Charting the Transformation of Enterprise Information

Management: AI Explainability and Transparency in EIM Practice

Lufan Zhang

and Paul Scifleet

Swinburne University of Technology, Melbourne, Victoria, Australia

Keywords: Enterprise Information Management, Explainable AI, AI Transparency.

Abstract: Today’s data-intensive environment poses significant challenges for enterprises in managing their vital

information assets that often exceed manual capabilities. Despite a promising potential to assist, there’s

mistrust and misunderstanding of the values AI presents to Enterprise Information Management. This paper

investigates the current state of AI-led changes to EIM practices and proposes an approach to improve

understanding of AI’s transformative role and impact on EIM. By charting AI use in EIM platforms across

five areas - AI development, AI techniques, AI-integrated EIM capabilities, AI applications, and AI impacts

– along with practice-based criteria for evaluating AI-integrated EIM solutions, this paper lays the foundation

for explainable and transparent AI in EIM.

1 INTRODUCTION

In the current massive data environment enterprises

face substantial challenges in managing their most

vital information resources. The last decade has

generated more data, documents and records than any

previous decade of human activity, however most

information resources remain predominantly

unstructured and poorly controlled (Kolandaisamy et

al., 2024), making them less reliable, retrievable, and

accessible than ever before (Jaillant, 2022). The

overwhelming volume of information is exceeding

in-house expertise and the manual or semi-automated

approaches that most enterprises usually take to

implement architectures for information control.

Consequently, it is not surprising to see increasing

attention being paid to applying Artificial

Intelligence (AI) and Machine Learning (ML) based

solutions in Enterprise Information Management

(EIM) (Baviskar et al., 2021) including for, the

classification of digital assets (Huddart, 2022),

authoritative records control, taxonomy and metadata

management (Duranti et al., 2022), and screening for

sensitive and confidential information communicated

via email (Schneider et al., 2019).

https://orcid.org/0009-0004-5148-564X

https://orcid.org/0000-0003-2776-1742

Despite the promising potential of AI-based

approaches to enterprises’ IM needs, current evidence

indicates that the inherent complexity and opacity of

AI cause mistrust and misunderstanding among EIM

practitioners about the transformational opportunities

and value AI presents (Adadi & Berrada, 2018).

While explainable AI (XAI) research initiatives aim

to improve the transparency and understandability of

complex AI solutions for end users, these approaches

are primarily algorithm-centric and highly technical,

often falling short in adequately addressing the needs

of non-expert users (Barredo Arrieta et al., 2020). In

contrast to the AI experts, programmers, and data

analysts, who typically interact with AI at algorithm

and model design levels (Bunn, 2020; Langer et al.,

2021), EIM practitioners do not need to understand

AI algorithm functions and the reasons behind the

generation of specific outcomes. Their first

experience of AI is often through interface

interactions. This might involve experimenting with

publicly available tools like ChatGPT or investigating

the use of AI product integration in other workplace

tools. EIM professionals are more concerned with the

practical applications and utility of AI across the

information management lifecycle (Haresamudram et

al., 2023). Solutions based on an explainable AI in the

context of EIM are required to meet the practical

Zhang, L. and Sciﬂeet, P.

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice.

DOI: 10.5220/0012951100003838

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 16th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2024) - Volume 3: KMIS, pages 60-73

ISBN: 978-989-758-716-0; ISSN: 2184-3228

needs and interests of IM practitioners. To address

this need, this study investigates the current state of

AI use in EIM systems and the changes this is

bringing to information management practice, this

paper presents the findings from an environmental

scan of AI integration into EIM platforms, focusing

on how understandable new AI-based solutions are

for practitioners and, on the characteristics required

for explainable AI in EIM.

The research findings present current issues and

challenges in EIM as prioritised by 20 leading EIM

platform providers between August 2022 and

November 2023. Research outcomes include the

categorisation of AI use by platforms into five areas

required to support the explainability and

understandability of AI for practitioners seeking to

adopt new approaches. These include describing 1)

how AI development is taking shape, 2) what

underlying AI techniques are being used, 3) how AI

integration maps to EIM capabilities, 4) what AI

applications are available for use, and 5) how AI

impacts on EIM practices. Moreover, to evaluate the

extent to which information provided by the 20 AI-

integrated EIM platforms is clear and transparent for

practitioners, an outcome of this research is a model

for evaluating AI transparency in EIM based on six

practical criteria: 1) Provision of AI development

details 2) Provision of AI function details 3)

Provision of AI impacts (benefits & risks) 4)

Provision of real-world use cases 5) User experience

design for AI-integrated interface 6) Human-AI

interaction.

The contribution of the research is twofold.

Firstly, it adds to knowledge in EIM by uncovering

the role and impact of AI in EIM practices, with five

practice-based categories to improve the description

and understanding of AI integration in EIM

recommended. This offers a practical contribution for

EIM practitioners seeking to leverage AI in their

work, and is supported by a further six criteria for AI

transparency, developed as an outcome of this

research, that both vendors and practitioners can work

towards achieving. Both contribute to the field of

Explainable AI by addressing the needs of non-

experts seeking to work with AI and, through this,

promoting human agency in explainable AI (XAI).

The following sections of the paper discuss key

topics in related research, followed by an elaboration

of the research design and findings. The paper then

concludes with a discussion of the current state of AI-

led changes in EIM practices and reflects on how AI-

integrated EIM practices can be facilitated.

2 RELATED WORK

2.1 EIM Issues and Challenges

Enterprise Information Management is an

overarching concept that encompasses a range of

related information systems and information

management work practices including Enterprise

Content Management (ECM), Electronic Records

Management (ERM), Document Management (DM),

and Knowledge Management (KM) (AIIM, 2024).

Notably, these terms are often used interchangeably

in industry (Scifleet et al., 2023). EIM can be broadly

defined as the integrated, enterprise-wide, strategic

management of all types (physical, digital, differing

sources and formats) of enterprise information assets

over their entire lifecycle of business use

(Jaakonmäki et al., 2018; Williams et al., 2014). The

term information asset encompasses all data,

information, documents, and records required for the

everyday work practices of business (Scifleet et al.

2023). Hausmann et al.’s research on enterprise

information readiness further summarises EIM as a

comprehensive initiative for managing information

assets throughout the entire lifecycle to unlock value,

with a focus on ensuring regulatory compliance. A

key goal is to eliminate information silos across

business departments and areas of work, ensuring the

availability of well-structured information when

needed (Hausmann et al., 2014).

Notably, many of the EIM issues identified for

business in an industry survey by Hausmann and

Williams et al. in 2014 remain relevant (Hausmann et

al., 2014; Williams et al., 2014), if not more

pronounced following a survey conducted a decade

later (Scifleet et al. 2023). Hausmann et al., (2014)

identified the challenges for practitioners in

managing an increased volume and variety of

business information and prioritised compliance and

assurance as central with new technologies and new

types of data, such as social media impacting

businesses. Both the 2014 and 2023 survey results

revealed that while enterprises self-rated highly in

achieving conformance goals, they continued to

struggle when working with information to achieve

performance objectives requiring timely access to

critical information, sharing information (both

internally and externally), managing the information

lifecycle, deriving value from, and delivering

actionable business intelligence (Hausmann et al.,

2014; Scifleet et al., 2023; Williams et al., 2014).

Additionally, the 2023 survey highlights the

compounding and negative effect of an

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

overwhelming growth in employee-created data

through an increasing number of applications.

In this research, we have built from the many

well-known challenges presented by Hausmann et al

(2014) and others to revisit priorities current in

practice today: with a focus on the role that platform

providers can play by providing advanced,

transparent, understandable and usable AI-based

solutions for acquiring, organizing, storing,

retrieving, and sharing information assets within an

organization. EIM systems are beginning to integrate

AI across areas of information management that are

traditionally labour-intensive and hard to achieve, e.g.

taxonomy development, classification, process

automation, search and findability (Duranti et al.,

2022). Despite the promising features that AI can

bring to the EIM space, we still lack a holistic

understanding of how AI is positioned in EIM

practice. Understanding the steps that are being

undertaken to achieve AI integration and transform

EIM practice is a critical area for research.

2.2 AI and ML in the EIM Context

Within the EIM field, we have found the term

Artificial Intelligence (AI) serving as an overarching

concept encapsulating many aspects of cognitive

computing and advanced programming aimed at

performing tasks that typically require human

intelligence, though in most cases it is being used to

describe narrow, task-specific uses of AI (Meske et

al., 2022).

As a subset of AI, Machine Learning (ML)

specifically focuses on developing algorithms and

models that enable computers to learn from data,

allowing them to undertake information processing to

deliver outputs and make predictions, or decisions,

without the need for additional explicit human

programming or human intervention (Barredo Arrieta

et al., 2020). Martens and Provost (2014) have

demonstrated how ML can be applied to document

classification tasks by selecting a minimal set of

words to successfully classify document features for

management purposes. In a study by Ragano et al.

(2022), semi-supervised ML models were used to

evaluate audio quality in digital sound archives,

demonstrating potential benefits for other

multimedia, such as video calls and streaming

services. Sambetbayeva et al. (2022) highlight the

document management challenges in the context of

the data deluge and propose the use of ML techniques

to enhance document retrieval. The potential of AI

tools for managing digital records has gained

widespread recognition in records management with

Duranti et al’ (2022) and others highlighting that the

requirements for collecting and indexing digital

records in a reproducible manner far exceed manual

capabilities (Duranti et al., 2022; Schneider et al.,

2019). AI technologies, such as Recurrent Neural

Networks (RNN) (Shabou et al., 2020), Handwritten

Text Recognition (HTR) (Goudarouli et al., 2019),

and Chatbots (Gupta & Kapoor, 2020) are all being

proposed as means for reducing labour intensive

work, increasing efficiency and effectiveness with

examples of their role in facilitating record

classifications, access to paper-based archival

information, and establishing new knowledge

available. What is at issue, is just how understandable

and useful these technologies are for IM practitioners

in everyday work, where they are tasked with

explaining application, use, and value to the

enterprise?

2.3 Explainable AI (XAI) and AI

Transparency

The practical implementation of AI and ML in real-

world business settings is often met with scepticism

by industry professionals (Modiba, 2023).

Predominately this scepticism stems from

perceptions of AI as an uninterpretable “black box”,

with significant concerns about transparency,

trustworthiness, and a need to improve understanding

of how AI systems produce their outcomes (Adadi &

Berrada, 2018). Consequently, there is a growing

prioritisation for Explainable AI (XAI), with an

overarching goal of improving the accessibility of

intelligent systems (Meske et al., 2022). The concept

of XAI demands a better explanation about how AI-

generated outcomes are achieved. This has resulted in

substantial progress within the AI community, where

XAI is seen as a sub-field of AI (Adadi & Berrada,

2018; Barredo Arrieta et al., 2020) that aims to

provide users with the ability to see inside the black-

box. This is typically facilitated through the lens of

another algorithm that has the role of describing the

logic and decision-making processes of the AI and

generating a report that confirms its operations.

However, in turning to computational solutions to

read an algorithm and and report on the veracity of

black-boxed behaviours so that the AI can be trusted,

for example in a legal claim, we are at risk of using

one opaque method to describe another with little

gain for practitioners (Adadi & Berrada, 2018). The

Human-Computer Interaction (HCI) community has

approached this differently, by bringing a human-

centred perspective to XAI that emphasises the

significant role of humans in the explanatory process,

arguing the importance of being able to engage

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

different user-stakeholder groups and their needs in

AI development and use (Meske et al., 2022).

HCI research has focussed on understanding

users’ perceptions of XAI methods, with research

evaluating different HCI-XAI methods for their

explanatory capabilities: having the right method in

place to explain AI to a user can, in turn, contribute

to better more understandable AI systems design

(Wang & Yin, 2021; Wanner et al., 2022). Even so,

many approaches to XAI still remain technical and

often struggle to be translated into practical

implementations when it comes to assisting non-

expert users in making decisions about AI in work-

based contexts (De et al., 2020). The need to improve

explainability for non-experts from a practice

perspective remains (Brennen, 2020; Bunn, 2020).

EIM practitioners’ engagement with AI will be

through those tools in EIM platforms that they use to

meet their daily information management needs. AI

will be assessed on the practical benefits gained.

Instead of delving into detailed micro-level

explanations of AI models, there is a need to link the

understandability of AI to EIM practices, with an

emphasis on being able to identify and explain

system-level applications in AI-integrated EIM

transparently, in as open and clear way as possible for

practitioners.

AI transparency has been identified as a key

requirement for AI technologies by the AI HLEG (the

European Commission’s High-Level Expert Group

on Artificial Intelligence), and the transparency of

data processing in AI applications is mandated by the

GDPR (General Data Protection Regulation)

(Felzmann et al., 2019). Broadly referring to the

principle of making AI systems understandable,

explainable, and accountable, AI transparency

concerns disclosing information about an AI system,

typically to support judgments regarding fairness,

trustworthiness, safety, efficacy, accountability, and

compliance with regulatory and legislative

frameworks (Andrada et al., 2023). The problem with

AI’s “black box” is a clear lack of transparency;

however, the concept of AI transparency itself is often

opaque (Kiseleva et al., 2022). AI transparency

research primarily targets algorithmic transparency,

aiming to provide visibility into the underlying

algorithms and neural networks to help rationalize the

outcomes produced by complex programming

(Andrada et al., 2023). However, algorithmic

transparency alone does not address the needs of

different AI stakeholder groups and thus fails to make

AI systems more understandable to non-experts

(Felzmann et al., 2019; Haresamudram et al., 2023).

To clarify different types of AI transparency and what

greater transparency might entail, Andrada et al.

(2022) offer a taxonomy that includes two main types

of AI transparency: reflective transparency and

transparency-in-use. Reflective transparency

encompasses information transparency, material

transparency, and transformation transparency.

Transparency-in-use focuses on ensuring the

interface itself is intuitive, allowing users to

understand and navigate systems to complete their

tasks. Similarly, Haresamudram et al. (2023) have

proposed three levels of transparency relevant to

diverse stakeholder groups and contexts: 1)

algorithmic transparency, 2) interaction transparency,

and 3) social transparency, however the categories

remain broad. Despite a better understanding of the

different aspects of AI transparency for various

stakeholder groups, research on how AI transparency

translates into applied settings is limited

(Haresamudram et al., 2023). This study addresses

the gap by operationalizing AI transparency with the

EIM context in mind, providing a set of six practical

evaluation criteria for AI transparency.

3 RESEARCH METHODOLOGY

AND DESIGN

3.1 Research Objectives

The integration of AI into EIM platforms will

transform information management practice not

simply as a by-product of smarter off-the-shelf

automation, but because EIM practitioners will adapt

to the use of AI by altering the conventions and

practical knowledge of everyday work in EIM, and by

contributing to the design of AI solutions specific to

EIM (De Certeau & Mayol, 1998) (Fensham et al.,

2020). This research then, contributes to an improved

understanding of the transformative role of AI and its

impact by investigating the current state of AI-led

changes to practice and presents the foundations for a

practice-based descriptive framework for explainable

and transparent AI in EIM. The approach taken is

based on a sociotechnical and practice theory

perspective (Orlikowski & Scott, 2016), holding the

viewpoint that both people (the practitioners) and

technologies (the platforms) have the agency to

influence and shape each other, and addresses the

following research objectives (RO) and questions

(RQ):

RO1 – To understand current EIM challenges

faced by practitioners in the massive data

environment.

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

 RQ1(a) What current EIM issues are

faced by practitioners?

 RQ1(b) How are the issues in EIM

practice being presented by platform

vendors?

RO2 – To describe how AI is being brought into

EIM practice by EIM platforms.

 RQ2(a) How is AI for EIM being

developed and integrated into platforms?

 RQ2(b) What are the impacts, benefits,

and advantages that AI is having on the

delivery of EIM services?

RO3 – To devise a working definition of AI

explainability and transparency for AI-integrated

EIM practices and evaluate the understandability of

AI-led changes in EIM platforms.

 RQ3(a) What does AI explainability and

transparency mean in applied EIM

contexts?

 RQ3(b) How transparent is the

information provided by the platforms

regarding AI applications?

3.2 Research Design

This study’s approach is qualitative and based on the

environmental scanning (ES) of publicly available

Web resources to establish awareness of products,

services and strategies constituting the relatively new

and emergent delivery of AI in EIM platforms. ES,

which has its roots in business analysis, is a method

applied by researchers to gather and analyse

information concerning the domain of interest from

publicly available resources to establish situational

awareness of the environment and plan actions

accordingly (Auster & Choo, 1994; Zhang et al.,

2011). While initially applied by businesses for

strategic purposes there has been a shift in the use of

ES in recent years to academic research, where the

identification and analysis of current, publicly

available information resources is critical to research

domain awareness (Lau et al., 2020; Yin et al., 2021).

As the AI-based changes that are taking place in EIM

are arriving from vendors integrating AI into their

platforms as products for clients, scanning publicly

available information from vendors’ websites

provides this study with a method appropriate for

establishing domain awareness and a starting point

for understanding the changes that AI integration in

EIM brings.

Data collection for the environmental scan was

undertaken in two stages between August 2022 and

October 2023, following a series of steps outlined for

qualitative media analysis (Altheide & Schneider,

2013), depicted in Figure 1.

Figure 1: Steps in Environmental Scanning.

Stage one resulted in data collection from 28

leading EIM platforms between August - September

2022, with stage two following in August- October

2023 concentrating on a subset of 20 EIM platforms

that were identified from the first stage, because of

their more focused discussion of AI integration.

While not initially planned for, this allowed the

research to map a significant industry change

corresponding to the public release of ChatGPT and

other OpenAI initiatives from November 2022

(OpenAI, 2022), with a burst of discussion about AI

and AI impacts taking place in EIM following

ChatGPT hype.

Steps 1 & 2 Scan Questions and Search Strategy

Starting an environmental scan involves initiating

a search strategy that is based on the research

questions. To locate relevant vendors, platforms and

service providers, we applied a search strategy

comprising broad terms, main terms and related

terms, relevant to the focus. Undertaking a Google

search with the broadest concepts first, including

information management, information management

services and information management service

providers. The same approach was taken for closely

related service areas of, content management,

document management, records management and

knowledge management.

Following initial search results, information was

collected from 140 pages, with the scan identifying

more than 70 EM companies including 42 service

providers and 28 platform providers. We consider

EIM platform providers to be companies that design,

develop, and provide integrated technology platform

solutions (all software, database, network and cloud

components) e.g., OpenText, Oracle NetSuite,

Objective, Hyland, Microsoft M365, and EIM service

providers as companies who focus on the provision of

information management consultancy services, e.g.

Access, Astral, TIMG, Cube Records Management,

Document Logistix are examples. While EIM service

providers also undertake software development to

further customise major platforms for clients, they are

not platform developers. For the purposes of this

study, our starting point to examining changes in

practice is the arrival of AI in EIM platforms.

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

We next focussed on applying inclusion and

exclusion criteria to refine the list of EIM platform

providers to a list of platforms with Web resources

describing explicitly, the incorporation of AI into

their offerings. As a result, a pool of 20 AI-integrated

EIM vendor platforms was selected for further data

collection and analysis

Steps 3 & 4 Data Collection, Analysis and

Reporting

The unit of analysis for the study comprised three

main components of information typically available

at each of EIM platforms’ websites: 1) platform

overviews, 2) AI feature descriptions, and 3) real-

world AI-based case studies. The presence of real-

world AI-based case studies showcased by the

vendors serves as a significant indicator of AI

transparency, by describing real EIM use cases for

AI.

A “three-level” data collection strategy was

implemented for collecting data from the vendors.

Navigating “three-levels deep” from the platform

overview page ensured that data collection was

independent from the platform’s website homepage

and architecture and helped to locate the same type of

information consistently. The “three-levels” are

defined as: Level 1, providing an overview of the

EIM platform; Level 2, offering a general insight into

AI features; and Level 3, providing specific details

about AI features. A data collection template was

used to ensure the same details were collected for

each platform including, platform provider name,

platform name, collection date, URLs, platform

overview, AI feature descriptions, and the presence of

real-world AI-based case studies. The raw data

collected for each vendor was initially saved to a

Word document using the template and then imported

into NVivo for further analysis. Excel has been used

for supporting analysis and to assist in the visual

presentation of the findings.

Thematic content analysis was used to analyse the

collected data, following an inductive, ground-up

approach in NVivo while acknowledging that the

starting questions and topics have framed the analysis

(Williamson & Johanson, 2018). The coding process

commenced without preconceived categories or

themes about AI integration in EIM; instead, the AI-

specific categories presented emerged from the

research findings during coding. This was supported

by sense-making and fact-checking aligned with the

current state of cognitive computing, AI, and ML. To

ensure reliability, the research team regularly

discussed and refined the coding, with inter-coder

reliability checks conducted to reach consensus on

themes, topics, and categories, leading to a rigorous

process of topic reduction and confirmation.

4 RESULTS

4.1 EIM Issues and Challenges

The thematic content analysis identified eight EIM

issues for practitioners as described by the 20 EIM

platforms: immature digitalisation, information

security, privacy, and compliance (ISPC) risk,

information silos, poor information findability,

information overload, poor information sharing, poor

information utilisation, and information quality

concerns (Figure 2).

Figure 2: EIM Issues as prioritised by EIM Platforms.

Immature digitalisation (N=6, 30%) refers to

lacking sufficient digitalising capabilities to leverage

digital technologies for the automation of manual

‘paper-driven’ processes. Although OCR techniques

have been widely utilised for converting documents

into digital forms, full automation remains a critical

challenge for many enterprises in managing their

information assets.

Information security, privacy, and compliance

(ISCP) risk (N=5, 25%) refers to the challenges faced

by enterprises in meeting information security,

privacy and regulatory compliance over their

information assets. While flexible and improved

collaboration enables better productivity,

unauthorised access to information and lack of proper

control continue to cause data breaches. Enterprises

also raise concerns about identifying and protecting

their customers’ sensitive and personally identifiable

information (PII).

Information silos (N=5, 25%) and poor

information findability (N=5, 25%) are closely linked

issues that significantly impact information sharing

(N=3, 15%). Information silos occur when data is

stored in multiple or geographically dispersed

locations, with fragmentation resulting from

information being spread across different systems,

tools, siloed repositories, and disconnected end-line-

of-business applications. Siloed information leads to

poor information findability, making it difficult for

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

employees to access information when needed. This

impacts productivity and hinders both internal and

external information sharing, resulting in lower

productivity and higher information risks.

Information overload is another critical issue

identified (N=4, 20%), that arises from the

overwhelming scale of information coming from

different sources and channels, and is closely linked,

in platform discussion, with poor information

utilisation (N=3, 15%). This includes challenges in

knowledge discovery and obtaining data-driven

business insights.

While not prioritised by EIM platforms (N=1,

5%), information quality concerns must be seen as

remaining critical. Enterprises raise concerns about

the quality of their information and the

trustworthiness of decisions made based on this

(Scifleet, 2023). The accuracy and reliability of

analytical insights always depend heavily on data

quality, and the success of AI projects in enterprises

will rely heavily on the quality of the datasets that AI

models are trained with.

4.2 AI Explainability in EIM Platforms

Table 1 categorises this study’s findings across five

key areas for explainability (XAI) that practitioners

can consider when seeking to understand and evaluate

AI use in EIM platforms: AI Development, AI

Techniques, AI-integrated EIM Capabilities, AI

Applications in EIM, and AI Impacts. These areas

support the explainability and understandability of AI

applications for EIM practitioners.

Table 1: Five categories of AI use in EIM.

Category

Definition

Development

Refers to the way AI is developed

inhouse or adopted by an EIM

platform, to develop specific AI

solutions and AI model training.

Techniques

Refers to types of approaches (e.g.

computer vision, generative AI, deep

learning) employed in EIM platforms’

AI offerings.

AI-integrated

EIM

Capabilities

Refers to the EIM capabilities for

managing enterprise information assets

through AI integration, e.g. AI-

powered information capture.

Applications

in EIM

Refers to the underlying AI

applications that support EIM

capabilities e.g. Automated data

classification, Automated workflows.

AI Impacts

Refers to benefits and advantages

identified for integrating AI into EIM

practices across the EIM lifecycle.

4.2.1 AI Development

We consider two sub-categories important as part of

AI Development: AI solutions and AI training and

deployment. AI solutions refer to the integration of an

AI capability directly into an EIM platform either as

native AI solutions (developed in-house) or by

including third-party AI solutions. Among the 20

platforms analysed, 12 are explicit about how they are

developing AI, while 8 are not. That 40% do not share

this level of detail remains a significant explainability

concern for practitioners. As shown in Figure 3, the

majority of the 12 platforms take a Native-AI

approach (N=9, 75%) to development. Others adopt a

third-party AI solution (N=3, 25%), working with

well-known AI service providers, including Clarifai,

Microsoft Cognitive Services and OpenAI, to offer

AI capabilities to their clients.

Figure 3: AI Development – AI Solutions.

Training and deployment encompass key aspects

of AI model development and use, including data use

with AI, human-AI interaction, pre-built AI models,

customisation capability, model performance

improvement, explainable features, and AI

limitations (Figure 4).

Figure 4: AI Development - AI Training and Deployment.

Data use with AI (N=8, 40%) involves datasets

for training AI models and data use by deployed AI

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

products. Understandability, about how AI works

with data is crucial for addressing concerns regarding

the security and sensitivity of proprietary datasets

within enterprises. While some platforms clarify that

datasets are used with enterprises’ consent for model

training purposes, concerns remain about the security

and sensitivity of data used in deployed AI products.

Human-AI interaction (N=6, 30%) represents a

key feature of AI development in the EIM platforms,

where the interaction between humans and AI can

take various forms. For example, humans can validate

AI-produced results and provide corrections or

feedback to help AI systems improve continuously.

Pre-built AI models (N=5, 25%) include pre-built

or out-of-the-box AI models that are shipped with the

platform. Without the burden of undertaking any

further development, practitioners can interact with

these pre-built AI models at the application level.

However, this does not lower the burden for

explainability as some platforms offering pre-built AI

models also provide AI customisation capabilities

(N=5, 25%), allowing integrated AI applications,

including AI models and generative AI prompts, to be

customised to specific use cases and business needs.

The options for customising AI models vary widely,

ranging from allowing practitioners to build an AI

model from scratch to simply tuning model

parameters or selecting from various models or model

versions to achieve optimised results. Model

performance improvement (N=3, 15%) includes

enabling continuous learning capabilities and

incorporating human-in-the-loop verification and

feedback based on proper business context.

Explainable features and limitations (N=3, 15%)

represent features available on EIM platforms that

indicate the accuracy and reliability of AI-generated

results. For example, platforms provide accuracy

rankings for suggested filing locations, or use

different colours to indicate certainty levels in

indexing. Despite its importance for improving users’

trust in AI-generated outputs, only a few platforms

offer this. Additionally, only one platform in our

analysis acknowledges the limitations of AI

applications, noting that AI performance relies

heavily on the quality of the training data used.

4.2.2 AI Techniques

Seven significant sub-categories of AI techniques

emerged from the content analysis: Advanced

Character Recognition (ACR), Generative AI, AI-

integrated Robotic Process Automation (RPA),

Computer Vision, Natural Language Processing

(NLP), Deep Learning, and Generic AI and ML

(Figure 5).

Figure 5: AI Techniques.

Among the seven main categories emerged from

AI techniques, it is not surprising that Advanced

Character Recognition (ACR) (N=12, 60%) appears

to be the most employed AI technique in the solutions

offered by EIM platforms. With a long history of

using OCR for information capture in EIM, ACR is

familiar to practitioners. ACR applies AI in various

character recognition technologies including Optical

Character Recognition (OCR), Intelligent Character

Recognition (ICR), and Zonal OCR. These

technologies identify and extract text from images,

scanned documents, or other visual sources,

converting it into editable, searchable text with high

accuracy and efficiency.

The second-ranked AI technique making its way

in EIM is Generative AI (N=7, 35%), including large

language models (LLMs) and generative AI-based

chatbots. Since OpenAI released ChatGPT in

November 2022, generative AI has significantly

changed the way AI tools are thought about for work

tasks and are now serving multiple purposes such as

generating text, images, and audio and videos. The

integration of generative AI chatbots in EIM

platforms is transforming EIM practices, including

search and retrieval, knowledge-based reporting and

digital asset management.

Other underlying AI techniques that are being

brought to EIM include AI-integrated Robotic

Process Automation (RPA), Computer Vision,

Natural Language Processing (NLP), and Deep

Learning. While disclosing information about these

AI techniques across the platforms provides some

transparency for practitioners, details can be dense in

terms of applying specific AI techniques across the

EIM Lifecycle to improve understandability. Adding

to this problem, we found that some platforms are

using generic AI and ML terms (N=3, 15%) without

explanation, providing no useful information to help

practitioners determine the use of these tools for

specific IM needs.

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

4.2.3 AI-Integrated EIM Capabilities

Six sub-categories of EIM capabilities that result

from AI inclusions were identified from the analysis,

we refer to these as AI powering of the capability: AI-

powered Business Process Automation (BPA), AI-

powered Information Capture, AI-powered

Information Search and Retrieval, AI-powered

Information Security, Privacy and Compliance

(ISPC), AI-powered Business Intelligence, and AI-

powered eDiscovery (Figure 6).

Figure 6: AI-integrated EIM Capabilities.

AI-powered Business Process Automation (BPA)

identified across almost all platforms (N=16, 80%),

refers to using AI to automate and streamline business

processes, e.g. automating repetitive tasks and

workflows, and aligns well with the most common

EIM issue of immature digitalization, particularly for

automating manual processes.

AI-powered Information Capture (N=16, 80%)

refers to the application of AI to automate the

collection, conversion, organisation and filing of data

from source materials. For instance, ACR is often

used in AI-powered information capture to process

large volumes of scanned paper documents, making

them readily available for downstream processing.

AI-powered Information Search and Retrieval

(N=15, 75%) involves the application of AI in

information search and retrieval processes, including

natural language and semantic search across textual,

visual, and multimedia data (text, image, video, and

audio files). The application of computer vision for

image searching is increasing and generative AI-

based chatbots are providing new interfaces for

employee search queries.

AI-powered Information Security, Privacy, and

Compliance (ISPC) (N=12, 60%) involves applying

AI to information security, compliance and

governance in EIM. This includes applying AI to

information security tasks, e.g. detecting threats, and

anomalies for data protection. Additionally, AI is

utilized for identifying and ranking sensitive and

confidential information, including Personally

Identifiable Information (PII) and proprietary

information. Importantly, AI is being applied in

information governance and compliance by

automating metadata management and retention and

disposal schedules.

AI-powered Business Intelligence (BI) (N=7,

35%) refers to utilising AI for various data analysis

and reporting tasks, including relationship analysis,

sentiment and behavioural analysis. AI techniques

like NLP are used to analyse large volumes of text,

identifying relationships across documents and

records that are not otherwise apparent.

AI-powered eDiscovery (N=4, 20%) refers to the

integration of AI technologies into the process of

identifying, preserving, collecting, reviewing, and

producing electronically stored information (ESI) for

use in legal contexts, e.g. court proceedings,

investigations, and other compliance matters. AI can

be applied to automate and enhance tasks that would

traditionally be time-consuming and labour-

intensive, such as identifying required documents

based on keywords, concepts, or patterns that occur

in a document or even predicting the relevance of

documents to a particular court case.

4.2.4 AI Applications in EIM

Closely related to AI-integrated EIM capabilities, AI

applications in EIM represent the underlying

applications of specific AI techniques that support the

high-level AI-integrated EIM capabilities. As shown

in Figure 7, eight sub-categories emerged from the

content analysis: Automated Workflows, Automated

Data Classification, Automated Content Creation,

Automated Information Recognition, AI Analytics,

Help, Assistance and Recommendation Services,

Automated Translation and Automated Security

Monitoring. Notably, a single AI-integrated EIM

capability can be supported by multiple AI

applications. For instance, AI-powered information

capture is supported by automated information

recognition, automated data classification and

automated workflows simultaneously, highlighting

the complex nature of AI-integrated EIM practices.

Automated workflows (N=18, 90%) are the most

common AI applications, where AI is leveraged to

automate various workflows, including document

workflow and document control workflow. These

applications help automate repetitive, manual tasks,

allowing employees to focus on higher-priority

activities. This aligns with the most common AI-

integrated EIM capability: AI-powered Business

Process Automation (BPA).

Automated data classification (N=16, 80%)

utilizes AI to classify data based on its content,

context, or other attributes. This includes tasks such

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

as metadata generation, auto-tagging, auto-indexing,

and document classification.

Automated content creation (N=15, 75%) uses AI

to generate content, including converting paper-based

documents into fully searchable digital documents,

automatic form creation and completion, document

summarization for reporting and knowledge creation,

transcript generation from speech to text, and alt text

generation for images.

Automated Information Recognition (N=13,

65%) uses AI to identify information or patterns, such

as passport numbers, phone numbers, or driver's

licenses from documents. This includes tasks like

data extraction and data validation.

AI Analytics (N=8, 40%) use AI to derive insights

from data and processes, including content analytics

(text analytics, image analytics, audio and video

analytics), sentiment analytics (intention analysis and

behaviour analysis), and process analytics.

Help, Assistance, and Recommendation Services

(N=7, 35%) use AI to provide users with various

recommendations. This includes suggesting similar

assets, recommending file storage options, providing

visualization suggestions and offering transformation

suggestions.

Automated Translation (N=3, 15%) uses AI to

translate text or speech from one language to another.

This capability enhances global business reach and

information sharing across different languages.

Automated Security Monitoring (N=2, 10%) uses

AI to identify and detect anomalies and threats in

content, generating timely alerts to users to protect

against data loss. This includes tasks such as anomaly

and threat detection, security alert generation, and

containing data leakage.

Figure 7: AI Applications in EIM.

4.2.5 AI Impacts

Six key categories identifying how AI integration

impacts EIM practices were found in the analysis: AI-

workplace benefits, AI-enterprise strategic, financial

and reputational benefits, AI-user experience

benefits, AI-information security, privacy and

compliance (ISPC) benefits, AI-information quality

improvements, AI-collaboration improvements, AI-

customer gains, AI-business sustainability and

continuity (Figure 8).

Figure 8: AI Impacts on EIM Practices.

Among these, AI workplace benefits (N=20,

100%) stand out prominently as advantages described

across all EIM platforms, including operational

benefits, collaboration improvements and employee

gains. This aligns with a commonly listed value

proposition for AI, that it enhances business

operations and employee performance. Associated

with this, AI-Enterprise strategic, financial and

reputational benefits (N=13, 65%) feature highly,

including strategic perspective (greater competitive

advantage, more informed decision-making, and

richer business insights for uncovering new business

opportunities), financial perspective (costsaving and

increased ROI), and reputational perspective (brand

consistency, cohesive and unified brand experience

and customer gains).

AI-user experience benefits (N=12, 60%) centre

around a user-friendly including natural language

interfaces, no-code environments, reduced

dependency on technical expertise, user autonomy,

and self-sufficiency, collectively enhancing the ease

of AI adoption for users. However, the connection

between both workplace and user benefits to AI

technology is rarely clearly presented.

AI-Information security, privacy and compliance

benefits (N=10, 50%) include enhanced information

security, data loss protection, privacy, compliance

and regulation adherence, and governance. AI-

Information Quality Improvements (N=8, 40%)

refers to the improvements regarding to all aspects of

Information quality, including integrity, accuracy of

data, completeness of data, reduced data errors, and

more.

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

4.3 AI Transparency in EIM

Based on the findings and literature, this study

proposes that a contextual working definition of AI

transparency in EIM that enables the evaluation of

transparency in AI-integrated EIM offerings is

needed. Critically we find that AI Transparency in

EIM solutions must encompass information

transparency (disclosing information relevant to EIM

practitioners) and transparency-in-use (intuitive user

interface and human in the loop) to enable a better

explainability: for understanding, trust and adoption

of AI applications for EIM practitioners, and note the

interdependency in these concepts.

4.3.1 AI Transparency Evaluation

To apply AI transparency in EIM, this study has

developed six practice-based evaluation criteria that

can be used by practitioners: information

transparency – 1) Provision of AI development

details (AI development and AI techniques), 2)

Provision of AI function details (AI-integrated EIM

Capabilities and AI applications), 3) Provision of AI

impacts (Benefits & Risks), 4)

Provision of real-

world use cases, and transparency-in-use – 5) User

experience design for AI-integrated interface, 6)

Human-AI interaction.

Table 2 evaluates the transparency of current AI-

integrated EIM solutions across the 20 platforms in

this study, based on these criteria

. In terms of

information transparency, all platforms offer insights

into how AI can support EIM capabilities and the

benefits it brings to EIM practices. However,

information regarding AI development and the

underlying AI techniques used is not fully disclosed,

with only a few platforms providing details regarding

data use with AI, including how datasets are utilised

in AI model training or operational deployments. This

lack of transparency on critical technical aspects such

as AI techniques and data handling, can raise

significant concerns for practitioners, particularly

around security, trust and AI adoption in EIM.

Notably, AI risk and proven AI success use cases

in the real world are rarely provided by platforms,

which can hinder the trust and adoption of AI-

integrated EIM offerings among practitioners.

Additionally, while most platforms highlight benefits

like intuitive, user-friendly interface designs for

practitioners to work with their AI products, there is

a lack of clarity regarding how humans can align AI

with their work practice.

Platform vendors are not identified by name in the

Table 2, details can be provided by the authors on

application.

5 CONCLUSION

5.1 Discussion of Findings

Through an environmental scan of 20 leading EIM

vendor platforms, this study identified eight

interrelated EIM workplace challenges prioritised by

the platforms. Compared to the EIM challenges faced

by enterprises a decade ago, these issues remain

unresolved, if not more problematic. With the

emergence and widespread use of generative AI,

there EIM systems are turning to AI to address these

challenges.

To understand how AI is transforming EIM

practices, this study starts by examining what is

available to practitioners, charting the role and impact

of AI across five areas: AI development, AI

techniques, AI-integrated EIM capabilities, AI

applications, and AI impacts. EIM platforms are

adopting a native-AI approach and developing AI

capabilities internally and this is likely to result in a

good fit-for-purpose. However, there is a clear lack of

information available to support the understandability

of AI’s role across the information management

lifecycle and this needs to be addressed.

Regarding AI-EIM capabilities and integration

into practice; AI-powered information capture,

search, and retrieval, supporting consistent

information filing and organization, enhancing the

discovery, sharing, and use of information, and AI-

powered business process automation are all

extremely promising. AI integration aims to facilitate

automated security monitoring and help address

compliance risks. In the face of information overload

and information silos, generative AI is valuable for

extracting relevant information and curating

knowledge tailored to practitioners’ needs. However,

the risks associated with AI use are rarely mentioned.

5.2 AI Explainability and

Transparency in EIM

The study’s findings address the need for

contextualized AI explainability and transparency in

EIM, and, in addition to developing evaluative

categories for understanding AI, we have presented a

working definition of AI transparency for EIM

practitioners with six criteria covering information

transparency and transparency-in-use available for

consideration. By evaluating the transparency of AI-

integrated EIM solutions offered by vendor platforms

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

(Table 2), we find that AI transparency can be

improved to facilitate explainability and

understanding, trust, and adoption of AI by

practitioners. When working with AI, practitioners

might not be interested in all the details but require

proven success in real-world scenarios that address

similar EIM needs.

Regarding ease of use and confidence in AI-

produced outcomes, more transparency is needed

about how practitioners can interact with AI. This

includes understanding how to determine the

accuracy or confidence in the results produced by AI

systems and how practitioners can be included in the

loop to improve this process.

5.3 Limitations and Outlook

There are two limitations to this work. Firstly, the

analysis is based on scanning the publicly available

information provided on the websites of 20 leading

EIM platforms. Both the size of the sample and the

information sourced are limited. Future work may

incorporate more public discourse on AI-integrated

EIM offerings, such as industry reports and platform

blogs, to achieve a more comprehensive analysis.

Secondly, while this study explains the need for

contextualized AI transparency for EIM practitioners

and proposes a working definition of AI

explainability and transparency with evaluation

criteria, more work is needed to verify these criteria

with EIM practitioners. That constitutes the second

stage of this study. Practitioner interviews have been

completed and will be reported at a later stage. This

research takes the first step towards making AI

applications more understandable, explainable, and

transparent in EIM. This work also contributes to the

field of Explainable AI by addressing the needs of

non-experts in applying and working with AI, thereby

promoting human agency in explainable AI (XAI)

initiatives.

Table 2: Evaluation of AI transparency in EIM.

Transparency Evaluation criteria

Information Transparency Transparency-in-use

Platform

1) Provision of AI

development details

Provision

of AI

function

details

3) Provision of

AI impacts

Provision

of real-

world use

cases

5) User

experience

design for

AI-

integrated

interface

6) Human-

interaction

Development

Techniques

Benefits

Risk

√ √ √ √

√

√ √ √

√

√ √

√ √ √

√

√ √ √

√

√ √

√ √ √ √

√ √ √

√ √ √ √

√ √

√

√ √

√

#10

√ √ √ √

#11

√ √ √ √ √

√ √

#12

√ √ √

√

#13

√ √ √ √

#14

√ √ √ √

√ √

#15

√ √ √ √

√

#16

√ √ √ √

√ √

#17

√ √ √ √

√

#18

√ √ √ √

#19

√ √ √ √

√ √ √

#20

√ √ √ √

√

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice

ACKNOWLEDGEMENTS

This research was funded by the ARC Industrial

Transformation Training Centre for Information

Resilience (CIRES). The authors gratefully

acknowledge the support provided by CIRES, which

has been instrumental in conducting this work.

REFERENCES

Adadi, A., & Berrada, M. (2018). Peeking Inside the Black-

Box: A Survey on Explainable Artificial Intelligence

(XAI). IEEE Access, 6, 52138-52160.

https://doi.org/10.1109/ACCESS.2018.2870052

AIIM. (2024, April 5). Intelligent Information Management

Glossary. https://www.aiim.org/what-is-information-

management

Altheide, D. L., & Schneider, C. J. (2013). Qualitative

media analysis (2nd edition ed.). Sage.

Andrada, G., Clowes, R. W., & Smart, P. R. (2023).

Varieties of transparency: exploring agency within AI

systems. AI & SOCIETY, 38(4), 1321-1331.

https://doi.org/10.1007/s00146-021-01326-6

Auster, E., & Choo, C. W. (1994). How senior managers

acquire and use information in environmental scanning.

Information Processing & Management, 30(5), 607-

618.

Barredo Arrieta, A., Díaz-Rodríguez, N., Del Ser, J.,

Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-

Lopez, S., Molina, D., Benjamins, R., Chatila, R., &

Herrera, F. (2020). Explainable Artificial Intelligence

(XAI): Concepts, taxonomies, opportunities and

challenges toward responsible AI [Article]. Information

Fusion, 58, 82-115. https://doi.org/10.1016/j.inffus.20

19.12.012

Baviskar, D., Ahirrao, S., Potdar, V., & Kotecha, K. V.

(2021). Efficient Automated Processing of the

Unstructured Documents Using Artificial Intelligence:

A Systematic Literature Review and Future Directions.

IEEE Access, 9, 72894-72936.

Brennen, A. (2020). What do people really want when they

say they want "explainable AI?" we asked 60

stakeholders. Conference on Human Factors in

Computing Systems - Proceedings,

Bunn, J. (2020). Working in contexts for which

transparency is important: A recordkeeping view of

explainable artificial intelligence (XAI). Records

Management Journal, 30(2), 143-153.

De Certeau, M., & Mayol, P. (1998). The Practice of

Everyday Life: Living and Cooking. Volume 2 (Vol. 2).

U of Minnesota Press.

De, T., Giri, P., Mevawala, A., Nemani, R., & Deo, A.

(2020). Explainable AI: A Hybrid Approach to

Generate Human-Interpretable Explanation for Deep

Learning Prediction. Procedia Computer Science, 168,

40-48.

https://doi.org/https://doi.org/10.1016/j.procs.2020.02.

255

Duranti, L., Abdul-Mageed, M., Hofman, D., & Sullivan,

P. (2022). I Trust AI, the latest InterPARES research

project. Anuario Escuela de Archivología(13), 36-55.

Felzmann, H., Fosch Villaronga, E., Lutz, C., & Tamò

Larrieux, A. (2019). Transparency you can trust:

Transparency requirements for artificial intelligence

between legal norms and contextual concerns. 6, 1-14.

https://doi.org/10.1177/2053951719860542

Fensham, R., Threadgold, T., Webb, J., Schirato, T., &

Danaher, G. (2020). Understanding Bourdieu.

Routledge.

Goudarouli, E., Sexton, A., & Sheridan, J. (2019). The

challenge of the digital and the future archive: through

the lens of the national archives uk. Philosophy &

Technology, 32, 173-183.

Gupta, A., & Kapoor, N. (2020). Comprehensiveness of

archives: A modern AI-enabled approach to build

comprehensive shared cultural heritage. arXiv preprint

arXiv:2008.04541.

Haresamudram, K., Larsson, S., & Heintz, F. (2023). Three

Levels of AI Transparency. Computer, 56(2), 93-100.

https://doi.org/10.1109/MC.2022.3213181

Hausmann, V., Williams, S. P., Hardy, C. A., & Schubert,

P. (2014). Enterprise Information Management

Readiness: A Survey of Current Issues, Challenges and

Strategy. Procedia technology, 16, 42-51.

https://doi.org/10.1016/j.protcy.2014.10.066

Huddart, K. (2022). Artificial intelligence powered digital

asset management: Current state and future potential.

Journal of Digital Media Management, 11(1), 6-17.

Jaakonmäki, R., Simons, A., Müller, O., & vom Brocke, J.

(2018). ECM implementations in practice: objectives,

processes, and technologies. Journal of Enterprise

Information Management, 31(5), 704-723.

Jaillant, L. (2022). How can we make born-digital and

digitised archives more accessible? Identifying

obstacles and solutions. Archival Science, 22(3), 417-

436.

Kiseleva, A., Kotzinos, D., & De Hert, P. (2022).

Transparency of AI in Healthcare as a Multilayered

System of Accountabilities: Between Legal

Requirements and Technical Limitations [Review].

Frontiers in Artificial Intelligence, 5.

https://doi.org/10.3389/frai.2022.879603

Kolandaisamy, R., Rajagopal, H., Kolandaisamy, I., &

Sinnappan, G. S. (2024). The Smart Document

Processing with Artificial Intelligence.

Langer, M., Oster, D., Speith, T., Hermanns, H., Kästner,

L., Schmidt, E., Sesing, A., & Baum, K. (2021). What

do we want from Explainable Artificial Intelligence

(XAI)? – A stakeholder perspective on XAI and a

conceptual model guiding interdisciplinary XAI

research [Article]. Artificial Intelligence, 296, Article

103473. https://doi.org/10.1016/j.artint.2021.103473

Lau, F., Antonio, M., Davison, K., Queen, R., & Bryski, K.

(2020). An environmental scan of sex and gender in

electronic health records: analysis of public information

KMIS 2024 - 16th International Conference on Knowledge Management and Information Systems

sources. Journal of Medical Internet Research, 22(11),

e20050.

Martens, D., & Provost, F. (2014). Explaining Data-Driven

Document Classifications. MIS quarterly, 38(1), 73-

100. https://doi.org/10.25300/MISQ/2014/38.1.04

Meske, C., Bunde, E., Schneider, J., & Gersch, M. (2022).

Explainable Artificial Intelligence: Objectives,

Stakeholders, and Future Research Opportunities.

Information systems management, 39(1), 53-63.

https://doi.org/10.1080/10580530.2020.1849465

Modiba, M. (2023). User perception on the utilisation of

artificial intelligence for the management of records at

the council for scientific and industrial research.

Collection and Curation, 42(3), 81-87.

OpenAI. (2022, November 30). Introducing ChatGPT.

https://openai.com/index/chatgpt/

Orlikowski, W. J., & Scott, S. V. (2016). Digital work: A

research agenda. A research agenda for management

and organization studies, 88-95.

Ragano, A., Benetos, E., & Hines, A. (2022). Automatic

Quality Assessment of Digitized and Restored Sound

Archives. Journal of the Audio Engineering Society.

Sambetbayeva, M., Kuspanova, I., Yerimbetova, A.,

Serikbayeva, S., & Bauyrzhanova, S. (2022).

Development of Intelligent Electronic Document

Management System Model Based on Machine

Learning Methods. Eastern-European Journal of

Enterprise Technologies, 1(2), 115.

Schneider, J., Adams, C., DeBauche, S., Echols, R.,

Mckean, C., Moran, J., & Waugh, D. (2019).

Appraising, processing, and providing access to email

in contemporary literary archives. Archives and

manuscripts, 47(3), 305-326.

Scifleet, P., Felsbourg, M., Dang, C., & et al. (2023).

Information governance and information management

as a service: survey report 2023 [Report]. Swinburne

University of Technology. https://apo.org.au/node/

323899

Shabou, B. M., Tièche, J., Knafou, J., & Gaudinat, A.

(2020). Algorithmic methods to explore the automation

of the appraisal of structured and unstructured digital

data. Records management journal, 30(2), 175-200.

Wang, X., & Yin, M. (2021). Are Explanations Helpful? A

Comparative Study of the Effects of Explanations in

AI-Assisted Decision-Making. International Conferen-

ce on Intelligent User Interfaces, Proceedings IUI,

Wanner, J., Herm, L.-V., Heinrich, K., & Janiesch, C.

(2022). The effect of transparency and trust on

intelligent system acceptance: Evidence from a user-

based study. Electronic Markets, 32.

https://doi.org/10.1007/s12525-022-00593-5

Williams, S. P., Hausmann, V., Hardy, C. A., & Schubert,

P. (2014). Managing enterprise information: Meeting

performance and conformance objectives in a changing

information environment. International journal of

information systems and project management, 2(4), 5-

36. https://doi.org/10.12821/ijispm020401

Williamson, K., & Johanson, G. (2018). Research methods:

information, systems and contexts (Second edition.

ed.). Chandos Publishing.

Yin, X., Liu, H., Webster, J., Trieu, K., Huffman, M. D.,

Miranda, J. J., Marklund, M., Wu, J. H., Cobb, L. K., &

Li, K. C. (2021). Availability, formulation, labeling,

and price of low-sodium salt worldwide: environmental

scan. JMIR public health and surveillance, 7(7),

e27423.

Zhang, X., Majid, S., & Foo, S. (2011). The Contribution

of Environmental Scanning to Organization

Performance. Singapore Journal of Library &

Information Management, 40.

Charting the Transformation of Enterprise Information Management: AI Explainability and Transparency in EIM Practice