Shulou(Shulou.com)06/03 Report--

Author | Nielsen translator | Summer Night planner | Tina artificial intelligence creates great opportunities for businesses and organizations around the world, and those enterprises and organizations that use artificial intelligence can respond to future challenges and opportunities faster and better.

At Nielsen, we continue to rely heavily on artificial intelligence to control the power of data, but this is only one of the ways that artificial intelligence helps our organization. As companies explore new ways to use artificial intelligence, they need to ensure that any associated risks are managed during artificial intelligence development. This is why our internal audit function has adopted an audit process for artificial intelligence. Organizations must remain vigilant and ensure that they make the right investments in the right place, so as to take advantage of the opportunities represented by artificial intelligence and ensure that risks associated with artificial intelligence are accurately assessed. ensure that artificial intelligence projects are consistent with organizational mission, values, culture, and larger technology strategies.

In this regard, internal audit should be a valuable ally, but from an audit point of view, artificial intelligence brings many challenges. First of all, the definition of artificial intelligence is somewhat vague. At the same time, in a wide range of business activities, organizations are actively seeking to develop or acquire artificial intelligence capabilities, artificial intelligence programs and processes may be very complex and highly technical. Finally, there is a lot of hype on the market about what artificial intelligence is and what artificial intelligence can do.

The practice of auditing artificial intelligence is developing. As for advanced artificial intelligence technologies, such as machine learning, there is no unified standard to regulate these artificial intelligence technologies. These standards, as well as a framework for auditing artificial intelligence technologies, are under preparation. The Institute of Information Systems Audit and Control (ISACA:Information Systems Audit and Control Association) has issued guidelines for applying its existing COBIT framework (for regulating enterprise-level IT) to artificial intelligence, and the Internal Audit Institute has also proposed an approach. At the same time, a number of organizations, including NIST,ISO and ASTM, are drafting artificial intelligence standards. However, with the continuous development of artificial intelligence audit practice, it is clear that internal audit needs to continue to clearly focus on strategy and regulation-to ensure that the organization has a reasonable artificial intelligence strategy. And there is a robust regulatory structure to support the implementation of the strategy.

Define artificial intelligence

If there is no general consensus on what is artificial intelligence and what type of technology should be considered artificial intelligence technology, it is difficult to establish standards for artificial intelligence audit. People have different views not only on what artificial intelligence is, but also on how long artificial intelligence has been around. Some people have adopted a broad view, believing that artificial intelligence appeared earlier than most people think. Others have a narrower view that artificial intelligence does not actually exist, and that the so-called artificial intelligence technology is actually misunderstood.

Technologies that are generally considered to belong to the field of artificial intelligence include: deep learning, machine learning, image recognition, natural language processing, cognitive computing, intelligence enhancement, machine intelligence enhancement, enhanced intelligence and so on. However, there are differences among various sources on how to classify and describe these technologies. For example, some people think that robotic process automation (RPA:Robotic Process Automation) is a form of artificial intelligence because it can execute highly complex algorithms on a variety of large data sets, and this kind of program makes decisions in a way that makes it look like it simulates intelligence. On the other hand, others, including ISACA, disagree, arguing that "artificial intelligence does not run based on a set of predefined rules" and that this rule-based function is "the hallmark of traditional software engineering".

Furthermore, artificial intelligence itself can be subdivided according to the type of technology. For example, in 2017, the American Institute of Internal auditors (IIA:Institute of Internal Auditors) released a global perspective and insights report (Global Perspectives and Insights) on the theme of artificial intelligence audit, which cites four categories of artificial intelligence (namely: I-reactive machine, II- finite memory, III- mental theory, IV- self-awareness). Meanwhile, in a 2019 study, Protiviti chose to combine machine learning, deep learning and natural language processing into a group and collectively referred to these technologies as "advanced artificial intelligence".

Exactly what artificial intelligence is, this may seem like a topic of debate among opinionated people, but it is actually related to the value delivered by artificial intelligence internal audit and audit functions, as well as meeting the expectations of stakeholders. When considering what kind of artificial intelligence technology needs to pass internal audit, the most important consideration is to understand how an organization defines artificial intelligence. By initiating an open dialogue proactively, internal auditors can answer the question: as an organization, what do we mean when we talk about "artificial intelligence"? Obviously, the consistency of answers to this question is very important for the expectations of management stakeholders, because this question involves the scope of artificial intelligence audit. However, the answer to this question can also provide insight into whether an organization's definition of artificial intelligence is broad enough-or narrow enough-to perceive risks in the market.

Audit artificial intelligence

Although people have different views on the nature of artificial intelligence, when it comes to the practice of artificial intelligence audit, the mainstream guiding ideology is that internal audit should still focus on strategy and supervision.

Strategy

Without a clear and periodically reviewed strategy, investments in artificial intelligence capabilities will only produce disappointing results, or even worse, they may lead to financial and / or reputational damage to the organization. The internal audit should be interested in confirming whether the organization has documented an artificial intelligence strategy and assessing the power of the strategy based on the following considerations:

Does it clearly describe the expected results of artificial intelligence activities?

Does it include plans to identify and address artificial intelligence threats and opportunities?

Is it developed in collaboration between business executives and technical executives?

Is it consistent / compatible with the mission, values and culture of the organization?

Will it be reviewed and updated?

Does it take into account the support capabilities required to take advantage of artificial intelligence technology?

Organizations need their internal audit department to ask these types of questions, more than once, and repeatedly. Research continues to show that organizations want their internal audit departments to be more forward-looking and provide more value, as this involves assessing strategic risks. When talking about support capabilities, it should be noted that in 2019 board members and C-level leaders (such as CEO, CTO, etc.) are most concerned that their existing operations and infrastructure will not be adjusted to achieve the expected performance among "gifted digital" competitors. Therefore, internal auditors can and should constantly address the question of whether the organization's artificial intelligence strategy is appropriate and whether it is actually enforceable.

Supervision and control

Like any other major system, organizations need to establish appropriate regulatory systems for artificial intelligence activities, ensure that organizations have proper control and responsibility for artificial intelligence, and determine whether artificial intelligence projects are working as expected and whether the goals of artificial intelligence projects have been achieved.

Similarly, there is no reliable template to manage artificial intelligence. "the script hasn't been written yet," one expert wrote. However, internal auditors should start with big data and explore the caution of business executives in developing a robust regulatory system to support artificial intelligence applications.

Big data forms the basic capabilities of artificial intelligence, which means that organizations should pay more attention to their data regulatory system. Internal auditors should understand how organizations ensure that their data infrastructure is capable of adapting to the scale and complexity of artificial intelligence activities (which were previously set in artificial intelligence strategies). At the same time, internal auditors should understand how organizations manage risks to data quality and consistency, including data collection controls, access rights, data preservation, classification (i.e. naming), editing and processing rules. Internal audits should also consider security, network resilience, and business continuity, and assess the extent to which organizations are prepared to deal with data threats, including threats to data accuracy, completeness, and availability.

The value and performance of artificial intelligence also depend on the quality of algorithms and computational accuracy. Artificial intelligence applies these algorithms to big data. Organizations must have documented methods for algorithm development and quality control to ensure that these algorithms are written correctly, that the results are unbiased, and that data is used appropriately. Internal auditors should also understand how to verify the correctness of artificial intelligence system decisions and whether they can be defended when they are challenged.

In addition to regulatory data and artificial intelligence algorithms, the regulatory system should also be checked to determine whether it meets the following:

Obligations, responsibilities and supervision mechanisms have been clearly established.

Policies and procedures are properly documented and followed.

Those responsible for artificial intelligence have the necessary skills and expertise.

Artificial intelligence activities and decisions and behaviors related to artificial intelligence are in line with the values, ethics, social and legal responsibilities of the organization.

Any supplier accepts the third-party risk management process.

Audit department

In order to audit artificial intelligence effectively, internal audit functions must ensure that they have (or can acquire) sufficient resources, knowledge and skills to audit around artificial intelligence, even if auditors do not always have expert knowledge. This has proved to be very challenging. According to the 2018 North American Internal Audit Pulse (North American Pulse of Internal Audit), 78 per cent of chief audit executives said it was extremely difficult to recruit staff with data mining and analytical skills. In any case, internal audit departments should steadily increase their AI expertise through training and recruitment, as organizations' reliance on AI will only increase in the next few years.

The success of artificial intelligence audit does not directly depend on professional and technical knowledge. Success depends on the assessment of strategy, regulation, risk and process quality-all of which are good at internal auditing from an independent, cross-sectoral point of view. When it comes to artificial intelligence, internal audits should continue to focus on these things.

Now it's time to focus on artificial intelligence audit.

Although the market is changing rapidly and artificial intelligence is sometimes vaguely defined, internal audit can provide valuable, basic and reasonable guarantees to ensure that the organizations they serve invest in artificial intelligence in the right direction, take risks and opportunities into account when it comes to artificial intelligence, and give strong execution to business goals. The sooner internal auditors do this, the better, because all forms of artificial intelligence will not disappear. The momentum of artificial intelligence is rising. The internal audit function, like other business processes, should be accustomed to auditing and providing security around artificial intelligence, as more and more business areas will use artificial intelligence technology in some way in the coming years. While new technologies and risks emerge one after another, a focus on strategy and regulation will enable auditors to do so.

Original text link:

Https://medium.com/nielsen-forward/dont-let-the-race-to-embrace-ai-overshadow-the-needs-to-audit-your-advancements-a806a3e47259

Https://mp.weixin.qq.com/s/7JlMFZDved-N3oo_ZV7tMQ

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.