TORCH Talks convenes debates and discussions on some of the most urgent and topical issues of our time from a cross-disciplinary perspective. We invite experts and practitioners from diverse disciplines and fields to share their insights on major global challenges and advances in knowledge. This edition of TORCH Talks is the first to feature on TORCH’s new podcast channel on Spotify and is now available for streaming on this link: Spotify Link
On 5 March 2025, TORCH hosted its second TORCH Talk of the year on the thought-provoking topic of AI and Human Creativity. TORCH Director Christine Gerrard moderated a panel discussion with three distinguished academics from the University of Oxford whose research proactively engages with AI: Professor Matthew Reynolds (English Faculty), Professor Kathryn Eccles (Oxford Internet Institute), and Dr Robert Laidlow (Music Faculty). Each brought their unique perspective to the discussion on the evolving role of AI in relation to defining human creativity, drawing from their respective fields of literature, cultural heritage, and music.
As large language models (LLMs) continue to transform educational and creative landscapes, the panel discussed how rapidly evolving developments in AI are reshaping artistic and intellectual expression. The discussion focused on issues of agency. To what extent is AI a threat and rival to human creativity? Or is AI best seen as a ‘tool’ used by humans to advance human creativity, a further technological advance in the age-old dialogue between the human, the machine and the creative impulse. The discussion framed these complex questions in ways that enabled each panellist to offer their unique perspective.
AI and its Impact on ‘Creative’ Work
Professor Gerrard reminded the audience that student use of AI is accelerating and omnipresent. According to recent studies, over 85% of students in the United Kingdom have used AI for their academic work. She asked the panel how AI has impacted their own work. LLM-based platforms like DALL·E and ChatGPT generate impressive content, yet their outputs often feel impersonal, stale, or uninspiring. If the panel use AI, could they define how or if it differs from the human ‘voice’?
Dr Robert Laidlow, a trained composer, compared the reception of AI with the reception of the saxophone at the point of its invention in the 1840s. Initially greeted with scepticism and disdain, the saxophone’s versatility and range soon earned it a place within the canon of respected musical instruments. Laidlow has positively embraced AI as part of his creative process rather than viewing it with suspicion and disdain. His work has thus evolved to explore the non-humanness of AI in music, celebrating it, and incorporating it into his work, navigating the boundaries of composition and musical instrumentation.
Professor Eccles’ work explores how AI curates knowledge and acts as a cultural intermediary. She shared an experiment she conducted with her student, Laura Herman: an exhibition showcasing and comparing AI-generated and human-curated images. One of the most interesting findings was that when surveyed, exhibition attendees expressed relief when they realised they preferred human curation over AI’s selections.
The discussion then turned to a critical debate over the AI ‘black box’, the enigma surrounding AI decision-making. There is a popular assumption that there are elements within AI models that are essentially opaque and unknowable, but the panel pushed back against this assumption. These systems are, in reality, an amalgamation of coded algorithms trained on specific sets of data. Making their processes more transparent is essential. However, not all stakeholders want this transparency—revealing how these models operate might expose uncomfortable truths about biases, data use, and the mechanics of AI decision-making deployed by big companies.
Distinguishing Between AI Creativity and Human Creativity
The discussion turned to whether we can divorce AI-generated creativity from human creativity. Can we genuinely disentangle AI from human influence? Is there any creativity within the algorithms themselves?
Professor Reynolds, who contends with this question in his academic work through the lens of translation, noted that standardised language has historically been a product of colonial regulation, impinging on linguistic rigidity. When using LLMs for translation, he found that these models revealed movement through language rather than striving for rigid accuracy, offering unexpected but thought-provoking results. He posited that incorporating AI in translation processes challenges traditional notions of linguistic equivalence, and has resulted in ‘creative’ outcomes that are at times very strange, but very intriguing.
Dr Robert Laidlow shared insights from a Barbican show he previously organised, where AI was used to generate and ‘fill in’ missing sections of Bach’s French Suite compositions. This exhibition highlighted the complexities of using AI in artistic interpretation—when does AI authentically enhance creativity, and when does it distort historical and cultural contexts? Who decides what kinds of creative outcomes gain legitimacy?
The discussion left the audience with a compelling thought: AI is an amalgamation of knowledge that it is trained on—but isn’t human experience, at its core, a similar exercise?
The session concluded with Dr Laidlow demonstrating his AI-trained instrument, the Stacco, which he will feature in his upcoming performance, The Techno-Utopia. You can hear a snippet of his performance on Spotify.
This session was recorded as the first episode on TORCH’s new podcast channel on Spotify. Be sure to check it out and participate in the poll, which will remain live for two more weeks.
