2026 HCI Prelim P1 SAMS

1 (a) Using Figure 1, explain which industry is most likely to adopt Artificial Intelligence (AI) widely. [2] ● Banking ● From Figure 1, it has the largest share/percentage of working hours that can be automated (optional: and augmented) by AI. (b) With reference to Extract 1 and using a suitable diagram, explain the source of market failure in the Artificial Intelligence (AI) industry. [4] Figure 1 ● From Extract 1, hyperscale data centers that produce AI capabilities emit planet-heating gases. Such gases increase the temperature around the data centres and third parties such as the residents living near the data centres would have to incur external costs such as spending more on electricity bills to cool down their homes. ● As marginal social cost (MSC) = marginal private cost (MPC) + marginal external cost (MEC), these external costs cause MSC to be higher than MPC. ● The market equilibrium quantity is Qm, where MPC = MPB (marginal private benefit) but the socially optimal quantity is lower at Qs, where MSC = MSB (marginal social benefit). ● Hence this overproduction of AI / overallocation of resources to AI results in a deadweight loss of area ABC. (c) Explain how the adoption of AI can increase labour productivity. [2] ● Labour productivity refers to the (value of) output per worker. ● For example, in Extract 3, AI assistance has helped Grab drivers get more jobs and reduces waiting times for riders. ● Hence this increase in number of rides per driver increases their labour productivity (as output per worker increases). Price MSC = MPC + MEC C MPC MEC at Qm B Ps Pm A MPB = MSB 0 QS Qm Quantity

2 (d) With reference to Figure 2 and a circular flow of income diagram, explain how the projected trend in US investment spending on AI will affect equilibrium national income in US. [4] ● From Figure 2, US investment spending on AI is rising. ● This results in an increase in (autonomous) injections in the circular flow of income shown above. ● This increased investment spending on AI creates more (factor) incomes for workers who provide factors of production to produce the higher output of goods and services in the AI industry. ● They then spend a portion of their additional income on more induced consumption (depending on the size of their MPC), while the rest (of their additional income) forms additional withdrawals such as savings, taxes and import spending. ● As one man’s spending is another man’s income, this leads to rounds of (progressively smaller) increase in induced consumption and withdrawals (via the multiplier process). ● This multiplier process stops when the rise in leakages = rise in injections (OR new total withdrawals = new total injections) and the equilibrium national income in US increases (by a multiple of the initial rise in investment spending).

3 (e) Discuss whether the adoption of AI will cause unemployment to rise in US. [8] R1: How AI can cause (structural) unemployment to rise ● Adoption of AI (automation) can replace some workers: Extract 2: “3,900 workers lost their jobs explicitly due to AI -driven automation” or “entry - level roles most vulnerable due to their routine nature” ● Adoption of AI creates jobs that require different skill sets: Extract 2: demand for AI -related roles, such as machine learning engineers, “AI ethicists, prompt engineers, and data annotators, has surged” ● The unemployed struggle to find new employment due to a mismatch between their current skill sets and the skills required in the changing job market; structural unemployment occurs when the skills of the unemployed do not match those required by available job vacancies. R2: how AI can cause (demand-deficient) unemployment to fall Figure 3 GPL Real Y AD0 AS Y0 0 Y2 AD2 P0 AD1 YFE Y1

4 ● From Figure 2, the adoption of AI entails a rise in US investment spending on AI. ● As investment is a component of aggregate demand, US’s AD will increase from AD 0 to AD1 and eventually to AD 2 due to the rounds of induced consumption explained in part d). ● Hence as concluded in part d), this will increase US’s real national output from Y0 to Y2 via the multiplier process. ● To produce the higher output, there will be greater demand for inputs such as labour and hence the higher derived demand for labour will create more jobs and reduce demand-deficient unemployment. Alternative approach for R2 (case of AS ) - 1. AS shift diagram - 2. Explain why AS shift down (COP ) &/or AS shift right (explain why productive capacity ) - 3. Derived demand / demand for labour inputs EV: Judgement about whether unemployment will overall rise or fall. [L1]: Clear stand why unemployment may rise / fall. [L2]: Stand justified with link to AI or US context. L2: 4-6m ● Makes use of the relevant tools of analysis to explain how structural unemployment may rise while demand -deficient unemployment may fall ● Includes the AD-AS model as a tool of analysis for demand- deficient unemployment ● Makes good use of case data to apply analysis to the context of AI adoption’s impact on unemployment. L1: 1-3m Lacking in any of the L2 criteria (f) Discuss the extent to which the adoption of AI can help a business become more profitable. [10] R1: Explain how AI adoption increases profits via increasing revenue o E.g. Extract 3: “drivers get more jobs” o Explain why AR & MR rise by linking to relevant demand factor

5 o How profit max. output increases (with profit max. condition) o How corresponding price increases o How profit increases R2: Explain how AI adoption increases profits via decreasing cost o E.g. Extract 3: “reduces waiting times for riders” -- > reduces petrol cost per passenger --> reduces variable costs OR reduces per unit cost --> reduces MC & AC o E.g. Extract 4: “AI could replace hundreds of human agent” -- > reduces labours costs --> reduces variable costs OR reduces per unit cost --> reduces MC & AC

6 o How profit max. output increases (with profit max. condition) o How corresponding price decreases o How profit increases SC: Judgement about extent to which adoption of AI helps a business become more profitable 1m: Clear justified stand about extent of rise in profitability with weak substantiation 2m: Clear justified stand about extent of rise in profitability with good substantiation from only cost OR revenue angle. 3m: Clear, contextualised & justified stand about extent of rise in profitability with good substantiation from both cost AND revenue angles. EV: Large extent / LR more profitable o E.g. How “OpenAI and Anthropic deploying dedicated engineering teams to work directly inside companies to co-build bespoke solutions” --> can increase a business’s success in AI adoption (be it to increase revenue &/or decrease costs) & hence increase chances of increasing profitability in the long run. EV: Small extent / SR not profitable ● EV: What may limit the rise in revenue o E.g. Extract 4: “wine company using app Cellar Tracker faced problem of sycophancy which resulted in accurate AI feedback” --> consumers cannot trust its recommendations --> may not help increase consumer demand for its wines and hence its revenue. o E.g. Extract 4: “being overly polite about poor -quality wine vintages may backfire as it can lead to erosion of consumer trust about the wine company’s products” -- > consumers switch to more trustworthy competitors --> this change in consumer taste and preference reduces wine company’s demand -- > offset earlier rise in AR & MR. o E.g. Extract 4” 40% of consumers are frustrated by the inability to reach a r