i-AID Governance

Turning the tide on corruption in India with predictive governance policing and building a real time government auditing framework through the use of Artificial Intelligence^[1]

Background

The topic of Artificial Intelligence (AI) has been much debated, since as early as the 1950s when Alan Turing put forth the question of whether machines could think.^[2] Although the definition of AI has been moulded to fit the needs of all the stakeholders involved, in simpler terms, it refers to the utilization of digital technologies for creating systems capable of performance of tasks, which are thought to be requiring ‘intelligence’. AI has been often termed as the Fourth Industrial Revolution and is driven by machine learning, which is a sub-concept of AI, wherein the digital systems may improve the performance of a given task through experiential learning techniques.

A good governance system, both at a macroscopic as well as a microscopic level, thrives on an efficient, accountable, responsible and transparent administration. Due to the magnitude of India’s population and the related diversity, for various years, the Indian governance structure had been plagued with bureaucratic delays and archaic processes. The present government, which has been in power since 2014, brought with it an influx of policies revolving around technological innovation and utilizing the same in fostering effective governance mechanisms.

Corruption Perceptions Index 2020, Transparency International^[3]

However, the Indian policy and governance scenario pertaining to AI was highlighted in the recent years, wherein better algorithms, access to substantive amounts of data and a surge in the computational power was proliferated into the system machinery. The efficiency, effectiveness and accuracy of results promised by AI has resulted in making it an imperative ingredient in almost every sphere and sector of the society today. NITI Aayog^[4], in 2018, came up with ‘AI for All’^[5]– a discussion paper pertaining to India’s national strategy on AI. The strategy seeks to position India as a global AI leader by introducing and promoting the use of AI in these sectors: Education, Agriculture, Healthcare, Smart Cities, Smart mobility and transportation. This year, the national government has doubled its investment in its innovation program known as Digital India to Rs3,063 crore (or $477 million) in order to fund advances in AI, machine learning, and 3-D printing.^[6] An Israeli company called Cortica is working with the Best Group “to analyze the terabytes of data streaming from CCTV cameras in public areas. One of the goals is to improve safety in public places, such as city streets, bus stops, and train stations.” The firm is “looking for behavioural anomalies’’ that signal someone is about to commit a violent crime.”^[7]

Therefore, although a step has been initiated in the policy framework of introducing AI in the varied sectors of the governance mechanism in India, a lot is yet to be conceptualised and implemented in utilising AI for curbing corruption in the country. AI is still at its nascent stage in India and its potential has not yet been utilized fully.

Need for AI based Governance System

The recent trends in India pertaining to bribery and corruption highlights bribes being paid to government officials through cash or cash equivalent, meals and entertainment, gifts, travel expenses and/ or payment of personal expenses, services, golf (or any other) outings, charitable donations, medical treatment, loans or jobs/ internships for relatives/ friends. Since a lot of the government processes are still manual and bureaucratic, and require multiple government approvals before a project can be implemented, the chances and risks of bribery and corruption are insurmountable in such cases.

The Prevention of Corruption Act, 1988 (PCA) is one of the predominant anti-bribery anti-corruption legislations in India which criminalizes the receipt of any ‘undue advantage’^[8] by a ‘public servant’.^[9] Therefore, in accordance with the PCA and other relevant statutes and legislations for preventing bribery and corruption in India, the following broad key elements are required to be kept in mind:

1. Payment, offer, or promise
2. Anything of value
3. Foreign Official
4. Corrupt intent
5. Obtain or retain business

Often the risks and chances of bribery are extended to third parties, while applying for/ making low level payments for permits, licences, renewals, etc by entities. These may surface as brokerage charges or fees, consultancy charges, covering charges, management fees, documentation charges, managing expenses, out of pocket expenses, protection fees, special expenses, licensing fee, liaisoning fee, clearance charges, NOC fee, rishwat, baksheesh, ghoos, hafta and/or chai paani. The chances of such illicit payment of bribes rises exponentially, due to a dearth of a predictive (as well as curative) policing system in the government processes and machinery in India.

As systems and procedures continue to become more digitized, there are more opportunities to leverage available data to find the red flags that can indicate corruption and other integrity risks.^[10] IBM researchers are working with the Kenyan government to climb the “Ease of doing Business” ranking. One measure has been to reduce the number of interactions with the government required to start a business. During the project, the number of interaction points has been reduced “from 11 to just three simplified steps”^[11]. They further plan to investigate the role of AI and blockchain technology to improve government service delivery. The word “corruption” is not mentioned, still this may turn out to be an alternative path in using technology to fight fraud or reduce the risk for corruption.^[12]

Given India’s democratic framework, its diversity and widespread existence of social and economic disparities across the nation, it is essential that AI development proceeds with the least possible social and economic disruption and with the maximum possible public support.^[13]

Governance using AI

The typical approach followed in the system of governance enlists auditing as a follow-up process to a completed transaction. The auditing process serves the purpose of ensuring that there is no lapse of procedure or any corruption involved during the transaction. Auditing has been an age-old conventional process which is reactive in nature and often fails to point out the loopholes in the government process and transactions at a proactive level. This cyclic loop of systematic inefficiencies leads to a poor governance across sectors and also leads to a poor behaviour amongst the government representatives.

With the advent of technology, it is extremely crucial to make the auditing process ongoing with the transaction. The transactions between the government and businesses or government and citizens are undertaken in multiple forms such as approval of business permits, approval for environmental licenses, health claim transaction for a government health scheme, fine for flouting traffic rules etc (as elucidated above). The list is endless and hence the sheer scale of government transactions are overwhelming to audit and resultantly, there is a prevalence of high levels of corruption herein.

A real-time Governance Auditing system which tracks the transaction – its genuinity, the process followed and the outcomes on a real-time basis is the need of the hour.

AI based Governance in Other Jurisdictions

A collaboration between Exiger and Transparency International (TI) in the United Kingdom aims to improve TI’s capacity to analyse public records to identify risk for corruption. In Ukraine, the local chapter of TI has developed its own AI tool to reveal fraudulent bids in public procurement. They named the tool Dozorro, as they deployed it to monitor the open source government procurement system Prozorro.^[14] The Brazilian Office of the Comptroller General has developed a machine learning application to estimate the risk of corrupt behaviour among its civil servants. Variables from criminal records, education registries, political affiliation, business relations, and more are included in the analysis.^[15]

Tax authorities in Australia, Canada, Norway, the United Kingdom and elsewhere have started to build AI-supported models that attempt to predict which high-risk individuals are most likely to react positively to different tax authority interventions for recovering revenue.^[16] Researchers from Valladolid, Spain, created an Artificial Intelligence system which can predict the risk of corruption in Spanish provinces and other variables associated with greater corruption such as the establishment of new companies, real estate tax, and others.^[17]

The Ministry of Transparency and Controller General’s Office, Brazil implemented a system to find evidence of deviations in the performance of public servants. The software uses machine-learning features, an AI technique that feeds data, presents criteria, and checks if the results of the analyses performed by the machine are within the expected range.^[18]

Proposed Model for India

I- AID- Governance (India’s Artificial Intelligence Driven Governance)^[19] is a proposed AI based system which shall capture the entire trail of a transaction and store every transaction as a blockchain and analyse every transaction trail to develop trends which can be later used as triggers to outline threshold for every transaction.

Every transaction shall be stored digitally and every representative ever involved in the transaction shall be digitally captured through a unique identifier (code). The transaction will be irreversible and shall establish an accountability on the representatives processing the transaction to follow the strict process. As soon as the process will deviate from its pre-chalked out flow, the auditing trigger will highlight the red flag/ deviation. For instance, in case a business required environmental clearance, it shall be required to pass through an environmental department officer followed by the environmental commissioner. In case the transaction instead passes through any middlemen, an alarm shall raise highlighting the deviation of the process flow.

Huge sets of data shall be analysed (on a real time basis), to determine the actual threshold of a process completion. This step shall also capture the time taken at every step for the approval process. This shall provide an average standard time and requisite documentation for an approval. In case the time taken for any step is noted to be significantly less than the average standard time, an alarm shall be raised. The triggers in this case are dynamic and hence can be updated to suit the current processing times. This shall ensure flexibility in accordance with the prevailing trends across the country. This will also aid in significantly reducing inefficiencies, highlight the impediments and raise an alarm in cases of bribery and corruption.

Such a system shall help the government in identifying geographical, demographical and sociological vulnerabilities among different regions within a state/ country, and accordingly target their efforts in a risk based pattern. By way of studying the life cycle of such government processes, better policies may be implemented to further improve the efficiency and effectiveness of the government’s policies.

Due to the maintenance of an audit trail for all the government processes, accrual of liability on the relevant stakeholders and maintenance of evidence in case of any misconduct shall be maintained and accessible to the law enforcement authorities, to act in the interest of justice. It shall be an open blockchain platform wherein the evidence, so collected can be seen by everyone on the network but can be changed only on the consent of everyone, which is practically not possible in cases where there is suspicion on the transaction or any of its stage.

A specific first hand implementation of the proposed application should take 2-5 years for building a prototype after collating data at a national level, beta testing at the local level within the different states of India, correction/ upgrading the project, if required and a national level launch in a phased manner.

Roadblocks to Success

To effectively reap the benefits of deploying AI in the government auditing systems, certain obstacles are required to be tackled such as inadequate expertise in the research and application of AI related technology, inadequate enabling data ecosystems to access artificially intelligent data, exorbitant resource cost and awareness regarding adoption of AI, data privacy and security related issues, lack of collaborative approaches for the adoption and application of AI, and lack of basic access to technology in most parts of rural and geographically tough locations.

The predictions and performance of algorithms are constrained by the decisions and values of those who design them, the training data they use, and their intended goals for use.^[20] It should be ensured that the data which is initially fed on the AI based system is non-biased and reflects high standards of transparency. If any inaccurate information is fed on the system, procedure for its rectification as well as liability for the mistake committed by the AI based system must be fixed. The system should also provide for an online grievance redressal service for lodging complaints relating to network/accessibility issues, amending the entries entered, excessive delay in the process etc. The red flags should additionally be responsive to the subjective considerations involved which may expedite/delay the matter.

Way Forward

Legal amendments are required to be introduced in the existing law to accommodate the liabilities accruing from the proposed model. Middlemen are the backbone of informal sector in India and people have often depended on them because of lack of knowledge and awareness. The government should encourage awareness drives regarding the use of the AI based system so that the reliance of the people on such middlemen is significantly reduced. Auditing standards are further required to be translated into programming codes which can be applied to the AI based system in order to bring it in line with the accounting and legal standards. Legal liability should be accounted for the acts of the AI based system either by conferring juristic personality to it or placing the government under radar for the acts of the system. This issue needs to be addressed with caution and care, because it is often argued that once an AI based system is put in place, especially the one which employs machine learning, the creator no longer has sufficient control over how the work is being created and under which conditions.^[21]

(This post has been authored by Tanya Ganguli. She is currently working as a Senior Associate at Law Offices of Panag and Babu)

References

1. Tanya Ganguli, Financial Crimes and Corporate Governance lawyer, India ↑

2. Turing, Alan M. (1950). Computing machinery and intelligence._Mind_59 (October) ↑

3. Corruption Perceptions Index 2020, Transparency International, accessible at https://www.transparency.org/en/cpi/2020/index/ind ↑

4. The National Institution for Transforming India (NITI Aayog) is the policy think tank for the Government of India for achieving sustainable development goals by fostering the involvement and the participation in the economic policy making process by the State level governments in India. ↑

5. Available at: https://niti.gov.in/writereaddata/files/document_publication/NationalStrategy-for-AI-Discussion-Paper.pdf ↑

6. https://www.brookings.edu/blog/techtank/2018/05/17/artificial-intelligence-and-data-analytics-in-india/ [Last Accessed on 23 August 2020]. ↑

7. John Quain, “Crime-Predicting A.I. Isn’t Science Fiction. It’s About to Roll Out in India,” Digital Trends, April 11, 2018. ↑

8. As per Section 2(d), Prevention of Corruption (Amendment) Act, 2018, ‘Undue Advantage’ refers to any gratification other than legal remuneration, not limited to being pecuniary in nature or estimable in terms of money. ↑

9. As per Section 2(c), Prevention of Corruption (Amendment) Act, 2018, ‘Public servant’ refers to any person who is in the service or pay of the government or a government -funded entity, or other body owned or controlled or aided by the government. ↑

10. https://blogs.worldbank.org/governance/can-artificial-intelligence-stop-corruption-its-tracks [Last Accessed on 23 August 2020]. ↑

11. https://www.ibm.com/blogs/research/2018/10/breaking-50-kenyan-government-ibm-research-collaborate-improve-countrys-ease-business-ranking/ [Last Accessed on 23 August 2020]. ↑

12. https://www.cmi.no/news/2149-is-artificial-intelligence-the-future-tool-for#:~:text=Artificial%20Intelligence%20(AI)%20can%20be,in%20financial%20transaction%20data%2C%20unparalleled. [Last Accessed on 23 August 2020]. ↑

13. Report of Task Force on Artificial Intelligence for India’s Economic Transformation, Ministry of Commerce and Industry (2018). ↑

14. https://www.u4.no/publications/artificial-intelligence-a-promising-anti-corruption-tool-in-development-settings/shortversion [Last Accessed on 23 August 2020]. ↑

15. Ibid. ↑

16. OECD, Advanced Analytics for Better Tax Administration: Putting Data to Work (n2), OECD Publishing (2016); https://resources.centreforpublicimpact.org/production/2019/01/CPI-AI-Case-Study-Taxation.pdf. ↑

17. Lopez-Iturriaga, Felix Javier and Pastor-Sanz, Iván, Predicting Public Corruption with Neural Networks: An Analysis of Spanish Provinces (November 22, 2017). Social Indicators Research, Forthcoming , Available at SSRN: https://ssrn.com/abstract=3075828 or http://dx.doi.org/10.2139/ssrn.3075828. ↑

18. Orgãos Públicos Usam Inteligência Artificial para Combater Corrupção, TECNOLOGIA – IG (Aug. 3, 2018), https://tecnologia.ig.com.br/2018-08-03/inteligencia-artificial-contra-a-corrupcao.html, archived at https://perma.cc/V9KQ-K6AF. ↑

19. The “I-AID-Governance” conceptual application is proposed by the author which incorporates AI in government processes and transactions for curbing corruption in India. ↑

20. https://www.oecd.org/corruption/integrity-forum/tech-topics/ [Last Accessed on 23 August 2020]. ↑

21. Spindler, G., Copyright Law and Artificial Intelligence. IIC 50, 1049–1051 (2019). ↑