AI is transforming how we research, write, think, and communicate. But can it really replace the human expertise and experience behind archaeological and historical writing and interpretation?
The short answer is no.
While AI can mimic structure and surface-level knowledge, it cannot provide vital nuance, context, ethics, and genuine and innovative insight and creativity.
Pirated Knowledge; Fabricated Sources
AI tools depend on vast datasets, but those datasets are increasingly built on unauthorised content. A recent exposé by The Atlantic revealed that Meta has used pirated books and articles to train its AI models, often without permission . A quick search unearthed several of my own publications in their system, with the ominous note: ‘there may be more.’
These models often present knowledge as original and uncited—or worse, invent sources entirely. This strips work of context and undermines the intellectual ecosystem of citation, collaboration, and debate. In history and archaeology, where interpretation is layered and evidence-based, this is more than unethical and unprofessional: it’s dangerous.
Emerging conversations around intellectual property, governance, and data curation highlight the tension between open access and responsible ownership. As discussed in DigVentures’ recent webinar the challenge lies in creating policies that protect intellectual property and empower individuals to own their narratives.
When Poor Data Meets Powerful Technology
AI is only as good as its data. And increasingly, that data is incomplete, biased, outdated, or wrong. The consequences are well-documented: Microsoft’s chatbot, Tay, turned toxic from bad training data; Amazon scrapped a hiring tool that penalised women.
The same risks apply to research. As Times Higher Education reported (May 12, 2025), a flood of low-quality, AI-assisted research is now overwhelming journals and peer reviewers, making it harder for authentic, original work to rise to the surface. A system built on speed and scale, rather than accuracy and integrity, threatens the foundations of scholarly and commercial written reports and publishing.
New findings from a Massachusetts Institute of Technology (MIT) study highlight these concerns. Researchers found that a heavy reliance on ChatGPT weakens creativity, memory, and cognitive engagement. AI users underperformed in originality and recall, and their essays were more formulaic and rated lower in quality. Participants even experienced lingering cognitive fatigue (see here). These results suggest a growing need to ‘think first, then use AI’ – as a tool for enhancement, not substitution.
Why Swallowtail Works Differently
At Swallowtail, we don’t outsource thinking to machines. Our work – led by archaeologist and editor Rachel Swallow – is grounded in experience, ethics, and an understanding that ensure that Words Work – authentically and creatively.
AI may be time-saving, but Swallowtail Archaeology’s services are directional, precise, and purposeful. Our arrowhead-inspired ethos ensures:
- Contextual understanding drawn from offline – and so, off-grid – sources that AI cannot access
- Expertly composed communications that consider the wider significance of the subject
And while we may use AI selectively (for example, to streamline formatting or check consistency), it never replaces the real work of truly original and innovative research, writing and dissemination.
Swallowtail Archaeology also acknowledges the broader conversation: while AI may democratise access to knowledge and increase public engagement, it cannot substitute wisdom, empathy, or lived experience. As the DigVentures session noted, the power lies in owning your own narrative and embedding understanding at the heart of your work.
No AI Can Lead a Walk, Teach a Talk, or Read a Ruin
Swallowtail doesn’t stop at what is written on the page. Our walks, talks, lectures, and courses – online and in person – are deeply engaging, human experiences shaped by curiosity, expertise, and place. AI can imitate language, but it can’t stand in a field, read the landscape, or bring history to life.
So yes, AI is just one tool in the kit – useful for research and writing when handled wisely by the experienced and well-informed. But relying on that tool alone risks eroding the authenticity and creativity at the heart of research and content creation in History and Archaeology. Without the expertise to recognise and challenge red flags—fabricated sources, biased data, decontextualised claims—AI can mislead more than it supports. It is only when used alongside the tools of human-led, ethically grounded, and unapologetically original research that AI can play a meaningful and trustworthy role.
Have you spotted the deliberate irony in this blog? The image has been created (beautifully, don’t you think?!) by AI.
0 Comments