Inteligencia Artificial en la auditoria interna – Cómo Auditarla y Cómo Usarla para auditar

Introduction to Artificial Intelligence in Internal Auditing

Overview of the Session

• The session begins with a warm welcome, acknowledging the audience regardless of their time zone.

• Paola Andrea González, an experienced auditor and technology consultant from Colombia, is introduced as the speaker for this special course on AI in internal auditing.

• Paola expresses her excitement about discussing AI's role in internal auditing and emphasizes its growing necessity.

Understanding Digital Evolution vs. Disruption

• The session aims to provide an overview of AI concepts related to internal auditing, focusing on strategic applications rather than technical details.

• Paola engages the audience by asking whether they perceive recent changes as digital evolution or disruption, prompting responses via chat.

• The consensus among participants leans towards viewing these changes as digital evolution, highlighting gradual advancements over time.

Examples of Digital Evolution

• Paola illustrates digital evolution using mobile phones' progression over years, emphasizing that change occurs gradually rather than abruptly.

• She contrasts this with digital disruption, which represents sudden and significant shifts that challenge existing processes and practices.

Implications of Digital Disruption

• An example is provided regarding the transition from candles to electric lighting; this shift required new skills and knowledge for those previously working with candles.

• This analogy underscores how disruptive technologies can render certain professions obsolete while creating demand for new expertise.

Current Risks in Organizations

• Reference is made to a report from Risc Focus Latin America 2024 by the Latin American Foundation of Internal Auditors, indicating that digital disruption ranks fifth among current organizational risks.

Understanding the Impact of Digital Disruption

The Necessity of Addressing Digital Risks

• The rapid pace of technological change necessitates that organizations take responsibility for the risks associated with digital disruption.

• Increased computing power and storage capacity lead to more information being processed, making it essential for audit functions to adapt strategically.

Transformation in Internal Auditing

• Internal auditing must evolve to meet new business model needs in increasingly strategic environments.

• Emphasis on using artificial intelligence (AI) is crucial for driving necessary transformations within organizations.

Defining Artificial Intelligence

• The term "artificial intelligence" was coined in 1956, indicating its long-standing presence in technology discussions.

• AI aims to enable machines to solve problems and make decisions similarly to humans, though current capabilities are still limited.

Current State of AI Capabilities

• Generative AI exists but does not yet replicate all human decision-making processes; it serves specific purposes like text or image generation.

• There are significant limitations regarding AI's ability to perform all human functions or make comprehensive decisions autonomously.

Components of Artificial Intelligence

• AI encompasses various disciplines such as data analytics, big data, machine learning, deep learning, and robotics.

• Data analytics involves analyzing past data trends to inform future decisions, which is a fundamental aspect of AI applications.

Machine Learning and Its Applications

• Machine learning enables algorithms to learn from initial data inputs over time by recognizing patterns and improving their outputs based on successes and failures.

• Everyday applications like Netflix use machine learning algorithms for personalized recommendations based on user behavior.

Understanding Deep Learning and Robotics

• Deep learning mimics brain function through interconnected nodes or neurons, enhancing machine learning capabilities.

• Robotics can either utilize AI (e.g., humanoid robots capable of interaction and decision-making) or operate without it (e.g., simple robotic arms).

Integrating AI into Organizational Processes

Introduction to AI and Auditing

Overview of Automation and AI in Organizations

• The speaker discusses the automation of processes within organizations, emphasizing the importance of data in anticipating risk events.

• Introduction of artificial intelligence (AI) concepts, particularly focusing on auditing AI and how it can enhance audit processes.

Understanding Deep Learning

• Reference to IBM's Deep Blue, the first computer to defeat a chess champion, marking a significant moment in recognizing machine intelligence.

• Discussion on emotional aspects of AI; algorithms can make bold decisions without fear, unlike humans who may hesitate.

Chatbots and Their Functionality

• Explanation of chatbots like ChatGPT as AI-driven tools that simulate human interaction through pre-defined responses.

• Emphasis on the potential of ChatGPT, which utilizes advanced algorithms trained on specific datasets for effective communication.

The Paradigm Shift in Auditing

Changing Perspectives in Audit Practices

• The need for auditors to shift their mindset and adapt to new methodologies rather than relying solely on traditional practices.

• Highlighting issues with sampling data instead of utilizing complete datasets available through modern technology.

Leveraging Technology for Data Analysis

• Encouragement to use advanced tools for analyzing various forms of data (e.g., PDFs, recordings), moving beyond outdated methods.

• Mentioning an example involving transforming recordings into actionable insights through data analysis techniques.

Addressing Challenges with AI Algorithms

Overcoming Misconceptions about AI

• Clarification that AI algorithms operate within technological frameworks requiring security measures similar to any other technology.

Building Knowledge and Skills

• Importance of continuous learning and adapting one's perspective towards new information regarding AI applications in auditing.

Cultivating a New Culture in Auditing Teams

Fostering Change Within Organizations

Data Analysis and Auditing in Technology

Importance of Comprehensive Data Analysis

• Emphasizes that data analysis should not be intimidating; auditors should utilize tools to handle large volumes of data, moving away from sampling when possible.

• Highlights the necessity for auditors to learn data analysis tools, stressing the importance of IT skills across all auditor roles, whether financial or operational.

The Role of Technology in Business Processes

• Discusses how business processes are increasingly supported by technology, reducing manual processes even in fieldwork scenarios.

• Suggests that auditors need foundational courses in technology to understand key terms like databases and servers, which are essential for effective data analysis.

Balancing Technical and Soft Skills

• Stresses the need for a balance between technical skills and soft skills within audit teams to enhance teamwork and overall performance.

• Notes the emergence of specialized analytics subgroups within auditing departments, requiring collaboration between tech-savvy individuals and experienced auditors.

Understanding AI Auditing Principles

• Points out that auditing artificial intelligence requires knowledge of both auditing principles and technical aspects, as well as an understanding of ethical dilemmas associated with AI.

• Introduces the concept that AI technologies pose risks related to data quality, confidentiality, and ethical considerations.

Ethical Dilemmas in AI Development

• Discusses ethical challenges faced by algorithms trained for decision-making based on human-like judgments influenced by personal beliefs and values.

• Warns about potential biases introduced during algorithm training due to developers' backgrounds affecting AI outcomes.

Governance Around AI Developments

• Urges auditors to assess governance structures surrounding AI developments within organizations, similar to existing information security frameworks.

• Advocates for clear policies regarding AI development aligned with organizational ethics and accountability measures.

Exploring Ethical Dilemmas: The Trolley Problem

Dilemma of Autonomous Vehicles and Ethical Decision-Making

The Trolley Problem in Context

• The discussion begins with the classic ethical dilemma known as the trolley problem, where one must choose between saving five people or one by diverting a tram.

• Responses to this dilemma vary based on personal values and beliefs, highlighting that there is no universally correct answer.

• This moral quandary extends to autonomous vehicles, which are programmed with algorithms that embody certain ethical principles for decision-making.

Ethical Algorithms in Autonomous Vehicles

• If an autonomous vehicle faces a situation where it could crash into five people or swerve to hit one, it should prioritize the greater good by saving more lives.

• There is concern about how these vehicles are trained ethically and whether they align with societal values when making such decisions.

Auditing AI Algorithms

• Organizations need to audit their AI algorithms for biases related to ethics and ensure alignment with organizational values.

• Key aspects of auditing include understanding governance models for AI, accountability structures, and performance measurement of algorithms.

Data Usage and Organizational Policies

• With widespread access to AI tools like GPT, organizations must consider policies regarding data usage and potential information leaks.

• Employees may inadvertently share sensitive information while using generative AI tools; thus, awareness of risks is crucial.

Importance of Data Architecture

• Understanding the architecture behind data systems used in AI is essential for ensuring quality data input/output processes.

• Organizations should focus on auditing data sources, quality control measures, and decision-making processes influenced by this data.

Real-world Implications of AI Decisions

Understanding the Role of Artificial Intelligence in Auditing

Overview of AI in Organizations

• The discussion begins with an emphasis on the importance of auditing within organizations that are developing artificial intelligence (AI). It highlights the need for organizations to ask critical questions regarding their AI implementations.

Access and Risks Associated with AI

• All users in organizations now have access to interact with public cloud-based AI tools, such as ChatGPT. This accessibility raises concerns about associated risks, which have been formally recognized by organizations.

Regulatory Challenges

• There is a global effort to regulate AI, but progress is slow. For instance, Colombia struggles to define what constitutes a cybercrime, indicating that comprehensive regulation may take time.

The Future of Auditors with AI

• A key takeaway from a conference is that while AI will not replace auditors entirely, those who do not adopt AI will be replaced by those who do. This underscores the necessity for auditors to embrace technological advancements.

Continuous Learning and Adaptation

• The speaker stresses the importance of continuous learning and adaptation within the auditing profession as technology evolves. Staying updated on developments in AI is crucial for maintaining relevance in the field.

Benefits of Automation in Auditing

Routine Task Automation

• One significant benefit of integrating AI into auditing is automating routine tasks. This allows auditors to focus on more complex analyses rather than repetitive data checks.

Predictive Analysis Capabilities

• By utilizing historical data effectively, auditors can perform predictive analysis to identify potential risks before they materialize. This proactive approach enhances risk management strategies.

Key Risk Indicators (KRIs)

• Well-designed KRIs provide valuable insights into organizational risks. Automating these indicators enables real-time monitoring and alerts regarding potential issues like cash flow problems due to inadequate revenue collection.

Real-Time Deviation Detection

Understanding the Impact of Automation on Risk Management

The Role of Automation in Reducing Human Error

• Automation can significantly reduce human error by minimizing manual processes, thus adapting to emerging risks.

• New business decisions and technologies introduce unforeseen risks that require careful analysis beyond current known risks.

Monitoring Emerging Risks

• Continuous monitoring of processes is essential to identify new risks associated with changes in technology or organizational environments.

• Implementing artificial intelligence (AI) in auditing involves a structured process that includes skills development, cultural adaptation, and strategic planning.

Case Study: Technological Innovation in Auditing

• The example presented revolves around "Lacaton," a competition for technological innovation in internal auditing organized by FELAB.

• Notable participants include Banco de Chile and BCP, showcasing advancements using data analysis and AI.

Disruption in Traditional Processes

• A significant shift occurred when a bank transitioned from face-to-face debt renegotiation to phone-based contracts, impacting audit practices.

• This change led to a reduction from 18,000 physical contracts to 8,500 phone contracts, raising regulatory compliance concerns.

Addressing Audit Challenges with Technology

• The challenge arose as auditors needed to ensure compliance without being able to physically review all contracts due to the volume.

• A bot was developed that processes each contract quickly—taking only 4 minutes per contract—allowing for continuous monitoring and efficient audits.

Conclusion: Enhancing Audit Efficiency through AI

How to Implement an AI-Powered Solution for Contract Auditing

Overview of the Solution Stages

• The solution is divided into four stages: audio-to-text transcription using AI, data validation, comprehensive result visualization, and deviation notification.

Functionality of the Bot

• The bot transcribes telephone contracts and highlights parameters for comparison, such as subscription language and contract terms.

Process Steps in Detail

• The initial step involves audio transcription to text. This is followed by data transformation and validation before results are visualized and deviations are notified.

• Transcription utilizes cloud technology where recordings are uploaded for processing. This method is noted to be significantly more accurate than standard mobile dictation features.

Programming Aspects

• After transcription, Python programming defines key variables that must be present in the audio, focusing on regulatory compliance elements within verbal contracts.

Dashboard Insights

• A dashboard displays search results including client names, contract dates, terms, and rates. It helps identify non-compliance cases effectively.

Performance Measurement

• The bot's performance is monitored through average processing time per contract. This ensures ongoing evaluation of its efficiency.

Summary Reporting

• A summary table outlines processed contracts versus those with discrepancies. Observed cases are flagged for further review by auditors.

Adaptation to Business Changes

• The implementation resolves previous business process challenges faced by auditors adapting to new operational requirements.

Future Considerations

• Emphasis on continuous learning and adaptation within auditing processes is crucial for future automation efforts.

Strategic Planning for Technology Integration

• Internal audit teams need a strategic plan that incorporates technological advancements as organizational projects requiring time and resources.

Conclusion & Invitation for Further Learning

Introduction and Course Overview

Closing Remarks and Recommendations

• The speaker expresses gratitude to all attendees, indicating a transition towards concluding remarks and recommendations for the upcoming course.

• Emphasizes the expectation of seeing many participants in the virtual class, acknowledging that not everyone can engage in practical sessions due to size constraints.