Charta is on Google Play: practise on the go. Get the app
Charta

Account for the main AI advances made since 2017, separating those pushed by supply from those pulled by demand.

FRM® Part II · Current Issues in Financial Markets · The Financial Stability Implications of Artificial Intelligence · FRM2.CI.2.a

The idea

AI advances since 2017 can be sorted by where the push comes from. Supply-side factors are developments in what the technology can do and what it costs. They are the transformer architecture published in 2017, which let models train in parallel on very large text collections, and far greater computing power from specialised chips and cloud capacity. They also include the growth of large digital datasets, and the release of pretrained and open-source models that others can adapt cheaply. Demand-side factors are the reasons financial institutions want the technology. These are pressure to cut costs and raise efficiency, competition from rivals and fintechs, heavier compliance and reporting workloads, and customer expectations of fast, personalised service. The two reinforce each other, since cheaper supply makes more use cases worthwhile. The usual exam error is labelling a firm's own motive, such as cost saving, as a supply factor.

A practice question

From the question bank, with the reason each wrong option is wrong.

Which of the following is a supply-side factor behind the advance of AI in finance since 2017?

  1. Far cheaper computing power from specialised chips and cloud capacity - correct
  2. Customers expecting fast, personalised service around the clock from their bankWhy not: Demand-pull factor: it describes what users want, not an improvement in what the technology can do.
  3. Pressure on firms to cut operating costs and raise efficiencyWhy not: Demand-pull factor: it is the adopter's motive for using AI, not a technological push.
  4. Growing compliance and reporting workloads at regulated firmsWhy not: Demand-pull factor: it creates a need for automation but does not expand technological capability.

Supply factors are improvements in capability and cost of the technology itself, such as compute, data, algorithms and pretrained models. Cost pressure, customer expectations and compliance burdens are reasons firms want AI, so they are demand factors.

A harder one

A bank's innovation head lists three reasons for its generative AI programme: (i) transformer-based pretrained models are now available through cloud interfaces, (ii) the compliance function faces a rising volume of regulatory reporting that it wants to automate, and (iii) very large collections of digital text are available to train language models. How should the reasons be classified?

  1. (i) demand, (ii) demand, (iii) demand, because the bank chose to adopt in every caseWhy not: Treats the adopter's decision as the source of every factor; the supply or demand label depends on whether the factor is a technology development or a user need.
  2. (i) supply, (ii) demand, (iii) supply - correct
  3. (i) supply, (ii) supply, (iii) demandWhy not: Reverses the last two: reporting workload is a user need (demand) and the text collections are an input to capability (supply).
  4. (i) demand, (ii) demand, (iii) supplyWhy not: Labels the availability of pretrained models as demand, when it is a technology development pushed by providers.

Pretrained models and large text datasets expand what AI can do and are supply-side. The reporting burden is a reason the bank wants automation and is demand-side.

The hardest tier

Which statement best describes how supply and demand factors interact in the recent growth of AI use in finance?

  1. Supply factors alone explain adoption, because financial firms' needs have not changedWhy not: Ignores demand: cost, competition and compliance pressures determine which supplied capabilities firms actually adopt.
  2. Demand factors alone explain adoption, because the technology has been unchanged since before 2017 and only needs have grownWhy not: Ignores the supply advances in architecture, compute, data and pretrained models since 2017.
  3. They reinforce each other: cheaper supply widens use cases, lifting demand and funding more supply - correct
  4. Supply factors only matter for technology firms and demand factors only for supervisorsWhy not: Mixes up the roles: both supply and demand affect financial institutions and supervisors alike.

The two sets are complements. Lower cost and greater capability widen the set of worthwhile applications, rising adoption then supports more investment, and firm needs shape which capabilities are developed.

In the app: 6 practice questions on this outcome, spaced review that brings it back before you forget it.

Start free →

More in The Financial Stability Implications of Artificial Intelligence