Appendix 1: NoFlake Business Case: TechnoTec has decided to develop an innovative financial product for consumers, NoFlake. The management has understood that borrowers who are at greater risk of having money stolen from them by hackers (or other forms of cyberattacks) are less likely to be able to re-pay their debts and are more likely to be a vector for infiltration of TechnoTecs IT systems. Therefore, to create a competitive advantage over the industry leaders, NoFlake will use an artificial intelligence engine to analyse both the credit risk and the cyber-security maturity of their borrowers. How Does NoFlake Work?

The AI engine uses an algorithm (see Appendix 2) to generate a predictive risk score between -100 and +100 for each potential borrower (with positive 100 being awarded to thelowest-risk borrowers and negative 100 being awarded to the highest-risk borrowers).

Who Built NoFlake?

The algorithm development team consisted of: Philippe Jankins-Chao (Team Leader, USyd IT graduate, age 26) Carmichael St Clare (Senior Developer, USyd IT graduate, age 24) Sebastian Montford-Smythe (Developer, Macquarie IT graduate, age 24) The development team hired Swizzle Pty Ltd to collect training data for the algorithm. Swizzle uses individual backpackers (paid minimum wage) to conduct surveys on major public streets in capital cities.

Financial Analysis of NoFlake:

TechnoTec currently pays 20 employees an average of $76,000 per year to determine which loan applicants should be approved, of which only two employees would be needed after NoFlake is introduced. NoFlake has two main goals: 1) to outperform human credit risk analysts by at least 20% (i.e. that the rate of loan non-repayments by applicants selected by NoFlake will be below 5.6% vs the current non-repayment rate of 7%); and 2) to lower by at least 40% (i.e. $600,000) the annual losses incurred by TechnoTec as a result of cyber-attackers gaining access to its internal systems using the legitimate log-in details of its customers. In 2020, TechnoTec lost $1.5m from such cyberattacks. On a loan-book totalling $50 million, the expected annual savings from NoFlake are: $50,000,000 * 0.014 (credit benefits) + $600,000 (cyber-security benefits) + $76,000 * 18 (labour cost savings) = $2,668,000. Given an expected cost of development of $1 200 000 and ongoing operational costs of $300,000, over a five-year period, the expected savings from implementing NoFlake are: $2,668,000 * 5 ($1,200,000 + ( $300,000 *5)) = $10,640,000. This equates to an 887% return on initial development costs, or 177% per annum over five years. Additional revenues could be generated by licencing the algorithm to third-party financial institutions (such as mortgage brokers) and paying a trailing commission to those brokers based on the value of the loans they originate which are approved by the algorithm. Those brokers would need to be trained in how to gather and fill in the information needed from borrowers to enable the algorithm to make its decisions. Question: With the appendix provided, give an analysis of the strengths and weaknesses (including unfairness, discrimination and biases) of the technology in the business proposal