← Course overview
Module 1 · Lesson

Where AI can improve efficiency

Start by separating two questions that are often blurred: can AI do this task, and should it? The tables below list common opportunities in the three areas named in the framework. The right-hand column rates risk using the rule from Module 0: risk rises as the task moves closer to a decision about an individual.

Recruiting

TaskHow AI helpsThe human’s roleRisk
Job descriptions and postingsDrafts from a brief; rewrites for clarity and inclusive language.Confirms requirements are real and lawful; approves.Low
Candidate outreach messagesDrafts personalised messages from a template.Checks accuracy and tone; sends.Low
Interview guidesProposes structured, job-related questions and scoring criteria.Selects questions; makes sure every candidate is asked the same core set.Low to medium
SchedulingAutomation offers times, books rooms, sends reminders.Handles exceptions.Low
Interview note summariesTurns notes into a structured summary against the criteria.Verifies the summary against the notes; makes the assessment.Medium
Candidate question-and- answer assistantAnswers questions about the process, benefits and location.Maintains the content; handles anything personal.Medium
TaskHow AI helpsThe human’s roleRisk
Screening, scoring or ranking applicantsA model filters or orders candidates.Must remain the decision-maker; needs bias testing, notices and legal review.High

Onboarding

TaskHow AI helpsThe human’s roleRisk
Welcome communicationsDrafts welcome emails, first-week schedules, FAQs.Verifies facts (dates, benefits); personalises.Low
30-60-90 day plansDrafts a role-specific plan from the job description.Manager edits and owns it.Low
Paperwork routing and remindersAutomation triggers tasks for IT, payroll and the manager.Monitors exceptions.Low
New-hire question assistantAnswers policy and “how do I” questions at any hour.Keeps content current; receives escalations.Medium
Training recommendationsSuggests learning content by role.Manager confirms relevance.Low to medium

Employee engagement

TaskHow AI helpsThe human’s roleRisk
Survey designDrafts clear, unbiased questions.Chooses what to measure.Low
Analysing open-text commentsGroups thousands of comments into themes with counts.Checks a sample; interprets; protects anonymity.Medium
Manager talking pointsTurns results into discussion guides.Manager adapts to the team.Low
Internal communicationsDrafts announcements and recognition messages.Checks facts and tone.Low
Predicting who might resignA model scores attrition risk.Treats scores as a prompt for supportive conversation, never as a basis for adverse action; legal review.High
Monitoring messages for sentiment or emotionA model infers mood from employee communications.In most cases, do not do this. See the warning below.Very high
WATCH OUTEmotion and sentiment monitoring

Inferring the emotions of individual workers from their messages, voice or face damages trust and is restricted by law in some places. The EU AI Act has prohibited AI systems that infer emotions in the

workplace since 2 February 2025, with narrow medical and safety exceptions. Analysing anonymous, aggregated survey comments that employees chose to submit is a different activity and is widely done.

KEY IDEAA simple risk ladder
  • Low: content and administration. Drafting, summarising, formatting, scheduling. Errors are easy to catch before they cause harm.
  • Medium: information and insight. Answering employees’ questions, analysing data. Errors can mislead people, so sources and checks matter.
  • High: evaluating or deciding about individuals. Screening, scoring, ranking, predicting. Errors can be discriminatory and unlawful. These need formal governance.

Good first projects come from the low rung. They build skill and trust while the organisation develops the governance needed for higher rungs.

Jot down key ideas in your own words. Saves automatically.