NAi Artificial Intelligence Model

AI Model for Analysis of NAi:

  1. Micro Financial Analysis:

Platform development comprise of building upon and securing fundamental smart contracts, added frameworks; such as a comprehensive oracle market, trading and management interfaces, service level app templates and customization tools, and integration with the future Ethereum substructure like state channels and stable coins. From a micro financial perspective, financial market could be highly influenced by the use of AI and machine learning in financial matters. In this section, possible changes in work pattern, inducement and pros and cons regarding financial stability are discussed.

Possible effects of AI and Machine learning on financial markets

To minimize the irregular flow of information, AI and Machine learning could play a vital role to strengthen the information function of financial system and improve the productivity of information processing. The means through which improvements could happen involves:

  • AI and machine learning may allow market participants to evaluate the information on a wider perspective, slowing them to understand the market value and how other components like sentiment analysis, can affect it. This will help in regularizing information effectively and bring stability in market.
  • AI and machine learning may lower market participation trading cost by enabling to balance trading and investment strategies in a smooth manner, through improving price discovery and minimizing the overall transaction cost in constantly changing market environment.

However, if many market participants use similar artificial intelligence and machine learning programs in areas such as credit rating or financial market activities, the resulting correlated risks can lead to financial stability risks. If machine-based operators outperform others, this could in the future lead many operators to adopt similar machine learning strategies (although this may also reduce the profitability of such strategies). Although there is no evidence that this is happening until now, this could become relevant with a greater adoption of these commercial strategies. As with any agricultural behavior in the market, this can magnify financial shocks. In addition, advanced optimization techniques and predictable behavioral models of automated trading strategies could be used by insiders or cybercriminals to manipulate market prices.

1.2 Possible effects of AI and machine learning on financial institutions

AI and machine learning is capable of increasing profitability and productivity while reducing the risk factor and minimizing the cost in financial institutes. This may lead to stable processing of system:

  • AI and machine learning may increase machine-based processing of several operation which will generate more revenue using less cost in financial institutes. For instance, if IA and machine learning is used to identify the potential customers’ needs or alter products according to targeted costumers, this will help in generating a wide range of customers and a market space to grow more. Comparatively less cost will be required for automated routine process in business matters.
  • AI and machine learning could predict or aware about the possible risks. Risk management through AI and machine learning can reduce risk by making decision based on previous records and comparing the prices of assets, enabling financial institutes to handle risks in a better way. It could also anticipate other suspicious activities such as frauds, cyber-attacks, mysterious transactions or other defaults. This will aid in better risk managements, using the tools to reduce tail risks in a system overall. Whereas, it might be unable to detect new types of risk due to rapid advancement in financial sector, while it keeps detecting accordingly to old risk analysis methods, missing out the new ways of possible fraud.
  • The data strength and open source nature of AI and machine learning can facilitate collaboration between financial institutions and other industries such as e-commerce and economy sharing businesses.

Despite this, risk management through AI and machine learning may create issues. Specifically, when human users at financial institutes try to analyze how decision making regarding investments and trading is being done. It could be challenging for the humans to comprehend the communication methods used by these applications. This may lead to users shutting off system manually and turning it on again will be possible only if other users across market turn it on in coordination. This could thus add to existing risks of system-wide stress and the need for appropriate circuit-breakers.

Moreover, in case of loss in AI and machine learning based decisions, it is hard to identify who is to be held accountable. For example, if loss occur due to AI and machine learning application built by third party who is to be held responsible? Institution who carried out the trading or the third party that built the application? Could the nature of supervision be affected if AI and machine learning is used widely even by the non-traditional market players? Moreover, several questions arise among collusion of trading applications that rely on deep learning. Specifically, if algorithms interact in ways that would be considered collusion if done by human agents, then as with human agents, proof of intent may be an issue. In this light, there may be a number of legal uncertainties.

Financial institutes may face greater risk due to unreliability of governance system of AI and machine learning. Investors need to fully understand the application and their possible tail risk to avoid uncalculated risks. However, financial institutes need to establish well-designed governance system and maintain auditability when installing the AI and machine learning applications. Which will assist in keeping out the additional cost for allocating possible risks due to uncertain governance system.

Finally, there may be important third-party dependencies. In the development of AI and machine learning to date there is a high reliance on a relatively small number of third-party technological developers and service providers. This third-party reliance could be relevant for market participants and financial institutions in the future. For instance, if a major provider of AI and machine learning tools were to become insolvent or suffer an operational risk event, this could lead to operational disruptions at a large number of financial institutions at the same time. These risks may become more important in the future if AI and machine learning are used for ‘mission-critical’ applications of financial institutions.

1.3 Possible effects of AI and machine learning on consumers and investors

By increasing the productivity of financial services and reducing the cost of AI and machine learning, customers could be immensely benefitted.

  • Investors and customers could benefit more by spending less on AI and machine learning financial services if borrowing cost and fees is reduced.
  • Financial services could be widely accessed by customers and investors. For example, applications of AI for robo-advice might facilitate people’s use of various asset markets for their investments. In addition to this, Ai and machine learning could make more sources of funds for consumers and small and medium enterprises (SMEs) by advance credit scoring for FinTech lending.

 

  • AI and machine learning could alter the financial services as per the need of consumers, making it more personalized through big data analytics. For example, AI and machine learning facilitate in analyzing big data, which could help in providing services as per the needs of customers according the pattern shown of consumers’ needs. The issue of information breaching of customers could be raised as AI and machine learning analytics analyze every customers’ characteristics through public data. Therefore, special measures should be taken to protect the output of customers’ analysis, allowing the safe and secure use of big data for efficient services. It further emphasizes on developing well-designed governance system for financial services delivered by AI and machine learning.

There are chances of AI and machine learning algorithms creating sensitive data information such as race, religion, gender etc. based on other provided information, on its own. There is a continuous research on how to overcome this shortcoming. This is a key area in the broader discussion on AI ethics (see annex B).

1.4 Current regulatory considerations regarding the use of AI and machine learning
Being relatively new, Ai and machine learning applications have no international standards. Still few international standard setters have noticed the risks associated with algorithm trading that may expose systemic risk. Examples include the following:

  • IOSCO (The International Organization of Securities Commissions) have made recommendations based on effect of newly introduced technologies on market surveillance, including algorithm trading to include data collection and cooperate across the border.

 

  • SSG (Senior Supervisors’ Group) which consists of senior representatives of supervisory authorities from around the world, have given out outlines to supervisors to consider when assessing practices and key controlling algorithm trading activities at banks.

According to national regulators, firms developing algorithm models based on AI and machine learning should identify potential risks at the developing stage to avoid market abuse or causing any kind of disruption in market values. This requirement is part of MiFID II, which will come into force in the first quarter of 2018 in Europe. US securities self-regulatory organizations have also implemented such requirements for algorithm model development.

BCBS (Basel Committee on Banking Supervision) has also emphasized on the safe and secure production of algorithm model that not only meet the firm’s internal policies and procedure, but also fulfills consumers’ requirements and is consistent with risk management and behavioral expectations of the firm. In order to support new model choices, firms should be able to demonstrate developmental evidence of theoretical construction; behavioral characteristics and key assumptions; types and use of input data; numerical analysis routines and specified mathematical calculations; and code writing language and protocols. It also advices to keep check on models during the development process of it.

  1. Macro Financial Analysis:

Financial system can be highly affected with the extensive adaptation of AI and machine learning in different ways, depending on the applications being used. AI and machine learning applications can escalate the economy growth through following means:

  • Enhance the efficiency of financial services: AI and machine learning can aid in better information processing on the fundamental values of assets, with effectual granting of funds to investors and projects. Aggregate system could also be benefitted by improved risk management of banks’ loan portfolio and insurers’ liabilities. Furthermore, increased speed and minimal costs of payment and transactions through AI and machine learning, i.e. perform transactions at times when counterparties are available with the corresponding demand, will increase transactions for real economic activities.

 

  • Facilitating collaboration and realizing new ‘economies of scope:’ AI and machine learning will help collaborate between financial services and other industries like ecommerce. Along with realizing new economies of scope and further nurture the economic growth. Like cooperation between ecommerce and financial services will advance through the customers’ analysis based on their transactions and payments activities.

 

  • Stimulating investments in AI and machine learning related areas: increasing interest of many firms to apply AI and machine learning to their business will further grow the economy through the investments on installation of this system.

From a macro-financial point of view, AI and machine learning have both pro and cons, and can affect the systematic importance of market participation degree of concentration and market vulnerabilities.

2.1 Market concentration and systemic importance of institutions

Concentration in financial markets may be affected by the adaptation of Ai and machine learning, as when concentration of financial services may be increased by third party that provides the Ai and machine learning services. Also, access to big data could be beneficial to firms to obtain substantial economies of scope. Lastly, large part of market will have access to advanced technology at affordable costs required for the installation and maintenance of the system.

With respect to banks’ systemic importance, if Ai and machine learning offer new set of financial services and unbundle the traditional banking system, banks then may focus more on narrow set of activities rather than running their universal services. However, taken as a group, universal banks’ vulnerability to systemic shocks may grow if they increasingly depend on similar algorithms or data streams. Simultaneously, if AI and machine learning is adopted by a bank that already enjoys public trust, to further empower their market value, its systematic importance may increase. Competitors providing same services on competitive price may affect the market entry cost and regulations, making it difficult to identify whether it is increasing or decreasing the degree of concentration.

2.2 Potential Market Vulnerabilities

Traditional trading could be affected through Ai and machine learning adoption. Financial stability could also be benefited by the introduction of different trading application by other market competitors. For instance, if an individual receives advise from robot-advisory which is powered by AI and machine learning, his activities would rather be more personal and customized rather than the traditional strategies set in the market. Investments in capital market could be increased by reducing the barriers for the retail consumers willing to invest. Similarly, the use of AI and machine learning for new and uncorrelated trading strategies by hedge funds could also result in greater diversity in market movements. More efficient processing of information could help to reduce price misalignment earlier and hence mitigate the build-up of macro-financial price imbalances.

Trading algorithm based on machine learning are capable of giving unexpected results rather than the rule-based application which follow the same pattern. This will allow firms using AI and machine learning to generate more revenue or reduce the trading cost, making Ai and machine learning is costlier. Unavailability of market-wide used data, AI and machine learning models could associate to market movement, leading to hindrance of market shock interpretations.

Regarding leverage, liquidity, and maturity transformation, the adoption of AI and machine learning by financial market participants such as hedge funds and market makers may also have both positive and negative impacts. AI and machine learning could increase liquidity in financial markets through enhanced speed and efficiency of trading activities.

AI and machine learning could be used to detect excessive risks and overly-complicated transactions and to design more effective hedging strategies for risk management by individual financial institutions. To the extent these tools enable the growth of new credit platforms to directly connect lenders and borrowers (broadly called FinTech credit), this could reduce reliance on bank loans, reduce banks’ leverage, and achieve a more diversified risk-sharing structure in the overall financial system. On the other hand, to the extent that market participants use AI and machine learning in order to minimize capital or margins or maximize expected returns on capital (within the constraints of regulations, and without paying due attention to risks), the use of AI and machine learning may increase risks. Specifically, it may allow for much tighter liquidity buffers, higher leverage, and faster maturity transformation than in cases where AI and machine learning had not been used for such optimization.

2.3 Network & Interconnectedness

AI and machine learning can increase the interconnectedness of financial markets and institutions in unexpected ways. Institutions’ ability to make use of big data from new sources may lead to greater dependencies on previously unrelated macroeconomic variables and financial market prices, including various non-financial corporate sectors (e-commerce, sharing economy, etc.). If institutes find an algorithm that produces uncorrelated profits, there are chances of it getting exploited on large level that inter-relatedness might really increase. Such unpredictable interconnections will only be identified once the technology is in use.

The interconnectedness in financial systems may let them share risks or minimize its effects. At the same time, it could give an extreme shock. Collective adaptation of AI and machine learning could bring new risks along. When financial institutions’ critical segments are dependent on same data sources and algorithm strategies, then under a specific change in market condition or new strategy that could exploit widely used algorithm strategy, can damage that segment alike, even if these segments are made up of hundreds or thousands of individual financial institutes.

2.4. Other implications of AI and machine learning applications

Moral hazard and adverse selection are inherited problems in insurance market which could be reduced with AI and machine learning applications. AI and machine learning may aid in reducing the moral hazard by continuously changing the insurance fee in accordance with the changing conditions of the policyholders. Adverse selection could also be decreased if AI and machine learning are used to provide customized insurance policies to each individual by analyzing their characteristics. This could also create challenging situations. For example, the more accurate pricing of risk may lead to higher premiums for riskier consumers (such as in health insurance for individuals with a genetic predisposition to certain diseases) and could even price some individuals out of the market. Even if innovative insurance pricing models are based on large data sets and numerous variables, algorithms can entail biases that can lead to non-desirable discrimination and even reinforce human prejudices. This has raised the discussion of how algorithms are developed, the desired extent of risk sharing and what information is permissible. AI and machine learning could help in financial institutions (RegTech) as well as supervisors (SupTech). Many of the uses described in section 3.4 could result in improvements in risk management, compliance, and systemic risk monitoring, while potentially reducing regulatory burdens.

Moreover, if AI and machine learning applications are adopted without proper training or means of receiving feedback, it could create new risks. For example, when AI and machine learning models are used in stress testing without adequate lengthy and diverse time series or required feedback in actual stress events, there are chances that consumers might not be able to identify possible institution-specific and systematic risks. These risks may be more prominent if AI and machine learning are used without the required knowledge of concealed methods and limits.

Over and above that, AI and machine learning techniques cannot be fully utilized unless the existing regulatory system is not revised accordingly to benefit from these techniques. Like in MiFID II, a firm is expected to submit a report when a notable event takes place, regulatory compliance is expected every time. In such case if AI and machine learning is used to identify whether the particular event should be reported or not, even if tools can point out the needed information for regulatory that could minimize the disruption in market, there are still high chances of disturbance in regulatory actions. In this case, more efficient results could be achieved if AI and machine learning is combined with human judgements and other tools to analyze the situation. More generally, the greater adoption of AI, machine learning, and other technological advances in finance may benefit also from more of a ‘systems’ perspective in financial regulation to contribute to financial stability in an increasingly complex system.

  1. AI and machine learning in regulatory compliance and supervision

AI and machine learning techniques are being used by regulated institutions for regulatory compliance, and by authorities for supervision. RegTech is often regarded as the subset of FinTech that focus on facilitating regulatory compliance more efficiently and effectively than existing capabilities. The total RegTech market is expected to reach $6.45 billion by 2020, growing at a compound annual growth rate (CAGR) of 76%. SupTech is the use of these technologies by public sector regulators and supervisors. Ai and machine learning is used by SupTech to increase the efficiency and productivity of supervision and surveillance. These two applications, RegTech and SupTech are discussed below and some of the examples used are from academic sector. These might not yet be used by regulatory or supervisory bodies, but there is a potential market for it in this sector. The use cases are grouped by the function for which they are used, namely regulatory compliance; regulatory reporting and data quality; monetary policy and systemic risk analysis; and surveillance and fraud detection.

3.1.1. RegTech: Applications by financial institutions for regulatory compliance

Machine learning when, combined with NLP, can be used by RegTech to analyze the unstructured data. Combination of machine learning with NLP may not only help in monitoring the behavior or applied to transparent communication of traders but can also interpret data inputs like from emails, instant messages, voice commands, documents, and metadata. This, in turn, begs the issue of the boundaries for the employee surveillance policy. Some of the regulatory institutions are also experimenting to be able to seek suitable product requirements through this.

NLP could be used by asset management firms to cope with new regulations. In the EU, investment managers have to comply with specific requirements in the Markets in Financial Instruments Directive (MiFID II), the Undertakings for Collective Investments in Transferrable Securities (UCITS) Directive, and the Alternative Investment Fund Managers Directive (AIFMD). Firms could potentially leverage NLP and other machine learning tools to interpret these regulations into a common language. Later they can analyze and make rules for the automation into the integrated risk and reporting system to let the firm obey the set rules and regulations. This will help in cutting down the cost, using minimal efforts in a

Comparatively less time to interpret and implement new and upgraded rules for fund management.

‘Know your customer’ or KYC is another area in financial industry where AI and machine learning are being used to make consumers’ and regulators’ experience better. In many sectors, the KYC process is often uneconomical, requires effort and is immensely duplicative. Machine learning is now rapidly being used in remote KYC financial services of firms to help them identify and pre-check their customers’ backgrounds. There are two basic ways in which it is used, firstly to match images from previous documents with the latest and secondly by calculating risk scores to let firms decide which of the applicants require to receive additional scrutiny. Machine learning-based risk scores are also used in ongoing periodic checks based on public and other data sources, such as police registers of offenders and social media services. These sources will allow to assess the risks and trust at lower cost in less time.

Firms can use risk scores on the probability of customers raising “red flags” on KYC checks to make a decision, either proceed with normal time and cost of full background check. Whereas some of the financial services have avoided using these tools because of the doubtful accuracy.

3.1.2. Uses for macro prudential surveillance and data quality assurance

Macro prudential can be improved by automating its analysis and data quality assurance using the Ai and machine learning methods. A series of new reporting requirements across jurisdictions has led to a greater volume and frequency of reported data, as well as greater resources required from financial institutions to complete reporting on time. In some cases (for example, transactions data in MiFID, AIFMD templates, etc.), data received by the authorities can be challenging, such that it cannot be used in its full potential via traditional methods. In addition to this, there are more prevailing chances of certain data qualities could be compromised such unfilled fields, substantial errors or other issues in new datasets. For this, additional tools will be required to assure the data quality. To prevent this, machine learning can be brought into use as it can identify in real time, the potential errors and flag them to the data-providing sources and/or statistician, enabling an economical, more effective, efficient, and quick data processing and macro prudential surveillance of data by the concerned authorities.

According to authorities, low data quality issue is one of the main challenges to make full use of TR data that needs to be addressed. AI and machine learning can help trade repositories (TRs) to ensure their data quality and add value of TR data to authorities as well as the public. The adaptation of machine learning techniques could assist TRs – for over-the-counter (OTC) derivatives or (where applicable) other types of transactions, such as exchange-traded derivatives or securities financing transactions – raise the quality of data. Adequately trained machines learning algorithms could be used to enhance data quality as they can identify data gaps, inconsistent data, fat-finger errors and also detect the duplication of data or fill-in the missing data. In this regard, the Autorite des marches financiers du Québec reports that it has successfully tested in its FinTech Laboratory that a supervised machine learning algorithm is capable of recognizing the distinct categories from unstructured free text fields in OTC derivatives data, such as the floating leg of swaps. Based on this algorithm, alerts implementations are on its way that allow to detect transactions which are not in accordance with obligatory clearing requirements.

3.1.3. SupTech: Uses and potential uses by central banks and prudential authorities

Machine learning can be applied to systemic risk identification and risk propagation channels. Especially NLP can help the authorities in various ways to detect, measure, foresee, and expect the market volatility, financial stress, unemployment, housing rates, and liquidity risks. . In a recent Banca d’Italia (BDL) study, still in progress, textual sentiment derived from Twitter posts is used as a proxy for the time-varying retail depositors’ trust in banks. The indicator is used to challenge the predictions of a banks’ retail funding model, and to try to capture possible threats to financial stability deriving from an increase of public distrust in the banking system. Moreover, at BDL, the required information is extracted from the newspaper articles available on the internet by processing them through suitable NLP pipeline to access their opinions. As per another research, academics developed a model using computational linguistics and probabilistic approaches to uncover semantics of natural language in mandatory US bank disclosures. This model was capable of finding risks in interest rates, capital requirements, real estate, mortgages, rating agencies and marketable securities of even back in 2005. Further studies can foresee and figure out the market outcomes and economic condition, its volatility as well as growth.

Supervisors are able to see the patterns of huge and complex data when using machine learning along with NLP. The combination of NLP with machine learning will let link trading database with other information of market participants. This will let to integrate and compare trading activity information with behavioral data like communications and to compare normal trading scenarios with those that may have substantial deviations, triggering the need for further analysis.

According to a survey conducted in 2015 regarding the central banks’ usage of big data, that it is expected that the interest of central banks in big data will increase for the sake of financial stability and macroeconomic purposes. It was expected in the survey that big data would be used to forecast the economic conditions, in regards to prices increase and decrease and possible inflation. . For instance, 39% of central banks expect to ‘now cast,’ or predict in real time, retail home prices using big data. AI will make it possible to predict the GDP, unemployment, industrial production, tourism activities, business cycle and other economy related conditions using the now cast and sentiment indicating techniques.

Researches have been carried out to highlight how Ai and machine learning can be used. According to a recent research at Columbia University, researchers combined the observational study with the newly advanced machine learning to let public authorities and market players (i) ‘score’ policy choices and link them to indicators of financial sector performance; (ii) simulate the impact of policies under varying economic and political conditions; and (iii) detect the rate of change of market innovation by comparing trends of policy efficacy over time. A study from the BDL employs a dynamic factor model and utilizes a dataset containing variables from different sectors of the economy with the intent of analyzing the re-distributive impacts of financial arrangement over various regions. In order to select the statistically most relevant independent variables they use automatic regression variable selection. According to another research being conducted at the Office of Financial Research (OFR), researchers are trying to look into the most popular financial innovations which are drawing a large number of market participants in financial publications. Researchers at OFR are also trying to analyze the correlation of news, attention and financial stability using machine learning to extract sentiment and key topics from financial publications.

3.1.4. Uses by market regulators for surveillance and fraud detection

AI is being used by regulators to identify frauds and AML/CFT detection. The Australian Securities and Investments Commission (ASIC) has been exploring the quality of results and potential use of NLP technology to identify and extract entities of interest from evidentiary documents. NLP along with other technologies is being used by ASIC to visualize and detect the extracted entities and the relationship among them. BDL has tried to overcome the banking crimes like money laundering by correlating the information from newspapers articles with the information extracted from bank transfers, which includes both structured and unstructured data. It is time taking as well as difficult to pin point the suspicious transaction. Machine learning can make it easier to highlight the unusual transactions as well as identify the complex patterns. Monitory Authority of Singapore (MAS) has been trying to use AI and machine learning to identify the suspicious banking transactions, letting supervisors focus more on the transactions which could possibly be riskier. Machine learning could be used to extract different types of data about transactions, user’s profile and other but it is also expected that machine learning can aid to identify other unusual banking patterns, the mysterious transaction, possible funding to terrorism which might not be effectively investigated otherwise.

Disclosure and risk assessment could be done by market regulators through AI and machine learning. The US Securities and Exchange Commission (SEC) staff leverages “big data” to develop text analytics and machine learning algorithms to detect possible fraud and misconduct. Assessment tools are now also being included into Ai as SEC staff has been using machine learning to identity pattern in SEC files’ text. With supervised learning, risks in investment manager files can be identifies by comparing the pattern of latest with the previous examination results. According to SEC staff, these newly introduced techniques are five times more reliable than rand to finding a language that merits a referral to enforcement. While the outcomes can create false positives that can be clarified by non-accursed activities and purpose, these, in any case, give progressively vital signs to grade the examination. For investment advisers, the SEC staff compiles structured and unstructured data. Unsupervised learning calculations are utilized to recognize one of a kind or outlier reporting behaviors – including both points demonstrating and tonality investigation. Results from the first stage are then compared with previous exams, leading to second stage where machine learning algorithm anticipate the presence of idiosyncratic risks at every investment advisor. In Australia, ASIC has also used machine learning software to identify misleading marketing in a sub-sector, such as unlicensed accountants in the provision of financial advice.

  1. Conclusion and implications for financial stability

Financial market provision is transforming with the advancement of AI and machine learning. Though data on the degree of selection in different markets is limited, but market participants’ communications show that certain part of the financial system is rapidly adopting the AI and machine learning, out casting the adoption of Fintech innovations, like distributed ledger technology or smart contracts. Applications such as for detecting frauds, managing portfolios or for customized capital are being extensively used, and expected to grow even further. This indicates that more focus should be put into financial stability beforehand. Moreover, many of the changes will not result in a material change to financial stability and hence fall outside the scope of this report.

Through AI and machine learning, financial services could be approved by providing more efficiency, effectiveness and regulatory and systemic risk surveillance. To further strengthen the financial system, proficient handling of data on credit risk and reduced cost for user interaction can build a strong foundation. AI and machine learning is capable of providing services such as to detect frauds, efficient risk management in low budget. In portfolio management, the more efficient processing of information from AI and machine learning applications could help to boost the efficiency and resilience of financial markets – reducing price misalignments earlier and (under benign assumptions) reducing crowded trades. Lastly, AI and machine learning have enhanced the productivity of supervisory and an improved system to analyze and mange risks in financial markets.

Due to vast usage with effectiveness, AI and machine learning are prone to increase the dependency on third party, giving rise to additional systematically vital players. Several large firms are providing the AI and machine learning services but likewise, there could be a great demand of third-parties to provide these services of same level as by the leading firms. There is the potential for natural monopolies or oligopolies. From economy efficiency point of view, these rivalry issues could give rise to financial instability if and when such technology firms have a large market share in specific financial market segments.

Existing AI and machine learning tool providers may fall outside the administrative boarder or may not be acquainted with applicable law and regulation. Where financial institutions rely on third-party providers of AI and machine learning services for critical functions, and rules on outsourcing may not be in place or not be understood. These servicers and suppliers may not be liable to supervision and oversight. Similarly, if suppliers of such tools start giving financial services to institutional or retail customers, this could involve financial services occurring outside of the administrative limits.

Models designed based on AI and machine learning techniques could be hard or at times impossible to interpret. Inability to interpret the AI and machine learning methods could add to macro-level risk if not appropriately supervised by micro prudential supervisors. This inability to interpret the models might be ignored in several situations as in, if the model’s performance exceeds that of more interpretable models. The effects of interpretability cannot be determined easily except in balance sheet of the firm, for example during a systemic shock. The models built using AI and machine learning might not be trained as they needed to be, in period of high volatility. This could affect the outcome of models as they might be unable to suggest possible actions to avoid financial crisis or aid in better risk management.

Widespread use of models based on AI and machine learning techniques could give out unexpected consequences, possibly consequences that could give negative effect on financial system. For Example, if a firm designs a model based on AI and machine learning but doesn’t understand its working due to its complex nature, it will be difficult for the firm or the supervisors to anticipate the possible actions made by the model and its consequences on the market. Similarly, there might be other unexpected consequences in application that would slowly affect the built-up as in detecting cyber threats, optimizing capital and credit scoring.

It is important at this point to check the AI and machine learning applications for risks, how they safeguard the date, handle risks and take actions regarding cybersecurity. There is also a need to further enhance the AI and machine learning applications, making it easier to interpret the decision and outputs of the algorithm. Special efforts are needed to make interpretations of AI and machine learning easier, for better risk management and also to enjoy great trust by the consumers, regulators, and supervisors in critical financial services. The complexity of the models may create problem for the users and developers to explain it and its working.

The use of AI and machine learning could further grow as there is a potential market to adopt the technology. AI and machine learning should be further monitored and polished to enable its usage in the other sectors than the particular cases mentioned in this paper.

NAi Technologies Europe plans to apply Artificial Intelligence capabilities on BlockChain to macro-economic phenomenons, as well, in addition to the financial markets. In the second phase of development, Gross Domestic Process (GDP), forecast & growth or decline rate predictions based on 100’s of years of data available for the countries will also be done. Our application will also be able to predict future unemployment rate, industrial productions, business cycles and other factors creating an impact on global and domestic economies of each country.