L’IA remplacera-t-elle les comptables ?

The honest answer depends on separating tasks from jobs. AI is very good at the structured, repetitive parts of accounting: coding transactions, matching bank lines to invoices, reading documents and flagging anomalies. Junior bookkeepers and staff accountants once spent much of their time on these tasks, and that share will keep shrinking. History offers a useful comparison. After electronic spreadsheets spread in the 1980s, the number of bookkeeping clerk jobs in the United States fell, while jobs for accountants and financial analysts grew. Cheaper analysis created demand for more of it. That pattern is not guaranteed to repeat, but it shows that automating calculation has not, in the past, removed the need for people who interpret numbers. Much of the profession also faces a supply problem rather than a surplus. In the US, fewer students have been choosing accounting degrees and sitting for the CPA exam while many experienced practitioners retire. The 150-hour education requirement is widely blamed, and several states have moved to allow other pathways to a license. Firms struggling to hire are adopting automation partly to cope. Demand is growing in work that needs judgment and trust: advisory and forecasting, tax planning, internal controls, sustainability reporting and assurance. Licensed accountants are also legally accountable when they sign audit opinions and tax returns, and software cannot take on that responsibility. The real risk is to the entry-level pipeline. If routine work disappears, firms must find new ways for juniors to learn the basics they used to pick up by doing that work. A common misconception is that accounting is mostly arithmetic. Arithmetic was automated decades ago. The core of the job is applying standards and law to messy facts and explaining the results to people.