In a post titled Skate to Where the Puck is Going, Armand Capisciolto, Chair of Canada’s Accounting Standards Board, recently set out some questions arising from ongoing changes in technology, relating to the Board’s ongoing research into how those changes are affecting the way financial information is consumed and accounting standards are applied:

Areas he raises include, on the consumption side:

• Will users of financial statements continue to read them in a linear fashion, or will they upload them into artificial intelligence (AI) tools that analyze the financial statements for them?
• Is disclosure overload still a problem? Will page count become irrelevant in a world where financial statements are analyzed by AI?

And on the application side:

• Will preparers and auditors use AI to apply accounting standards? I think we already know the answer is yes – but to what extent, and how will the use of technology interact with professional judgement?
• Are standards written in a way that AI can apply them appropriately – and should they be?
• How will we know that AI tools are using the right knowledge bases to apply standards? Many models are effectively a “black box,” and users don’t always know what they’ve been trained on. To properly apply IFRS® Accounting Standards, it’s not just the words of the standards that matter – you also need to consider the Basis for Conclusions, illustrative examples, and IFRS Interpretations Committee agenda decisions.
• Is technology the great equalizer? Will organizations that have historically lacked the capacity to deal with complex transactions and standards be able to use technology to manage those constraints – or will investment in technology widen the gap between the “haves” and the “have‑nots”? How might this affect our work on scalability?

He also asks whether digital reporting taxonomies will continue to be needed if AI tools can read and analyze financial statements directly, amusingly commenting: “If taxonomies are no longer needed, Canadian securities regulators will look like futurists for never requiring XBRL reporting.” I’ve long expressed skepticism about the supposed importance of XBRL (in 2019 I noted that “the reliance of XBRL on an initially laborious tagging process already seems old-fashioned and frankly, unexciting, compared to the possibilities of artificial intelligence” – yep, folks, that was in 2019) although if Canadian regulators do indeed come out looking good in this respect, it will be more through luck than skill. And although my own futurist capacities are as patchy as anyone else’s, I’ve written many times in this space about the failure of standard-setters and others to adequately grapple with changing times, and about the dubiousness of continuing to trot out “disclosure overload” as an excuse for anything, at a time when humans claiming to be suffering from such an affliction will only be bolstering the case for their own redundancy.

Another astute question: “How will junior accountants be trained to assess whether AI is producing reasonable results if AI is doing much of the work that junior accountants traditionally performed? Where will they develop the professional judgement needed to apply principle‑based standards?” This adds another dimension, I would add, both to the ongoing shortage of people entering the profession, and to the remorseless drive to improve audit quality: any AI-fueled squeeze on new blood flowing into the system (both in terms of the volume of recruits and the quality of experience they gain) can only make it harder to maintain the flow of reliably seasoned human wisdom flowing out. Indeed, an implicit question underlying all of Capisciolto’s points, as with much writing about AI in accounting, concerns the place of humanity in the whole infrastructure of producing and consuming financial information. Perhaps we need to collectively consider, if only as a thought experiment, the pros and cons of a future state in which financial information is generated and verified and fed into investment models without any human intervention at all. No doubt the quality of the information would suffer in various ways, but maybe, returning to the opening metaphor, it’s not possible to skate to where the puck is ultimately going, because it’s headed permanently off the ice…

That is, in a world of extreme volatility and lack of visibility about even short-term prospects, and of often deranged price multiples, the capacity to make major long-term capital allocation decisions based on fundamental financial analysis will likely only decline, with the great majority of trading being even more opportunistic than it is now (it’s notable that the S&P 500 recently hit a record high based on optimism about the American-Iran war coming to an end, notwithstanding the ample reasons for broader pessimism). Core financial information will continue to provide fuel for that opportunism, but the detail that comes with it may be basically gift-wrapping, its fate merely to be ripped open and tossed aside for the sake of getting to the good stuff…

The opinions expressed are solely those of the author.