Boring stuff that might change the world
Accounting and accountants are boring. Yet a new way of doing debits and credits could enable AI to disrupt finance and change the way every business runs.
The scope of change
I realise debits and credits may not be of such great interest to the general reader. That got me thinking about the wider implications.
Accounting essentially consists of aggregating a large number of individual business transactions then analysing and presenting them according to well established principles. The principles are actually quite simple. Double entry bookkeeping, classifying everything as either assets and liabilities (the balance sheet) or trade (the profit and loss account.)
Over time, those principles have become embedded in a variety of customs, standards and rules. Most of Dan's article refers to published accounts. These are subject to rigorous rules such as GAAP and IFRS. (note: I would guess public accounts are less than 1% of the actual accounts produced in the world but maybe 5-10% of the cost and effort.)
There is a complex industry of accounting professionals inside and outside companies supported by specialist IT systems, processes, workflows and a whole lot of spreadsheets that does nothing but apply these rules. Another, closely related, industry audits the outputs of all that work.
What if AI could reinvent the entire process?
Unlike many proposed applications of AI, this seems to me to be quite feasible. Essentially, you would build a programme with all the accounting rules and principles embedded in it. Then feed it all your transaction data and get it to do the job.
With only a little imagination, this could produce both the accounts required by the rules and the type of financial analysis that Dan Davies suggests. Logically, different audiences would want different views on the numbers. So you would put the AI in users' hands and enable them to generate what they need.
Start with the giant hairball…
This could happen in a couple of stages.
The first step would be to automate what happens today. For large companies, producing both internal and published accounts is a surprisingly messy business.
Rules and standards vary by country. A typical enterprise has hundreds of companies in many jurisdictions as well as the main quoted company that analysts pore over.
Add to that management accounts at various organisational levels. None of them align with legal entities. All are compared to budgets and forecasts. In a process freighted with incentives and potential for conflict.
Between SAP or Xero and figures that anyone actually looks at there is a giant hairball of spreadsheets, stress and friction fuelled by gallons of coffee.
…then throw it all away
Cleaning this up would save a nice chunk of overhead. And it would allow AI to be tested and proven, building confidence in the outputs before taking the bigger step.
Once you go beyond the mechanics, you are asking AI to aggregate, analyse and present the transactions. Different users asking different questions would get different analyses and presentations based on the same data.
I hope it's obvious that this would mean much more than simply cutting out some overhead. This would disrupt and rebuild the whole finance function.
That is very different from simply taking away some junior jobs. Business finance functions are just part of a wider community of professionals. That includes external accountants, auditors, financial and investment analysts and a fair chunk of the management consulting industry. And it spreads across all organisations. private, public and third sector. Small, medium and large. Formal and informal. Legal and illegal.
Disrupting this web of people and services opens up all sorts of fundamental changes:
Can the big 4 adapt their business models to meet this challenge?
If a new model of auditing emerges, does that affect confidence in the public markets?
Finance functions are part of the glue that binds big organisations together. How do things change when everyone can get their own numbers for themselves?
Accounting is also a great industry for developing smart people into good business managers and leaders. What happens to that talent pipeline if the whole industry is disrupted?
That could be a long list of questions. And it's not hard to see a similar outcome across some other back office functions. Supply chain, HR, Legal and Risk all have some of the same patterns.
Not all of this will happen and none of it will happen fast. Yet I find it helpful to think this way. Over my lifetime, tech has changed how business works in multiple waves. AI will be the same. So we should think about the shape of the new world. And not worry so much about how AI affects the old ways of doing things.