Skip to content

exercise.md

Latest commit

 

History

History
128 lines (86 loc) · 8.75 KB

exercise.md

File metadata and controls

128 lines (86 loc) · 8.75 KB

Will AI Take Our Jobs?

CAUTION: This is AI-generated overview on the given topic.

TL;DR: AI won’t replace all jobs, but it will reshape tasks within most jobs, speed up some careers, diminish others, and create new roles. The outcome depends on how we adopt AI, how quickly we reskill, and which policies we put in place.

1) The wrong question: jobs vs. tasks

When people ask “Will AI take our jobs?”, they usually imagine whole occupations disappearing. History (electricity, PCs, the internet) suggests something subtler: technologies automate tasks, which changes job content, productivity, and demand. That’s why economists talk about task-level automation rather than job-level extinction.

  • Automation: AI directly performs parts of the work (e.g., summarizing emails, drafting text, classifying support tickets).
  • Augmentation: AI helps humans do the work faster or better (e.g., code suggestions, data cleaning assistants).
  • Reallocation: Time saved on automated tasks shifts to higher‑value activities (customer interaction, design, problem solving).

A good rule of thumb:

Today’s reality What changes with AI
Jobs are bundles of many tasks. Some tasks get automated; the rest rise in relative importance.
Experience → productivity gains. AI can compress the experience curve for juniors.
Skills decay without practice. Continuous learning + AI tools becomes part of the job.

Key idea: Tasks are automated; jobs are redesigned.

2) What the best recent evidence says

Short‑term productivity effects (micro evidence).

  • A large field study in a Fortune 500 contact center found that giving agents a generative‑AI assistant increased issues resolved per hour by ~14%, with the biggest gains for the least‑experienced agents—and better customer sentiment.1
  • A preregistered experiment with 444 professionals on writing tasks showed time to complete fell ~37% and quality rose ~0.4 SD with access to a chatbot; benefits were strongest for lower‑ability writers.2

Medium‑term labor‑market exposure (macro evidence).

  • The IMF estimates about 40% of global employment is exposed to AI (higher in advanced economies), with mixed effects: some jobs will be complemented (higher productivity, possibly higher wages), others displaced.3
  • An ILO global analysis (using ISCO data + GPT‑4 task mapping) suggests most occupations are more exposed to augmentation than full automation; clerical work shows the highest automation exposure.4
  • The OECD Employment Outlook finds early deployments more often change job quality and tasks than reduce employment outright—so far. Effects on job quantity may lag adoption.5

Forecasts (always uncertain).

  • The WEF Future of Jobs 2025 projects ~170 million jobs created and ~92 million displaced globally by 2030 (a net +78 million), driven by tech, green transition, and demographics. Skills needs will shift toward AI/data, cybersecurity, and adaptability.[^wef2025]
  • McKinsey modeling suggests ~27–30% of hours in Europe/US could be automated by 2030 (genAI accelerates this), implying millions of occupational transitions and a major skills upgrade agenda.[^mgi2024]

Takeaway: In the near term, AI looks more like a broad productivity booster and task reshuffler than a universal job destroyer. Over time, transitions could be large—and uneven.

  1. Who’s at risk, who benefits?

Higher exposure (declining demand without reskilling):

  • Routine office support & clerical roles (data entry, scheduling, basic reporting).
  • High‑volume customer support that is largely scriptable.
  • Repetitive content generation without domain expertise.

Lower exposure / likely complemented:

  • Roles with complex human interaction (healthcare, education, management).
  • Skilled trades and on‑site physical work (until robots catch up).
  • Jobs requiring domain knowledge + judgement + accountability (software, product, data, law—augmented rather than replaced).

Early career boost: The contact‑center study and lab experiments show AI can encode and spread best practices, letting juniors perform closer to seniors sooner. That raises productivity—and sometimes wage pressure—but may also compress career ladders unless organizations redesign roles.

  1. What actually changes inside a job?
  1. Time allocation shifts: less rough‑drafting, more reviewing, prompting, verifying.
  2. Quality bars rise: if AI drafts “good enough”, human value moves to problem framing, data/context curation, taste, and verification.
  3. Skills portfolio updates: promptcraft, toolchain literacy (APIs, automation), security/privacy hygiene, critical reading of AI output.
  4. Metrics change: throughput and cycle time improve; judgement errors (hallucinations, misclassifications) become the new bottlenecks to manage.

Design principle for teams: Automate the boring, amplify the human.

  1. What should you (a first‑year IT student) do?
  • Master the basics of data + automation: get comfortable with CSVs, APIs, and workflow tools (n8n, GitHub Actions).
  • Practice “prompt → verify → ship”: use AI to draft, but always check sources, test outputs, and document assumptions.
  • Build a visible portfolio: small public repos showing before/after with and without AI (README demos, notebooks).
  • Invest in durable skills: systems thinking, communication, security mindset, and domain context.
  • Work in pairs/teams: peer review catches model errors and improves your prompts and specs.

A simple checklist for any AI‑assisted task:

  • Define the task and acceptance criteria.
  • Draft with AI and log prompts.
  • Verify: test, cite sources, run sanity checks.
  • Document what the tool did vs. what you did.
  • Commit in small steps with clear messages.
  1. Policy and management levers (why this matters for outcomes)
  • Reskilling at scale (short, modular programs) to help workers cross into rising occupations.
  • Job redesign: let humans own goals and judgement, while AI handles defined sub‑tasks.
  • Guardrails: data protection, bias testing, provenance/watermarking, and incident response.
  • Safety nets: transition support where displacement is concentrated.

Reality check: AI’s net effects depend on choices—of firms (deployment), governments (policy), and individuals (skills).

  1. So… will AI take our jobs?

AI will take parts of many jobs and create parts of new jobs. Some roles will shrink; others will grow; almost all will change. If you learn to use the tools, validate their output, and ship work that combines AI scale with human judgement, you’re positioning yourself on the opportunity side of that change.

References & further reading

#Endnotes (inline citations)

Footnotes

  1. Brynjolfsson, E. et al. (2023). Generative AI at Work. NBER Working Paper w31161. ↩

  2. Noy, S. & Zhang, W. (2023). Experimental Evidence on the Productivity Effects of Generative AI. MIT Working Paper. ↩

  3. Georgieva, K. (2024). AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity. IMF Blog. ↩

  4. ILO (2023). Generative AI and Jobs: A global analysis of potential effects on job quantity and quality. ↩

  5. OECD (2023). Employment Outlook: Artificial ↩