The future of ABC (audit, blockchains and code)

The views and opinions expressed in the text belong solely to the author, and not necessarily to the author’s employer, organization, committee or other group or individual.

The audit profession has undergone numerous iterations in the last few decades with the onset of greater regulatory requirements, stakeholder needs and technological changes.

There is also a greater level of concern being expressed across all quarters about the quality of audits. For instance, half of all audits monitored by global regulators had deficiencies according to the International Forum of Independent Audit Regulators (IFIAR).

In a world where trust is eroding, it will be important to understand what the introduction of blockchain technology coupled with AI powered by stable algorithms supported by machine learning could mean for the world of audit and assurance in a broader sense.

Without going too much into the description as to what a Blockchain is (essentially a distributed ledger that can only append and not delete transactions posted by different parties or peers in a network), I will attempt to provide a view of what the vision of the future could look like.

It will also be important to note that there are different types of Blockchains. You have on one hand a public or permissionless Blockchain which allows anyone to participate in it – Bitcoin runs on a public Blockchain. You also have a private or permissioned Blockchain which is only open to a defined group of individuals or entities (and where the consensus approach to proof of work may not be required).

You can read more about the underlying principles of Blockchain here or learn more about it here (edX).

 

Imagine

It is the year 2022 and the world’s largest FMCG company, Duopivot, has a permissioned Blockchain where all of its financial and operational transactions are recorded across all of its numerous entities globally.

Duopivot’s auditors, MonTian, gets its partner to run an algorithm, with a series of conditions, assumptions and principles, agreed mutually by Duopivot and Montian, on the permissioned Blockchain. Rather than a sample of tests, a complete test of all transactions is made by the algorithm resulting in queries or anomalies which are raised automatically by the AI built within the algorithm to Duopivot’s finance team

The finance team respond and make the necessary adjustments and the financial statements are declared true and fair and the audit is complete. Within 3 days. With no other interventions from auditors MonTian other than the audit partner running the code on the Blockchain.

In addition to a test of balances and figures provided by Duopivot, the algorithm also tests the stability of the Blockchain and detects any abnormal transactions raised across any of Duopivot’s entities by cross-referencing them to external datasets (such as consumer behaviour, sentiment, overall sales figures for the industry and GDP growth of the multiple markets Duopivot operates in). The AI is able to apply the principles of scepticism to the data provided by Duopivot using its own massive datasets and get a high level of assurance on whether the transactions recorded on the permissioned Blockchain is valid or not.

Following the review and sign-off by MonTian, Duopivot’s largest institutional investor WhitePebble, decide to run their own algorithm on Duopivot’s permissioned Blockchain to test the figures themselves and to also assess future investment potential in order to aid them for points to highlight during the AGM which takes place within 30 days of year-end.

MonTian’s algorithm and AI also are able to assess based on Duopivot’s data whether they have utilised the optimal tax channels to achieve greater tax efficiency, as the AI has access to the datasets of every single tax regulations globally and can therefore chart the route to ideal tax positions.

 

Implications

This is the world we could face where audit teams are reduced to individuals running code and managing code rather than managing audit team members and delivering results near instantaneously. It also allows for institutional investors to test the financial statements on their own without relying solely on auditors and also using the information for their investment decisions and approach.

The AI built within the algorithm could also provide recommendations for business decisions on areas such as game theory and provide a probabilistic approach to decision making that will be more attuned to market needs and conditions.

Changes in legislation or reporting standards can also be applied almost instantaneously (even where judgment is required) and be tested to a very high degree of certainty.

 

What is the likelihood of this?

Very high.

We already see how PwC’s GL.ai is an innovation made of algorithms. It analyses billions of data points in milliseconds and applies judgment to detect anomalies in the general ledger.

Northern Trust (Nasdaq: NTRS) also currently do this in a similar vein where they launched their Blockchain technology for private equity and allowing for audit firms to carry out audits through access on their own blockchain node, providing access to relevant fund data. Northern Trust currently allows for audit firms to obtain the master records to their own systems or to complete the audit on their blockchain itself. You can find out more here.

The ability of algorithms supported by machine learning , particularly augmented data learning and deep learning models that account for uncertainty (such as utilising the Bayesian approach to machine learning), the role, shape and future of audit and assurance as we know it will change dramatically.

 

Advertisements

Robots! Clear and Future Danger For Economies

I was at a conference recently and there was a speaker who was extolling the power of robots, technology, automation and artificial intelligence (AI) in the modern workplace and how it was going to revolutionise the global economy.

There was quite a catalogue of achievements as a result of increased robotics and AI including lower ‘FTE’ (or ‘Full Time Equivalent’ of human labour) requirements and greater efficiency, productivity and decreased errors and mistakes. These were achievements that were backed by undisputed statistics and data.

The ability to create consistently high economic value using systems, robots and AI which do not make mistakes, which do not break down often, which can even be self-correcting becomes very appealing.

However amidst the glories of robotics and AI, I felt increasingly concerned about where the world was heading with the increased introduction of automation, robotics and AI and the impact this was going to have on employment, social mobility and income equality.

My concerns

Technology as a displacer of jobs.

Technology, automation and robotics initially replaced blue-collar jobs and roles from the economies. Increasingly greater sophistication of AI means that white-collar jobs are also being replaced. We read various reports about the jobs of the future being technology-related roles that help create, maintain and repair robots and their related technology, but I postulate that robots can fix themselves (and their ‘peers’) better than people ever can and over time, robots can create other robots to do the tasks which they need done.

In the past, technology was an enabler. It was a great source of enhanced productivity for nations’ economies.

However, technology has now become a replacer or displacer – of jobs, of people, of roles. It has now become a tool to enhance economic output but ends up depleting people and their earnings.

This is going to be a longer-term fundamental problem and challenge to societal and economic growth and development.

The impact on developing economies

Let us consider Philippines and India. They have spent billions of dollars investing in the infrastructure and ecosystem to help create thriving shared services and business process outsourcing (SSCs / BPOs) businesses. This was to help meet the needs of multinational companies. However, with AI and automation increasingly taking on a majority of the roles and jobs that are currently being done by millions of people in both countries, it is going to lead to a significant job loss and risk the potential collapse of the SSCs and BPO sector in both countries.

Over time, with increasing automation and AI, multinationals need not outsource various roles to locations of lower labour cost. They will instead seek to outsource the roles to nations with the lowest tax and the best technology infrastructures in which they can base their systems and robots. 

The moral obligation and income inequality

With increasing AI and automation, I struggle to see how the job losses faced by millions as a result of robots taking on their roles are going to be mitigated. There also seems to be little alternative sources of formal employment.

Whilst it is easy to highlight how automation can reduce expenses by 66% and reduce ‘FTEs,’ I think we need to look at people beyond merely being an ‘FTE’ or as a mere factor of production.

 

Over time, it is going to also exacerbate the issues of income inequality which is already one of THE pressing moral issues of our time. I’ve covered this topic at length previously.

The factors of production, the technologies, the AI and robots are going to be in the control of a very small segment of society. Whilst it may create vast economic growths, it does not lead to growth in income or wealth for the majority of the people. This will lead to societal fractures which can be devastating to nations and society.

What then the moral obligation to people and society?

Possible solutions?

Leaving this issue to be dealt with purely by market forces will not result in resolution and frankly will be disastrous in my opinion. There needs to be a concerted governmental approach to resolving this and finding solutions that work.

Using levers such as tax policies will be ineffective, particularly in a world with little tax harmonisation. For instance, increased taxation for robotics-led solutions will only encourage a beggar-thy-neighbour policy and in a world with little tax harmonisation, it becomes a useless endeavour.

 

If we accept that robotics and automation are an inalienable part of the development of society, then we need to accept that the current economic models  will not be best suited for what the world needs. Maybe it is time for us to seriously consider and contemplate universal income as a way to mitigate and tackle some of the problems coming our way as a result of robotics and automation.

Universal income is something a number of countries are experimenting with to tackle income inequality which as I’ve explained earlier will only be growing with greater automation and robotics. Finland for instance has started a pilot programme, the Swiss held a referendum in June 2016 to consider universal basic income which did not pass as only a quarter of the Swiss agreed with it, the Dutch will be carrying out a pilot programme this year, and this is just a start.

What is increasingly clear is that it is not enough to simply hope the challenges brought on by AI and robotics are going to go away, there needs to be a concerted and strident efforts made to mitigate them.