Ever since the advent of artificial intelligence, people across industries have struggled with a common question: how long before it comes for their jobs? According to a Goldman Sachs report cited by Business-Today, the answer depends on the nature of the work. Routine and repetitive jobs across financial services, retail trade, healthcare, and education face the highest risk of automation, while roles requiring greater expertise and judgement are more likely to be complemented by generative artificial intelligence (Gen-AI) than replaced.
Task Complexity Determines AI's Impact
Goldman Sachs groups workplace tasks into different difficulty levels. Under its framework, level 1 tasks represent routine work with the highest likelihood of substitution by Gen-AI. As the difficulty level increases, tasks become more complex and are less likely to be replaced. Instead, higher-level work is expected to continue requiring human judgement and expertise, with AI serving as a tool to support rather than substitute workers. The report stated: "Skilled-service sectors appear better placed for AI complementarity, while routine service sectors face higher substitution risk."
Financial Services: Analyst vs. Back-Office
The report highlighted financial services as one such example. It said analyst roles could use Gen-AI to automate parts of basic modelling work, which falls under difficulty level 3. In contrast, routine back-office operations, including basic compliance checks, are more likely to be replaced as they are categorised under difficulty level 1. The sector accounts for around 6% of total employment, excluding agriculture.
Retail: Cashiers Most Exposed
Retail trade presents a similar pattern. According to the report, AI can assist sales staff with inventory management across difficulty levels 2 to 4. However, cashier roles are more exposed to substitution through self-checkout systems, which are classified as level 1 tasks and are already being used in some cases. Retail employs around 20% of the workforce outside agriculture, making it one of the largest employment sectors.
Healthcare: AI Strengthens Clinical Work
In healthcare, Goldman Sachs said Gen-AI is more likely to strengthen clinical work than replace it. Doctors could use the technology for diagnostics and treatment planning, activities classified at difficulty level 6. Administrative responsibilities such as appointment scheduling, however, are considered level 2 tasks and are more likely to be automated. Healthcare accounts for around 3% of employment excluding agriculture.
Education: Mixed Impact
The report also pointed to a mixed impact in education. Tutors are expected to benefit from AI-powered personalised learning tools covering difficulty levels 3 and 4. Meanwhile, repetitive work such as grading assessments and multiple-choice tests with standardised answers, classified as difficulty level 1, is considered more vulnerable to automation. The education sector accounts for around 6% of total employment outside agriculture.
| Industry | Employment Share (excl. agriculture) | Highest Risk Tasks (Level 1) | Moderate Risk/Complementarity (Levels 2-4) | Low Risk/Complementarity (Levels 5-6) |
|---|---|---|---|---|
| Financial Services | 6% | Basic compliance checks | Analyst modelling (Level 3) | – |
| Retail | 20% | Cashier work | Inventory management (Levels 2-4) | – |
| Healthcare | 3% | Appointment scheduling (Level 2) | – | Diagnostics & treatment planning (Level 6) |
| Education | 6% | Grading standardised tests | Personalised tutoring (Levels 3-4) | – |
Overall, the report said Gen-AI is expected to automate routine, low-complexity work across major service industries while complementing workers engaged in more complex tasks. Rather than replacing jobs across the board, the technology is expected to reshape how work is carried out within these sectors. For enterprise technology decision-makers, understanding this risk gradient is crucial: investments in AI should target high-complexity, high-value roles where AI augments human capability, rather than merely chasing automation of routine tasks that may already be commoditised.
Sources:
