The third Board meeting of the year was held from September 9 to 12, 2025, in Lisbon, Portugal.

Thank you to hosts the Ordem dos Contabilistas Certificados (Order of Certified Accountants) for welcoming the IPSASB back to Lisbon and hosting both the September Board meeting and the 5th Public Sector Standard Setters Forum. The IPSASB also extends its thanks to its members, technical advisors, and standard setters from around the globe who joined the Forum to shape the future of public sector standard setting.

The IPSASB held its second meeting of the year from June 10 to 13, 2025, in Toronto, ON, Canada.

At our meeting last week, we continued our analysis of responses to the Climate-related Disclosures and Tangible Natural Resources exposure drafts, and we progressed several other projects that will enhance IPSAS Standards, which provide the cornerstone for effective and efficient public financial management.

Thank you to The World Bank for hosting the IPSASB®'s first meeting of the year from March 18 to 21 in Washington, D.C., USA.  

We progressed several projects that will make our IPSAS® Standards, which are cornerstones of strong public financial management, easier to use and more effective.

Informed decision-making matters.

Reliable financial information supports sound policymaking, enabling governments to allocate resources effectively to achieve their goals.

Welcome to the sixth Market Scan from the IAASB's Disruptive Technology team. Building on our ongoing work, we issue a Market Scan every 2-3 months. Market Scans cover exciting trends, including new developments, corporate and start-up innovation, noteworthy investments and what it all might mean for the IAASB.

In this Market Scan, we explore Robotic Process Automation, a technology used for executing repetitive tasks that has applications across the audit process, from data transformation to workpaper creation, as well as potential for forming part of an audited entity’s IT environment.

Technology Landscape, September 2022

We cover:

• What is Robotic Process Automation and why is it important?
• The latest developments
• What this might mean for the IAASB

What is Robotic Process Automation and why is it important?

Robotic Process Automation (RPA) is a technology that involves creating software robots or “bots” to perform repetitive, routine manual tasks, such as extracting data, filling out forms or moving files. By completing rules-based actions that emulate human processes, RPA tools can autonomously complete various activities across multiple unrelated software systems. 

What is RPA (Robotic Process Automation)? Two-minute watch, Eye on Tech

What is RPA (Robotic Process Automation)? Two-minute min watch, Eye on Tech

RPA has evolved in recent years with companies such as Automation Anywhere, Blue Prism and UiPath developing ever more sophisticated technologies to automate tasks, workflows and processes. Coupling RPA with artificial intelligence technologies, such as machine learning, has enabled businesses to automate more complex tasks that may involve more judgment or use of unstructured data. This combination of technologies is often referred to as Intelligent Automation or Cognitive Automation.

RPA within the audit process

There are several areas across the audit where RPA can be deployed to drive efficiency in the audit process as well as improve audit quality by enabling consistency of outputs.

• Data preparation – combining data sets (e.g., general ledger transactions and trial balance information) and carrying out transformation actions to prepare data for use in automated tools.
• Transaction matching and reconciliation – automating the process of matching transactions to external datasets (e.g., bank statements)
• Basic documentation – completion of basic standardized documentation, such as workpapers, checklists or templates based on prescribed inputs (e.g., trial balance information or information documented elsewhere in the audit file)

RPA within audited entities

Audited entities may use RPA to support certain business processes or repetitive task activities, such updating records, processing transactions or completing reconciliations.

Benefits of Robotic Process Automation, Digital Beanie

Recent Noteworthy Developments in Robotic Process Automation

These recent developments that may signal future disruption in this area. It is not a complete list of all activities in the field of RPA. For a reminder of Key Venture Capital and Investment terms please refer to the first Market Scan.

1.  RPA market experiences slowing growth

The RPA market is dominated by some key players, with the 10 largest RPA vendors accounting for over 75% of the market. There have been some indications in the RPA market that growth in this technology sector may have begun to slow, although double digit growth is still predicted for 2023. UiPath, one of the leading companies in RPA software, has seen its stock value continue to fall since its IPO in 2021. Whilst Automation Anywhere, another well-known RPA provider, sought to secure a $200 million loan and Blue Prism, another large player, agreed to a $1.5 billion takeover offer from Vista Equity Partners in September 2022.

2.     Startups with Low-Code and No-Code RPA Solutions

Whilst the traditional RPA industry may be slowing down, business process automation— a term used to describe the use of technologies, including RPA and AI, to streamline and automate more complex business processes—has seen continued growth, particularly with low-code or no-code solutions that can be deployed by those without coding skills.

• In May 2022, UiPath announced a new partnership with airSlate, a workflow automation company, to advance the development of workflow solutions to enable individuals and small businesses to digitally transform their organizations.
• In June 2022, Next Matter, a workflow automation platform for business operations, raised $16 million in a Series A funding round to expand its team and operations.
• In November 2022, Rewst a no-code RPA provider for managed service providers, raised $21.5 million in Series A funding to continue to build out its Robotic Operations Center.

What this might mean for the IAASB

The IAASB is interested in maintaining its collective knowledgebase on digital technologies (including on specific sub-topics such as robotic process automation), promoting digital readiness and enablement through its engagement with stakeholders, and encouraging action by others to supplement and support the IAASB’s standard-setting activities.

The IAASB is also keen to explore how technologies could be used to enhance interaction with auditing standards. Subject to the IAASB’s work plan decisions, possible use cases of digital technologies for audited entities and audit engagements might provide input to further modernize the IAASB’s standards to be adaptable to and reflect the current business and audit environment (while recognizing that the standards would address digital technologies in a principles-based manner).

The increasing accessibility of RPA through the growth of low code and no code technologies offers opportunities for those without extensive programming knowledge to develop solutions to address specific pain points or low value time-consuming activities in their everyday work, including for audit practitioners or audited entities.  

Where RPA or other robotics-based technologies are deployed by auditors to complete audit procedures, adequate governance and approval of these technologies needs to be in place to ensure they are designed effectively, fit for purpose and used appropriately, including provision of training or enablement where needed. Availability of no-code or low code technologies may give rise to wider use of small-scale RPA to address specific needs which has the potential to improve audit efficiency but if ungoverned, could impact audit quality.

The use of RPA in audited entities to automate actions previously performed by a person may change how a process and its related controls operate, and therefore the risks within the process and the appropriate audit response, which may include understanding how the use of RPA is being governed and controlled. As the use of this technology develops further, particularly when coupled with artificial intelligence technologies, this may present new challenges for the auditor and may, therefore, have implications for the IAASB’s work, whether in terms of future standard setting related to technology or facilitating or supporting action by others (e.g., developing guidance).

Useful Articles and Resources 

• Low/No-Code RPA: What it is & Development Best Practices | AI Multiple, four-minute read
• Robotic Process Automation for Auditing | Journal of Emerging Technologies in Accounting, eight-minute read
• Robots strengthen the Digital workforce | Deloitte Switzerland | Audit, four-minute read
• Is hyperautomation worth the hype? | E&T Magazine, five-minute read
• How do you audit a robot? | ICAEW, four-minute read
• What is business process automation? |Red Hat, three-minute read

Interesting story

KFC Draws Backlash for Automated Kristallnacht Promo Message - BNN Bloomberg, two-minute read

KFC, the popular fast-food chain, experienced a public relations disaster when its automated communications process messaged customers in Germany about a promotion to “commemorate Kristallnacht”. Kristallnacht was a series of Nazi pogroms that took place against German Jews in 1938. The company blamed an error in the automated process which linked calendar events to promotional content.

The IPSASB held its fourth meeting of the year in Toronto, Canada on December 6-9, 2022. 

Welcome to the fifth Market Scan from the IAASB's Disruptive Technology team. Building on our previous work, we issue a Market Scan approximately every two to three months. Market Scans cover exciting trends, including new developments, corporate and start-up innovation, noteworthy investments and what it all might mean for the IAASB. Special thanks to Maddie Zietsman for helping to author this edition.

In this Market Scan, we explore Homomorphic Encryption for Analyzing Encrypted Data, a technology which has applications within Protecting Information. This technology has the potential to impact how data is used in the audit—creating opportunities for greater collaboration and access to specialist skills.

We cover:

• What is Homomorphic Encryption and why is it important?
• The latest developments
• What this might mean for the IAASB 

What is Homomorphic Encryption? Why is it important?

Homomorphic Encryption (HE) is a set of algorithms that allows computations to be done on encrypted data without the need for decryption. Homomorphic encryption lets data be protected while “in use”, so analysis can be run directly on encrypted information without disclosing it and providing complete confidentiality during analysis.

Source: What is Homomorphic Encryption?, OpenMined

There are two main types of homomorphic encryption, Partial Homomorphic Encryption, which supports only a single operation over encrypted data, and Fully Homomorphic Encryption, which supports multiple operations. Federated Learning (FL) is another privacy-enhancing technology that distributes machine learning across devices or servers, thereby reducing latency and security risk whilst protecting privacy.

Fully Homomorphic Encryption: Why it Matters, IBM News, three-minute watch 

Homomorphic encryption has many potential benefits for a wide range of industries from healthcare to financial services. From an audit and assurance perspective, there are several areas where homomorphic encryption can be leveraged.

• Using aggregated data to securely achieve common goals – Audit firms or other organizations that may perceive privacy or confidentiality risks when working together could collaborate using encrypted data to achieve a common goal such as developing fraud pattern detection applications. Using homomorphic encryption, encrypted data sets from multiple sources could be linked together, used to train an AI application, and develop a technology product for all parties to use.
• Enabling use of third parties without compromising data privacy – Homomorphic encryption may enable audit practitioners to leverage third parties with greater analytics capabilities or expertise to perform analysis on encrypted data to support audit procedures—an approach that would be difficult if not impossible without the encryption technology.
• Enhancing effectiveness of cross-border audits – Homomorphic encryption could be used to enable data analysis across borders while respecting data residency and privacy laws. This would be particularly beneficial to group audits with components in jurisdictions with strict data residency restrictions.
• Greater capability to perform benchmarking – Homomorphic encryption could be used to provide benchmarks across industries, including competitive companies, without exposing market sensitive data. Benchmarking data may be used when performing an audit, for example when performing analytical procedures.
• Mitigating bias whilst stress testing models – Using homomorphic encryption, machine learning models and algorithms could be stress tested using encrypted data sets, so the data could not be fitted to the model ahead of time.

All these areas focus on homomorphic encryption’s ability to increase the data analysis that can be done while still ensuring data security and privacy. As the technology gains wider traction, it offers audit and assurance practitioners opportunities to increase their analytical capabilities and leverage the specialized skills of other entities or parties, without compromising data privacy.

Recent Noteworthy Developments in Homomorphic Encryption

This section is designed to provide examples of recent developments that may signal future disruption in this area. It is not a complete list of all activities in Homomorphic Encryption. For a reminder of Key Venture Capital and Investment terms please refer to the first Market Scan.

1. Big player activity

Homomorphic Encryption is gaining traction and growing fast. Top companies, such as Intel, Microsoft and Google, are leveraging the power of this technology and working to develop its use in various sectors of the economy.

• Study shows growing interest in homomorphic encryption technologies

A December 2021 study by Deloitte noted 19 different “publicly announced pilots, products, and proofs of concept for homomorphic encryption”. Companies that are leading these pilots include large companies like Apple, Google, Microsoft, Nvidia, IBM, and more. Finance, health, and social care currently dominate the pilot projects, but the expectation is that more industries will reap the benefits from the technology as it continues to gain leverage. These pilots also present opportunities for the audit and assurance industry to capitalize on the power of homomorphic encryption and how the using data encryption may contribute to or enhance the performance of quality audit or assurance engagements.

Keeping AI private/Homomorphic encryption and federated learning can underpin more private, secure AI, Deloitte Insights , six-minute read

• Intel and Microsoft announce collaboration on security technology

Intel and Microsoft have partnered together as part of a DARPA program to focus on reducing the overhead that is associated with using homomorphic encryption. To reap the benefits of this technology, it is important that it is both accessible and affordable. Both Intel and Microsoft’s investment in time and research in homomorphic encryption reveals the importance that both companies see in the technology’s ability to change the working world. As these large companies continue their research and testing, it may not be long before homomorphic encryption is accessible to companies of all sizes.

2. Start-up activity

As homomorphic encryption becomes more prominent, there have been a few key start-ups that have spearheaded its development, including those highlighted below.

• Duality advances Homomorphic Encryption Landscape

Duality, a leader in enabling privacy-enhanced collaboration on sensitive data launched their open-source fully homomorphic encryption (FHE) library, OpenFHE, in July 2022. This was a collaborative project with other leaders in cryptography including Intel, Samsung, University of California-San Diego and MIT. OpenFHE is considered a next generation open-source FHE software library providing even greater security, robust privacy protection and wider useability.

• Enveil announces new encrypted training solution

Enveil, a start-up company founded in 2016, has been a key leader in developing homomorphic encryption and federated learning technologies. In June 2022, Enveil announced a new solution called ZeroReveal ML Encrypted Training (ZMET), which enables encrypted federated learning and usage of decentralized datasets for machine learning applications.

• New start-up sees web3 opportunity using homomorphic encryption

Brand new start-up, Sunscreen, just raised $4.65 million in seed funding to develop advanced privacy technology for the next generation of the world wide web, web3. Currently zero-knowledge proofs (ZKPs), which allow for a transaction to be verified on a blockchain without the underlying data being shared, are seen as the main solution for improving privacy in web3 but require significant processing power. However, co-founder and CEO of Sunscreen, Ravital Solomon, thinks “fully homomorphic encryption is even more promising in its potential to bolster privacy in web3.”

What this might mean for the IAASB

The IAASB is interested in maintaining its collective knowledgebase on digital technologies (including on specific sub-topics such as homomorphic encryption), promoting digital readiness and enablement through its engagement with stakeholders, and encouraging action by others to supplement and support the IAASB’s standard-setting activities. The IAASB is also keen to explore how technologies could be used to enhance interaction with auditing standards. Subject to IAASB’s work plan decisions, possible use cases of digital technologies for audited entities and audit engagements might provide input to further modernize IAASB’s standards to be adaptable to and reflect the current business and audit environment (while recognizing that the standards would address digital technologies in a principles-based manner).

Access to appropriate and reliable data is fundamental to being able to use automated tools and techniques in the audit. Considerations around data protection and privacy are key to this approach and homomorphic encryption presents a potential solution to current data access restrictions.

Homomorphic Encryption also offers opportunities for those in audit and assurance to develop advanced analytics, machine learning and AI technologies through enabling more options for data management.

However, using this technology may present unique practical challenges with applying certain principles set out in existing standards that address aspects of, for example, quality management, audit evidence, service organizations and using the work of others which may indicate the need for additional guidance. Given the nascent nature of this technology it is too early to fully comprehend the practical implications related to its use, but the IAASB will continue to monitor developments in this area.

Useful Articles and Resources

• Basics
  • What is Homomorphic Encryption? Benefits & Challenges [2022] AImultiple.com, three-minute read
  • What Is Homomorphic Encryption & How Is It Used | Venafi, four-minute read and three-minute video
• Further reading
  • Everything You Wanted to Know About Homomorphic Encryption (But Were Afraid to Ask) | Forbes.com, four-minute read
  • What is Federated Learning? | Tech in 3, four-minute read
  • Privacy Enhancing Technologies (PETs) | Enveil, infographic
  • What Are the Latest Developments in Homomorphic Encryption? [Ask the Experts] | Venafi, two-minute read and four-minute video

Interesting story

Ink filled with secret molecules can hide encryption key in a letter | New Scientist two-minute read

Scientists at the University of Texas have developed an ink containing polymers that can store data and have used it to write a letter containing a hidden message. “The idea of writing a message but the real, hidden message is contained in the molecular structure of the ink is fascinating, although maybe not the most practical method,” says Alan Woodward at the University of Surrey, UK.

What do you think about this bulletin?

Please take the time to fill out our quick survey to let us know your thoughts about this bulletin, how it can be improved and what you would like to hear about going forward.

The IPSASB held its third meeting of the year in Lisbon, Portugal on September 12-16, 2022. 

Welcome to the fourth market scan from the IAASB's Disruptive Technology team. Building on our previous work, we will issue a Market Scan on topics from the report approximately every two to three months. Market Scans will consist of exciting trends, including new developments, corporate and start-up innovation, noteworthy investments and what it all might mean for the IAASB.

In this Market Scan, we explore natural language processing (NLP), a technology that has applications within Accessing Information & Data (NLP and Computer Vision for Digitizing Documents) and within Assessing Internal Controls (Optical Character Recognition, NLP and Machine Learning for Intelligent Document and Voice Analysis). This technology has the potential to impact many areas of the audit—enhancing the way auditors work and providing opportunities for greater insight. 

We cover:

• What is natural language processing and why is it important?
• The latest developments
• What this might mean for the IAASB

What is Natural Language Processing? Why Is It Important?

Natural language processing (NLP) is a branch of artificial intelligence that is concerned with giving computers the ability to understand, interpret and manipulate human language, both written text and spoken words. 

Artificial Intelligence Technologies with Natural Language Processing highlighted

NLP uses a combination of technologies, including computational linguistics (rule-based modeling of human language), statistical modelling and machine learning. NLP involves both natural language understanding (NLU) and natural language generation (NLG). See the video from Simplilearn below for an explanation of how the technologies fit together.

What Is NLP And How Does It Work? | Simplilearn, five-minute watch

There are many benefits of having technology that can fully understand human speech and text (including tone and meaning), determine the right response and provide that response in a human-like format. These include: 

• Enabling better decision support through analysis of large quantities of unstructured data (e.g., emails, documents, social media) to understand sentiment, identify text similarity or discover specific themes.
• Supporting improved communication through effective translation applications (e.g., Google Translate)
• Providing around-the-clock support to customers, clients, users and other help-seekers through chatbots, voice assistants and applications that use semantic search functionality
• Assisting with efficient documentation through generative language applications that can automatically create text content, such as high-level insights or summaries.

In an audit and assurance context, NLP-based technologies can be utilized to support and enhance auditor activities. For example:

• Providing auditor assistance by delivering relevant guidance to the auditor, when and where they need it, through voice assistants and help bots.
• Enriching risk identification and assessment activities by providing valuable insights about the entity and its environment, from analysis of unstructured data from a variety of sources such as regulatory notices, social media, and news articles.
• Supporting understanding the entity's internal controls by extracting and summarizing what has been written down in process documents, emails, articles, and from employee inquiries.
• Augmenting documentation activities through automatic text generation, for example, by providing commentary from data analysis results. See the below video from Wordsmith for an example of technology that can do this.

Wordsmith: Extend the Power of Tableau, Automated Insights, two-minute watch

These technologies offer opportunities, such as those described above, to the auditor or the audited entity, which could reduce manual, time-intensive activities; enhance the effectiveness of certain procedures, tasks, or actions; and augment decision making. However, they are not without their risks and adoption needs to go hand in hand with a recognition that imperfections in outputs could arise. 

Recent Noteworthy Developments in Natural Language Processing

These recent developments may signal future disruption in this area; this is not a complete list of all activities in the natural language processing. For a reminder of Key Venture Capital and Investment terms please refer to the first Market Scan.

1. Transformers—boosting NLP development

Transformers are deep learning models used in natural language processing for translation and text summarization. Recent developments have seen these technologies trained with large language datasets, leading to a growth in pre-trained systems like Generative Pre-trained Transformer (GPT) and Bidirectional Encoder Representations from Transformers (BERT). Some notable developments include:

• GPT-3 from OpenAI is a hugely popular model that can generate large amounts of text based on a small amount of text input. It is used in many copywriting and content generation applications available today. However, the solution has gained notoriety for the machine learning bias it picked up from the human bias in the internet text it learned from. The Gender Bias Inside GPT-3 | Made by McKinney, five-minute read

What is GPT-3 (Generative Pre-Trained Transformer 3) – Tech Target,  three-minute watch

• Google’s BERT is the underlying model for the Google search engine—and due to be replaced by MUM, a Multitask Unified Model, during 2022 and described as 1,000 times more powerful than BERT.
• Megatron-Turing NLG is the latest model from Microsoft and Nvidia and described as the largest and most powerful generative language model with 530 billion parameters.

2. Start ups

• Enterpret launches with $4.3M, NLP technology to decipher customer feedback | TechCrunch
  • Enterpret, a start-up founded by brothers Varun and Arnav Sharma, is developing natural language models to improve insights from customer feedback. This technology, which is focused on enabling more informed product development, could have widespread application across all industries with digital presence.
• Cohere raises $125m in Series B Funding
  • Cohere builds large language models and makes them available through an Application Programming Interface (API). In November 2021 Google Cloud announced a multiyear partnership with Cohere to provide infrastructure for development of their platform.
• Popular AI Writer Jarvis Rebrands to Jasper
  • This popular Y-Combinator backed startup has built an AI copywriting assistant to write original, creative content including blog articles and social media.
• OpenAI develops new tool to create code
  • OpenAI, which developed GPT-3 for generating text, has leveraged this technology to develop a tool called Copilot to help generate code. It is part of a growing trend in developing technology that enables everyone to become a code creator. 

What this might mean for the IAASB

The IAASB is interested in maintaining its collective knowledge base on digital technologies, including on specific sub-topics such as NLP, promoting digital readiness and enablement through its engagement with stakeholders, and encouraging action by others to supplement and support standard-setting activities.

The IAASB is also keen to explore how technologies, such as NLP, could be used to enhance interaction with auditing standards. Subject to the IAASB’s work plan decisions, possible use cases of digital technologies, such as NLP for audited entities and audit engagements, might provide input for further modernizing the IAASB’s standards to be adaptable to and reflect the current business and audit environment (while recognizing that the standards would address digital technologies in a principles-based manner).

NLP-based technologies present exciting opportunities for auditors to enhance the effectiveness of procedures, tasks, or actions. Widespread adoption is also likely to be more straightforward than for other technologies as it can often be introduced with limited training requirements. From the perspective of both an auditor performing an audit engagement and a firm’s quality management around the technology resources used by engagement teams, the opportunities as well as the potential challenges should be considered.

Technologies that make suggestions to the auditor, for example, about risks they may not have considered (using data extraction) or on appropriate next steps in the audit process (using a chatbot or voice assistant) offer just-in-time guidance but may give rise to automation bias. That is, the tendency to over-rely on automated aids or, more broadly, the outputs from technology solutions. In addition, challenges are likely to exist around the availability of suitable data sets for training these support technologies in order to provide appropriate recommendations given that facts and circumstances will differ from audit to audit.

Technologies that help auditors create documentation, for example, those that summarize documents or those that provide commentary on data analysis, warrant consideration about how they may blur the lines of responsibility. With these technologies often acting so seamlessly with humans, reviewers of documentation prepared by a combination of computers and humans may find it difficult to determine who did what and therefore decide on the appropriate level of scrutiny required. A principles-based approach continues to recognize that the responsibility for documentation remains that of the auditor, including that the documentation provides a sufficient and appropriate record of the work performed and conclusions reached. 

Useful Articles and Resources

• Basics
  • What Is Natural Language Processing | MonkeyLearn.com, five-minute read
  • What is NLP Natural Language Processing, advances and uses | optocrypto.com, four-minute read
• Uses in Audit and Assurance
  • Natural language processing: A more fluent audit|KPMG Australia, two-minute read
  • Citi transforms critical internal audit with machine learning, NLP and AI |IBM.com, three-minute read

Interesting Story

AI helps historians complete ancient Greek inscriptions damaged over millennia | TechCrunch, two-minute read

Deepmind, the Google-owned AI company behind the AlphaGo program that was the first to beat a professional Go player, has now developed a new tool called Ithaca. Ithaca is a machine learning model that guesses what the missing words might be in incomplete ancient Greek texts potentially shedding light on the meaning of inscriptions that are thousands of years old.

What do you think about this bulletin?

Please take the time to fill out our quick survey to let us know your thoughts about this bulletin, how it can be improved and what you would like to hear about going forward.

The IPSASB held its first meeting of the year in New York on March 21-25, 2022. 

Welcome to the third Market Scan from the IAASB's Disruptive Technology team. Building on our previous work, which included the Innovation Report created with Founders Intelligence and discussed at the January 2021 IAASB Meeting, we issue a Market Scan focusing on topics from the report approximately every two months. Market Scans consist of exciting trends, including new developments, corporate and start-up innovation, noteworthy investments and what it all might mean for the IAASB.

In this Market Scan, we explore Artificial Intelligence (AI), which is used in a broad range of technologies across the audit and assurance value chain. This Market Scan provides a high-level primer on Artificial Intelligence as it is one of the most significant and potentially disruptive technologies in audit and assurance. Future Market Scans will build on this by focusing on some of the specific AI-powered technologies highlighted below.

We will cover:

• What is AI, including related concepts of machine learning and deep learning
• AI use cases in audit and assurance
• AI challenges
• AI developments

What is Artificial Intelligence?

Artificial Intelligence (AI) is a broad discipline of computer science that refers to the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision making, and translation.

AI also describes a broad range of technologies shown in the diagram below. Many of the technologies we use every day contain one or more of these capabilities; for example, a smart speaker contains speech recognition (to turn our speech into text), natural language processing (NLP) (to understand the request and generate a response) and machine learning (to improve the quality of responses over time). 

Overview of AI Technologies

Intelligence in this context is the ability to perceive or deduce information, retain it as knowledge and apply it to making decisions. In computers this is done by analyzing large quantities of data using advanced statistics (including probability analysis) to find patterns and make predictions.

Types of AI

What is an algorithm?

Algorithms are in use all around us, although the term may not be fully understood as frequently. Think of it as a recipe used by computers: a finite sequence of well-defined instructions, typically used to solve a class of specific problems or to perform a computation. An algorithm takes an input (e.g., a dataset) and generates an output (e.g., a pattern that it has found in the data). It is like taking your ingredients and following a recipe to bake a cake.

Algorithms are not exclusive to AI. They are likely used in every audit to complete procedures such as identifying sample sizes or performing data analytics, such as ratio or regression analysis.

What Is Machine Learning?

Machine learning is about using algorithms to guide predictions. The goal of the machine learning process is to create a model, which is based on one or more algorithms. The model is developed through training with the goal that the model should provide a high degree of predictability.

One of the earliest examples of a machine learning system was a computer checkers game created by Arthur Lee Samuel at IBM. Arthur demonstrated how machine learning could work by creating a computer function to measure the chance of winning based on the position of pieces on the board. The computer then used function to determine the move most likely to lead to a successful outcome, that is, winning. The computer learns by using the feedback from playing as its data and using Arthur’s function to guide its prediction model to get to a preferred outcome.

In its simplest form, machine learning requires a five-step process:

1. Get and organize the data
2. Choose a model (one or more algorithms)
3. Train the model (using training data, about 70% of your data set)
4. Evaluate the model (using test data, about 30% of your data set)
5. Fine-tune the model and implement

Machine Learning Process

The main challenges with implementation of machine learning are in relation to what data to use (and how to get it) and what model to use, that is, which algorithms to apply.

Machine learning approaches

There are three main types of learning approach used in machine learning; determining which approach to use largely depends on what data you have available.

Supervised learning is an approach used when large amounts of labelled data are available. This enables the technology to learn by comparing its results to the correct answer. There are effectively two types of algorithms that are used within supervised learning—one is classification, where you divide the dataset into common labels. A common form of classification algorithm is called Naïve Bayes Classifier, which is used in text analysis (e.g., for sentiment analysis, email spam detection). It uses frequency and patterns in data to come up with a prediction model based on probabilities.

The other type of algorithm used in supervised learning is regression, which finds continuous patterns in data. A common form of regression algorithm is linear regression, which shows the relationship between variables and uses this to predict outcomes based on inputs, e.g., predicting expected sales per square foot of sales floor space. 

Supervised vs Unsupervised Machine learning: What’s the difference?
(Eye on Tech video, two-minute watch)

Unsupervised learning is used when the available data is unlabeled, so the algorithms used seek to put the data into groups. The most common approach is called Clustering, which is grouping similar items together and then iterating the model to get better results. There are a variety of quantitative methods, i.e., ways of grouping items. Common uses of unsupervised learning are customer segmentation for targeting marketing messages where similar customer characteristics are expected to share similar preferences.

Finally, Reinforcement learning is commonly used in gaming and robotics, effectively learning through a process of trial and error to get the most effective outcome (such as winning the game or navigating successfully around a space). 

Useful Resources

• General AI vs Narrow AI | Levity, five-minute read
  • differences and examples
• Artificial intelligence explained in 2 minutes: What exactly is AI? | KI-Campus, two-minute video
  • simple explanation with examples
• AI and Machine Learning Explained with Examples | ClassicInfomatics, seven-minute read
  • machine learning methods explained with examples
• Top 15 Hot Artificial Intelligence Technologies | Edueka!, six-minute read
  • short explanations of key AI technologies

What Is Deep Learning?

Deep learning is a subfield of machine learning that uses neural networks for learning and bear some resemblance to how the human brain works. This way of processing data is more granular than with machine learning and involves more layers of analysis. Although the concept of deep learning has been around since the 1970s, its recent growth is due to the significant advancements in computing power. It is commonly used for speech and image recognition.

An artificial neural network ingests data through an input layer, processes it through a complex network (known as the hidden layer or layers) to provide an output. The word “hidden” in the hidden layer simply refers to the fact that the units in the layer are not visible to external systems and are “private” to the neural network. 

Example of a neural network used to identify the number 4
(From Deep Learning with Python by Francois Chollet)

Each of the processing units in the network is called a neuron. A neuron is a container with an input value, a weighting, and a bias (which is a constant). These are computed together and then an activation function is applied, which is effectively a mathematical operation that normalizes the inputs and produces an output that is then passed onto neurons in the next layer.

The weightings along with the bias can change the way the neural networks operate and are used to refine the model to get to the preferred outcome.

The most common types of neural networks are called fully connected neural networks, referring to all the neurons having connections from layer to layer. Other neural networks include recurrent neural networks, convolutional neural networks and generative adversarial networks.

In Recurrent Neural Networks (RNNs), the function not only processes the input but also prior inputs across time. An example of this is with predictive text, as you start to type, different word options are presented based on what the system predicts you are typing.

In Convolutional Neural Networks (CNNs), data is processed in stages from easy to complex with each of the stages being a convolution. CNNs are often used in computer vision applications such as image recognition software.

Generative Adversarial Networks (GANs) are a relatively new but powerful class of neural network used for unsupervised learning. They are made up of a system of two neural network models (a generator and a discriminator) that compete with each other and are able to analyze, capture and copy the variations in a dataset. It is this technology that gave rise to creation of deepfakes; they have also begun to be used by the financial services sector to help with fraud identification.

Useful Resources

• What is a Neural Network? | DeepAI, four-minute read
  • simple explanation and example
• What is a Neural Network? | Simplilearn, five-minute video
  • neural network basics and examples

AI Use Cases in Audit and Assurance

There are many ways that AI may be deployed to support the audit process.

Audit Planning

• Resource optimization using AI technology to analyze staff profiles and experience to bring together the best team for the type of audit engagement
• Client acceptance procedures using AI to analyze data from non-traditional sources, such as social media, emails, phone calls, public statements from entity management, etc., to identify potential risks relevant to client acceptance and continuance assessments.

Understanding the entity and its systems, and identifying risks

• Using natural language processing and machine learning AI technologies to analyze structured and unstructured information, such as global regulatory notices, industry reports, regulatory penalties, news, public forums, etc., to detect risks of audit relevance
• Intelligent document analysis, such as optical character recognition natural language processing and machine learning technologies, to derive insight from unstructured data sources like email, documents, transcribed voice, images, etc. to support understanding of the entity’s information system and related controls.
• Quickly and more efficiently understanding the entity's internal controls by summarizing and extracting what has been documented in process documents, emails, articles, and from employee inquiries.
• AI-powered behavioral analytics to identify suspicious or unusual entity employee behavior and intent, such as data exfiltration, employee collusion or abuse from privileged users.
• Enhancing an audit team's judgments on higher-risk areas of audit engagements by using AI to identify common risks relevant to entity’s industry, regulatory environment, operating locations and other external factors.

 Substantive Procedures

• AI tools, benefiting from increases in the quality and quantity of available “training” data, can be applied to data sets to algorithmically identify outliers and anomalous data and to perform predictive analytics for use in areas such as testing large transaction populations, auditing accounting estimates and going concern assessments.
• Document processing, review and analysis by using optical character recognition to identify and extract key details from contracts (e.g., leases) and other documents (e.g., invoices)
• Inventory and physical asset verification procedures through use of drones with computer vision (image recognition) particularly for larger capital assets, such as trucks, or the inspection of large-scale business sites, such as wind farms.

Conclusion Procedures

• AI technologies to support auditors’ work on financial statement disclosures enabling easier identification of missing disclosure requirements and non-compliance.
• AI technologies to support tick and tie of underlying audit work through to financial statements and related disclosures

Some of these technologies will be explored in more detail in future Market Scans.

Many organizations are expanding their use of AI across parts of their business with the goal of driving operational efficiencies, better informed decision making and generating growth through innovation. As a result, it is likely that this technology will become a relevant consideration when performing audit procedures, particularly regarding risk identification and assessment, and risk response activities.

Useful Resources

• The Intelligent Audit | ISACA, nine-minute read
  • examples of use of AI in the audit
• AI in the Accounting Big Four – Comparing Deloitte, PwC, KPMG, and EY | Emerj, 11 minute read 

AI Challenges

Where AI is deployed, whether by the auditor in carrying out their procedures or by an audited entity within their business operations, the associated risks need to be identified and appropriately managed. Many assurance firms and organizations have developed methodologies that provide a framework for identifying and managing AI related risks. In September 2021, COSO issued new guidance setting out how to apply “the COSO framework and principles to help implement and scale artificial intelligence”.

This guidance identifies five areas of AI related risks:

• Bias and reliability breakdowns due to inappropriate or non-representative data
• Inability to understand or explain AI model outputs
• Inappropriate use of data
• Vulnerabilities to adversarial attack to obtain data or otherwise manipulate the AI model
• Societal stresses due to rapid application and transformation of AI technologies

It concludes that appropriate risk management is needed to ensure that AI solutions are “trusted, tried and true”.

Auditing AI may require a different set of skills to those currently applied in today’s audits and many firms are updating their recruitment strategies, training curricula and audit methodologies to respond to the growing need for AI competencies. Future Market Scans will explore some of these challenges in more detail.

Useful Resources

• 6 lessons from audit experts who adopted AI early | Journal of Accountancy, five-minute read
  • transcript from interview with auditors who have implemented AI
• Auditing Artificial Intelligence | Corporate Compliance Insights, six-minute read
  • impact of AI on planning an Internal audit
• COSO Releases New Guidance: Realize the Full Potential of Artificial Intelligence | COSO press release, two-minute read

Audit and Assurance Publications

• The Data-Driven Audit: How Automation and AI are Changing the Audit and the Role of the Auditor | AICPA
• A CPA's introduction to AI: From algorithms to deep learning | CPA Canada

AI Developments

The global AI market is expected to achieve a compound annual growth rate of nearly 40% over the next five years and whilst AI technologies such as natural language processing and speech recognition are maturing, others such as deep learning and Generative AI have significant scope for development.

Here are some recent noteworthy developments:

Regulation and Explainable AI

One of the issues that has arisen with AI is around the negative impact of biases in algorithms and the harm that this can cause. In a recent survey more than one in three companies surveyed disclosed that they had suffered losses (revenue, customers or staff) due to AI bias in their algorithms. In response, there is an expectation regulation will be established in the near future. The EU, in its white paper, “On Artificial Intelligence—A European Approach to Excellence and Trust”, noted that explainability is a key factor to improving trust in AI. Many companies are, therefore, expected to look to implement explainable AI in which the results of the solution can be understood by humans.

Efficient AI

DeepMind, the company behind the AlphaGo program that was the first to beat a professional Go player, has developed an AI large language model—that is, a statistical tool to predict words—called RETRO (Retrieval-Enhanced Transformer). This AI technology, built to generate convincing text, chat with humans and answer questions is said to match the performance of neural networks 25 times its size through use of a text database.

Decision Intelligence

One of the top technology trends for 2022 noted by Gartner is decision intelligence, which is using AI to enhance and support human decision making. Peak.ai, a UK based start-up, raised US $75m in series C funding in August 2021 to enable it to build out its “decision intelligence” platform to expand into new markets and help non-tech companies make AI-based decisions.

Funny Story

AI argues for and against itself in Oxford Union debate: Megatron, an AI developed by Google and Nvidia, was given access to huge quantities of data to enable it to both defend and argue against the motion, “This house believes that AI will never be ethical”. It’s not clear which argument was more compelling!

What do you think about this bulletin? 

Please take the time to fill out our quick survey to let us know your thoughts about this bulletin, how it can be improved and what you would like to hear about going forward. 

What next? 

Our next Market Scan bulletin will be distributed in April 2022.