Assurance over machine learning and algorithms

Article

Assurance over machine learning and algorithms

Trust in the age of algorithms, robots and cognitive technologies

Driving innovation and business opportunities

Algorithms reduce time spent on repetitive, time-consuming tasks and allow people to focus on exercising human empathy and judgement focusing on what matters most.

In a world increasingly run by algorithms, such as those used by banks in the decision to grant a home loan or by Google in the development of self driving cars it is important  to understand what could go wrong.

Poorly designed algorithms can promote a fake news story, make a pricing decision that could ruin a reputation of a business or create regulatory risks of an enormous magnitude.

  • “20,000 people sent Centrelink 'robo-debt' notices found to owe less or nothing” The Sydney Morning Herald, September 2017
  • “Irma: Airline ticket prices surge up to 600% as incoming hurricane sparks mass evacuations” The Independent, September 2017
  • “On a tragic day for the country, some big tech companies stumbled. Both Google and Facebook widely promoted false stories on their sites.” Fortune Magazine, October 2017

Case Studies of Algorithm Assurance in action

Helping clients gain confidence in their data, algorithms and processes across sectors and industries

  • Financial Services
    We conducted a review of the processes, data and the underlying algorithm that classifies home loans into “owner occupier” and “investors” for a bank. We identified discrepancies in the algorithm logic to bank’s expectation and methodology.
  • Energy and Resources
    We analysed the fixed asset registry depreciation procedures and algorithm to provide insight into how the changes made to the algorithm over the course a few years have impacted the P&L. We validated that the current methodology set out by finance is accurately reflected in the algorithm.
  • E-Commerce
    We reviewed the algorithmic product recommendation engine of a health insurance comparison and recommendation service. We identified a bias built into the algorithm that contradicted the marketing message of an independent advice in question.
  • Services
    We analysed 100% of transactions for a health services provider to prove that the revenue generated through automated billing algorithm was in line with the contractual arrangements in place.

Rest assured, Deloitte will help you gain trust into your algorithmic operations

With increased volumes of data, processes automation and decisions being made by algorithms, you will need assurance that your algorithms are working as intended and achieving the desired business outcomes.

Our offering specifically focuses on areas where you might be most vulnerable, algorithms operating in environments outside of their ERP, potentially posing real or dormant business and reputational risks.

Our A&A practice is uniquely positioned to help you gain confidence in your algorithms, mitigate potential risks and realise potentially untapped business benefits of automation. We combine our knowledge and capability of people, process, technology, data and long standing assurance capability to deliver this offering.

At Deloitte we have spent years assessing the best way to review and provide you assurance that your algorithms are robust, working as intended to and that they are achieving the business outcome it was originally designed for.

If you would like more information, assistance or have any questions, please feel free to reach out to one of our team members below.


Published: March 2018

Did you find this useful?