🤖Platform 4: IRSHAD for Shariah Advisory

GPT-3 powered AI Chatbot

Intelligent Robo Shariah Advisor (IRSHAD) is an AI-driven solution developed by IBF Net Group. The term IRSHAD implies “guidance” or “the guiding hand”. In line with this connotation, IRSHAD is a system that emulates, or acts in all respects, with the advisory and guidance capabilities of a human Shariah Advisor. The system in different phases of development seeks to be functionally equivalent to an Islamic legal expert in tackling specific problem domains of reasonable complexity.

The initial module of IRSHAD aims to automate, smoothen and accelerate the process of consultation concerning (1) estimation of zakat liability of individual donors, a complex accounting process; (2) allocation of zakat among eligible beneficiaries, all of which require a sound knowledge of Islamic legal rules concerning zakat management. It offers a channel of communication with potential zakat payers through which the latter can seek guidance regarding their zakat liability accompanied by explanations and how best to pay the zakat due. While the English version has gone online, the Bahasa Indonesia version is currently in the testing phase.

Another module of IRSHAD currently under development is about Islamic wills and inheritance. Muslims can seek guidance regarding how best they should distribute their wealth and assets in accordance with the Shariah. This is extremely valuable especially for people living in Muslim-minority countries where the laws of intestacy are not based on Shariah. In these countries, it is important to have a will in conformity with Shariah, else, the assets of a deceased Muslim will be distributed according to domestic law, not Islamic law.

Rationale

What is the rationale underlying creation of an artificially intelligent Shariah expert, if a human expert can solve the same problem? Some answers to this question may be rooted in behavioural factors.

• Human Shariah experts are expected to be available for consultation only during stipulated hours or time periods. IRSHAD is available 24/7 with high reliability.

• Human experts are vulnerable to behavioural idiosyncrasies, may display cognitive or other forms of bias or even logical fallacies.

• Human experts may not make themselves available in hostile and hazardous environments. IRSHAD will be immune to any fear factor.

• IRSHAD will essentially clone the experts.

• It would explain the reasoning without bias or attitude.

• IRSHAD will not forget, while human experts may.

• Human experts are influenced by recency effects (most recent information having a disproportionate impact on judgement). They are also influenced by primacy effects (early information dominates the judgement). IRSHAD will be free from such effects.

• IRSHAD will be consistent. It will handle similar transactions in the same way. It will make comparable recommendations for like-situations.

Will IRSHAD increase efficiency in terms of costs? IRSHAD can be cloned easily without any upper limit on reproduction. One can make multiple copies of its system with further customization if required, while training new human experts is usually time-consuming and expensive. Although the IRSHAD system may be expensive to build and maintain, it is inexpensive to operate. Its development and maintenance costs can be spread over many users.

Will IRSHAD increase efficiency in terms of better analysis of complex rules and improved decisions? If there is a maze of rules (as usually the case with Shariah), then the IRSHAD system can unravel the maze with greater diligence. Human expertise in such cases is likely to be scarce and costly.

A computer-based system like IRSHAD provides permanent documentation of the decision process. It does a better job in terms of completeness. It can review all the transactions, while a human expert can only review a sample. It can produce more timely results, faster decisions and better prevent fraud and/or errors. Most importantly, it uses cutting-edge knowledge and combined wisdom of multiple human Shariah experts.

IRSHAD is a chatbot built using Natural Language Processing (NLP) technology and Neural Network/Artificial Intelligence. In addition to answering queries from user, IRSHAD can also help calculate user’s zakat liability by seeking answers to queries related to the value of his/her zakatable assets and/or earnings.

Performance

IRSHAD learnt to acquire more and more artificial intelligence as it increasingly interacted with humans. It is, however, expected that IRSHAD may not be able to respond creatively to unusual situations. And of course, it will fail to produce an answer when the problem is outside its area of expertise. Notwithstanding these limitations, the promise of IRSHAD has proved to be irresistible. How does IRSHAD compare with ChatGPT, the latest AI- based solution that has taken the world by storm? Because ChatGPT seeks intelligence on everything under the sun, it perhaps produces an inferior outcome when it faces more intensive questions in a narrow domain. For example, ChatGPT when confronted with questions related to zakat, completely failed to differentiate between the types of zakatable assets and the zakat rates applicable thereto. It applied a uniform “2.5 percent on total value of the assets” rule to every asset, while displaying ignorance about finer points and nuances of zakat estimation and allocation process. IRSHAD on the other hand, is able the handle queries relating all conceivable assets and/or earnings that are subject to zakat and estimate the zakat liability with an error percentage of 6 percent as defined by a case where a user shows disagreement or dissatisfaction by pressing a button to discontinue the session mid-way. Over time, this percentage steadily declines as lessons learnt from user feedback are quickly incorporated into the knowledge base.