Impairment
A capitalised AI model is tested for impairment when indicators suggest its carrying amount may not be recoverable, such as a superior base model superseding it, a fall in usage, or production underperformance. Where the recoverable amount is below the carrying amount, the asset is written down and the loss recognised.
Impairment indicators for AI
Indicators include a materially better external or base model becoming available, declining internal usage, failure to meet the performance that justified capitalisation, and changes in the business that remove the need for the modelIAS 38§111. The presence of an indicator triggers a recoverable-amount test.
How the write-down is measured
Where the recoverable amount, the higher of fair value less costs of disposal and value in use, is below the carrying amount, the asset is reduced to the recoverable amount and the impairment loss recognised in profit or loss. Amortisation in later periods is then based on the revised carrying amount.
An entity carries a capitalised model when a markedly stronger base model is released, cutting the internal model's expected remaining usefulness. The recoverable amount is assessed below the carrying amount, and the model is written down, with the loss taken to the P&L. All figures are illustrative.
- S1IAS 38 Intangible Assets, IFRS Foundation (IFRS). https://www.ifrs.org/issued-standards/list-of-standards/ias-38-intangible-assets/
- S4FASB Accounting Standards Codification, PwC Viewpoint (US GAAP). https://viewpoint.pwc.com/dt/us/en/fasb/GAAP/Codification/Codification/228073.html