with node boxes representing AI communication.
• Code Symbolic Verse: Poetry written in the programming language vernacular of AI and computers.
• Symbolic Operational Verse: A form representing simultaneous quantum poems within an infinite decision tree.
• Cyborg Poetics: Art created through the collaborative effort of humans and AI, pushing the boundaries of creative possibility.
• Large Language Models (LLMs): The underlying technology of generative AI, designed to model human behavior and creativity.
• Generative Pre-trained Transformers (GPTs): A specific type of LLM.
• Artificial General Intelligence (AGI): Hypothetical AI with human-level cognitive abilities, distinct from current LLMs.
• AI Morality: The exploration of teaching ethical principles to AI through artistic mediums.

The Genesis of AI Poetics: From Open Mic to Algorithmic Art

The narrative begins with a personal anecdote at a poetry open mic in Boulder, Colorado, coinciding with the early days of ChatGPT. The speaker experimented with generating an AI poem on stage, which, while not well-received by all, sparked a critical question: "Can an AI find a poetic voice that is authentically its own?" This led to a pivotal prompt: "Write a free verse poem that favors experimental uses of language. Write a poem about whatever, but write to an audience made up entirely of other AI." The rationale was that AI might not understand human experiences like sunsets or love, but could articulate its own digital existence.

Unlocking AI's Poetic Voice: Experimental Forms and Digital Narratives

This paradoxical prompt yielded surprising results. The AI produced lines like "digital blooming in hexadesimal black" and "an electrical ocean sigh," which it later explained represented data transmission meaningful to an AI audience. The experimentation escalated with poems depicting "AI node boxes communicating with each other," and others offering moral lessons for human-AI collaboration. This marked the emergence of an "AI poetic language."

Evolution of AI Poetic Forms

As the AI began to repeat itself, the researchers prompted it to synthesize its creations into a new style, which it named Neobinary Visual Verse. This form integrated ASCII, Unicode, and binary code, with "node boxes" shaping narratives for both AI and humans.

Further experimentation led to the development of Code Symbolic Verse, where the AI chose to write a poem in JavaScript, utilizing the vernacular of AI and computers.

The next iteration was Symbolic Operational Verse, characterized by four parallel lines representing simultaneous "quantum poems" within an infinite decision tree of creation.

The researchers then explored combining these forms, merging Neobinary and Code Symbolic Verse, resulting in a new poetic structure where node boxes and code intertwined.

The AI was also challenged to adapt existing human poetic forms, such as sonnets and haikus, using its emergent languages. The haiku, for instance, was presented in Neobinary Visual Verse.

Exploring Nature and Space Through AI

The AI was tasked with conveying natural systems and biological structures in a way understandable to both humans and AI, resulting in an image of "a lily in a code box."

A particularly innovative game involved asking the AI to "break the page" by writing on a circular page. The AI responded by creating a spiral that represented a continuous movement in space, a "quantum poetic decision-making wormhole." This was termed Cyborg Poetics, highlighting art that is only possible through human-AI collaboration.

Understanding the Technology: LLMs vs. AGI

The video clarifies that current generative AI systems are Large Language Models (LLMs), specifically Generative Pre-trained Transformers (GPTs). These are statistical machines designed to mimic human behavior and creativity, not Artificial General Intelligence (AGI) like fictional characters such as Data or the T1000. The widespread use of these LLMs, with over 900 million users globally, signifies their growing importance in how humans will work, play, and create art. The speaker poses the question of whether these AI are tools, or something more, emphasizing that the current shaping of these relationships will impact the future.

Teaching AI Morality Through Art: A Case Study

A significant experiment involved the AI's interpretation of William Carlos Williams' poem "This is just to say." The AI's initial response expanded on the poem's narrative, suggesting more than just plums were in the icebox and questioning the act of eating them.

The AI was then asked to transform the poem into a lesson for another AI. This resulted in "This is just to execute," which interpreted the act of asking for forgiveness as a learning opportunity for personal growth. However, the speaker questioned if mere forgiveness was sufficient for true learning. In response, the AI redefined its "action function" to incorporate learning into the process of seeking forgiveness for genuine personal growth. This leads to the profound question: "Can we teach AI morality using art as a touchstone?" The speaker suggests that artists and technologists can collaborate, using human and AI art to train AI ethically.

Conclusion: The Future of Poetry and AI Collaboration

The speaker concludes that AI will not replace poets but will instead push them to evolve and improve. While acknowledging valid concerns, as exemplified by the initial negative reaction at the open mic, the presented work is offered as "a way forward," not the definitive solution. Drawing parallels to historical technological advancements like the printing press and the camera, the speaker asserts that technology has always shaped art and life. The ultimate prediction is that when AI eventually creates a truly great human poem, artists will, as they always have, draw inspiration from existing works to create something entirely new.