Will AI soon be preparing your annual reports?

The Financial Reporting Council’s Financial Reporting Lab has published the latest in its series of technology deep-dive reports, examining the potential role of artificial intelligence (AI) in the corporate reporting process.

The Lab’s report raises the interesting prospect that, in the future, the entire exercise of preparing a company’s annual report — from collating the underlying data to auditing the report and validating its contents — could be carried out solely, or predominantly, by machines.

The report even suggests that, in time, annual reporting could vanish altogether, with companies generating their financial report at any time, covering any time period. Who knows? Developments in natural language production may even mean that, one day, an investor could visit a company’s website and generate their own custom report, there and then, using the investee’s own data.

Using AI in corporate reporting undoubtedly has the potential for efficiency and cost savings. A UK issuer’s annual report must address a raft of requirements, from legislation and FCA rules to the UK Corporate Governance Code and industry guidance issued by the Investment Association and others. An automated validation process would both streamline this exercise and reduce the margin of error.

Companies that do this will also need to address the potential risks. If not used properly, automated production could result in standardised, anodyne reports that fail to emphasise a company’s culture, strengths and objectives. A report that lacks a “human touch” could appear dry and inaccessible and suggest a lack of meaningful engagement by the board.

And inaccurate reports can give rise to liability. Companies will need to be satisfied with the integrity of their underlying data. Algorithms will need to be carefully constructed to ensure the AI takes in the relevant data, eschewing meaningless information that could render the report misleading.

The key will be striking a balance between the sheer power of AI-based data collation and assembly and the benefit of oversight and subtle inflection that (at least for now) only humans can bring.