The Ascent of AI ...
How should we allocate credits in the new Human-AI collaboration
Last week we published the latest post on a contemporary or near contemporary polymath – the subject being Jacob Bronowski. David has followed Bronowski since the publication of his magnum opus, The Ascent of Man – an examination of humanity’s intellectual and technological evolution – in the 1970s. In writing the post we could have gone through a process of discussion, dialogue and desk research followed by drafting, iterating and honing a piece of writing … we of course just engaged with ChatGPT. This leads us to an important question … who gets the credit? Who or what authored the piece? If we should say that 95% of the words are from ChatGPT and 5% from us, is it correct to say that we – David and Adam – authored the piece?
In the age of generative AI, questions around authorship, originality, and collaboration are more pressing than ever. How do we measure the contribution of human effort versus AI in a piece of writing? Should a publisher, tutor, or reader simply categorize a text as ‘95% AI-generated’ and ‘5% human-generated’?
While this may seem like a straightforward metric, it doesn’t fully capture the complexity of the collaboration between a human and AI, particularly in crafting thoughtful, nuanced work like our piece on Jacob Bronowski.
Here’s what really happened behind the scenes in creating this post:
Prior knowledge and contextual awareness
David has been aware of Jacob Bronowski since his youth and has developed a view of his influence as a polymath and public intellectual. His familiarity with Bronowski’s work—especially The Ascent of Man—gave him a foundational understanding of his ideas and their relevance. This pre-existing knowledge shaped the direction we wanted the piece to take.
Engagement with primary and secondary sources
Recently, David revisited Bronowski’s book The Ascent of Man and studied it closely, which provided insight into his themes of interdisciplinary thinking, creativity, and ethics. He also watched an interview with Michael Parkinson, recorded shortly before Bronowski’s death (see below), where he eloquently articulated his key ideas. These engagements deepened his understanding of Bronowski’s work and helped him shape the piece's narrative.
Critical exploration
Before drafting the piece, David explored critiques of Bronowski’s work using ChatGPT, discovering that while his interdisciplinary approach had some general detractors, there was no specific critique of his intellectual rigor or professionalism. This informed our decision to position him as a largely unchallenged figure in the domain of interdisciplinary thought while acknowledging potential counterpoints.
Strategic structuring and framing
We provided ChatGPT with a detailed outline for the piece, specifying its structure: an exploration of Bronowski’s educational background, a discussion of his seminal work The Ascent of Man, an analysis of his principles, and reflections on his relevance in the AI era. The clarity of this structure ensured that the AI-produced draft stayed focused and coherent.
Incorporating key ideas and perspectives
From our knowledge of Bronowski, we introduced specific ideas, such as his likely views on AI and its implications for the future of work, creativity, and meaning. This required a nuanced understanding of his philosophical and ethical stances—something AI alone could not provide without human direction.
Fact-checking and validation
Throughout the process, we used ChatGPT not only to generate text but also to verify facts, cross-check information, and ensure that the output aligned with Bronowski’s actual ideas. This iterative process refined the draft into an accurate and polished piece.
Selection of quotations
We explicitly requested key quotes from Bronowski’s work to add authenticity and provide a direct flavor of his thinking. These quotes were then carefully integrated into the final piece to reinforce key arguments.
The reality of AI-human collaboration
So, can we really say that this post is ‘95% ChatGPT and 5% human’? While it’s true that ChatGPT produced the majority of the draft text, the human contribution encompassed far more than merely typing prompts. Our involvement entailed:
Conceptual direction: Defining the purpose, scope, and tone of the piece.
Knowledge integration: Drawing on personal expertise and external sources to shape the narrative.
Critical judgment: Evaluating the AI’s output for accuracy, relevance, and coherence.
Creative framing: Incorporating fresh ideas, such as Bronowski’s hypothetical reactions to AI, to expand the depth of the discussion.
The human role, in this case, was akin to that of a director or editor, orchestrating the process and ensuring the final product reflected a blend of human insight and AI capabilities. Without our detailed input and judgement the piece would arguably lack the depth and precision it ultimately achieved.
The question of whether a piece was ‘AI-generated’ or ‘human-generated’ needs a more nuanced answer than presenting word count metrics. While ChatGPT provided the majority of the text in this case, the final product was the result of a sophisticated human-AI collaboration. Publishers should consider how much human expertise was involved in guiding and refining AI’s output, rather than simply quantifying word counts.
For educators, this raises questions about how to assess student work in an AI-enabled world. Students using AI are not simply handing over the reins; they are often shaping, structuring, and iterating the AI’s output. Evaluating the quality of their contributions—such as their ability to guide AI effectively, integrate knowledge, and think critically—may become as important as assessing the final written product.
In summary, the collaboration between humans and AI is far more complex than a simple percentage breakdown. In the creation of this piece on Jacob Bronowski, AI acted as an immensely powerful tool, but one that required direction, judgment, and knowledge to produce something meaningful. The result is not purely the product of machine intelligence, nor purely of human effort, but a synthesis of both—a modern testament to the power of interdisciplinary collaboration that we hope Bronowski himself would likely have admired.