AI at AM
Exploring our position on AI
AM has a long history of harnessing technological innovation to transform historical content. It's at the heart of our mission.
Our focus is on technology which can directly benefit historical primary sources and their impact for new generations of learners and scholars.
Provenance, discovery and critical thinking
AM was the first primary source publisher to bring handwritten text recognition (HTR) to our digital resources, relying on early neural network technology to enable transcription of handwritten script. Whilst generative AI is not a single, holistic solution, and the protection of provenance will always be vital for our custodianship of historical source material, we strongly support transformative technologies which privilege critical thinking and deepen discoverability of unique source content.
Most of the students in my Digital Humanities courses at UCLA did not come from the Arts & Humanities; they came from Engineering, STEM, Economics and other Social Sciences, and they learned text analysis by applying it to historical primary sources in my computational qualitative analytics classes.
Our guiding principles
Provenance and transparency
The protection of intellectual property and data is paramount to our ethical editorial approach, as is our trusted relationship with the world’s leading libraries and archives. Our role as an academic publisher is to protect intellectual property rights for content within our resources as well as asserting rights for any onward use of content. We do not allow content shared with Large Language Model (LLM) platforms to be used for training or any commercial purposes without express permission, neither do we allow for data retention of content on LLMs. As we continue to explore the opportunities for AI to enrich discoverability in our resources, our clean user interface will favour transparency about how and where any AI functionality may be enabled so that our customers, users and partners have insight and clarity.
Human oversight and accountability
We support a conscious, human-centred approach to AI use. Any use of specific AI-led technologies within our resources is closely overseen by our specialist in-house teams. For instance, we are exploring the enrichment of metadata at scale but this work needs a clear, ongoing focus on standards alignment and accuracy. Any commissioned pilots are fully tailored to the specific needs of the source material and to privilege the content at its centre.
Critical thinking and scholarly integrity
AM avoids using AI to generate secondary content within our resources. We prioritise authorship in editorial work and aim to promote deep engagement with primary sources. Our focus is on discoverability and depth instead of summarised chunks of interpretive content. This protects academic integrity by valuing and directing users to original sources over generative content. As an academic publisher, we consider that we have a responsibility to support AI literacy, encouraging interdisciplinary approaches in historical research and study.
AI integration examples
Focusing on our key priority of enhancing discoverability, the following examples show AI integration in practice at AM so far:
Transforming metadata
Metadata is a key part of the process of building new digital resources. We inherit metadata as part of our agreements with libraries and enrich this, in doing so creating standardized and interoperable records. Our in-house Metadata and Discovery team applies technical expertise and consistency between collections.
Focusing on human oversight and expertise, we are piloting new workflows using AI-led technologies to expand metadata terms at scale. For example, one pilot focuses on data modelling and automatic data extraction to create new workflows and enhance metadata for twentieth-century diplomatic documents.
Improving transcript accuracy
Transcripts enable rich discovery of primary sources, allowing for greater surfacing of hidden voices within historical archives and drawing out valuable connections between sources. We also work with linguists who use LLMs to support metadata translation for non-English language sources.
Improving transcript accuracy is vital for fuelling scholarship but also ongoing accessibility improvements. Our in-house teams work with prototypes using LLMs to improve transcript generation, accuracy, and structure recognition for real gains in quality and accuracy. This is a groundbreaking area where fast-moving technologies will transform researchers' and learners’ ability to interact with primary sources in future.
Your feedback is important
We welcome your questions on AI use within and when using our resources. Please get in touch to share your feedback.
Frequently asked questions
AM uses AI-led technologies in the creation and production of our digital resources in specific ways as described and referred to above, including metadata enhancement and transcription improvement. Our platform technology continues to evolve and, in future, AI will be integrated into aspects of our platform functionality and engagement with digital assets. Full transparency will be provided as to where such AI-led technology is enabled in order to transform search and discoverability.
AM does not allow the ingestion of AM data into LLMs for either training or data retention purposes - however, we welcome research initiatives and digital humanities projects that incorporate LLMs for non-commercial purposes.
Yes, our licence agreements allow for customers and users to understand text data mining for research and non-commercial purposes. Use this link to submit a text and data mining request. We would love to hear about any research undertaken with AM content, so please share details of any projects underway.
Yes, we have implemented protection across our sites to ensure web crawlers are prevented from accessing secure content.
There is no use of customer data to train AI models.
AM Quartex customers can implement optional services such as OCR, HTR, and AV transcription on assets ingested into the AM Quartex platform.
This page and our technology area will be used to provide high level updates. More specific information will be provided to existing customers through the Quartex Community Platform.