How AI is changing entry-level work, task by task
By First Circuit Editorial · · 3 min read
Headlines about AI and jobs swing between "nothing will change" and "no one will hire graduates." Neither is a useful plan. A more practical way to think about it is that AI tools change specific tasks inside jobs, and the people who do well are the ones who understand which tasks those are.
Tasks, not jobs
Every job is a bundle of tasks. A junior marketing coordinator drafts copy, books vendors, pulls reports, answers questions and sits in planning meetings. Some of those tasks, such as first drafts and summaries, are things current AI tools can speed up. Others, like negotiating with a vendor or noticing that a campaign clashes with a local event, depend on context, relationships and accountability.
BLS distinguishes AI exposure from predictions of job loss. In practice, a tool that drafts faster can shift a job toward editing, checking and deciding. Be skeptical of precise predictions about how many jobs will disappear. [1]
What it looks like in different fields
| Field | Tasks AI can help with | Tasks that still need you |
|---|---|---|
| Business and operations | First-draft emails and reports, spreadsheet formulas, summarizing documents | Checking numbers against sources, judging what matters to a client, owning the decision |
| Arts and design | Mood boards, variations, rough mockups, transcription | Taste, a coherent point of view, rights and licensing, working with a client's brief |
| Engineering and technical | Boilerplate code, test scaffolding, explaining unfamiliar code | System design, safety, debugging what the tool got wrong, knowing when a result is plausible |
| Research and writing | Finding leads, outlining, rephrasing | Verifying every claim and citation, original argument, ethical sourcing |
Check everything
AI tools produce fluent text and confident code that can be wrong. Treat output as a draft from a fast but unreliable collaborator:
- Verify facts, numbers and quotes against an original source. Don't cite a source you haven't opened.
- Run code and test it with real cases, including edge cases.
- Read for tone and fit. Generic phrasing is easy to spot.
- Keep notes on what you changed and why. That record is useful evidence of your judgment.
Protect privacy and follow the rules
Don't paste confidential information, such as an employer's documents, customer data, classmates' work or anyone's personal details, into a tool unless your employer or instructor has approved that tool for that use. Follow your school's academic integrity policy and any class-specific rules about AI; university career offices such as UC Berkeley's offer guidance on using generative AI in a job search, including reviewing and personalizing anything it drafts. In a job, ask what's allowed before you start; many organizations have policies.
Show judgment, not just speed
If AI makes first drafts cheap, what stands out is what comes after: choosing the right problem, checking the work, explaining trade-offs and taking responsibility. In applications and interviews, describe how you used a tool and how you checked it. "I used an AI assistant to draft three versions, then tested each against our survey data and kept the one that matched" says more than "proficient in AI."
The NIST AI Risk Management Framework is written for organizations, but its plain ideas, such as knowing a system's limits, measuring how well it works and keeping humans accountable, are a good mental model for individual use too.
Questions to ask employers
Interviews are a good place to learn how a team actually works with AI. Useful questions include: Which tools does the team use, and for what? How are outputs reviewed before they reach a client or the public? What data are staff allowed to use with these tools? How has the work of junior staff changed in the past year? The answers tell you what you'd learn in the role and how carefully the organization manages risk.
What to do this semester
- Pick one task in your field and try doing it with and without an AI tool. Compare time and quality honestly.
- Learn the fundamentals the tool depends on: the statistics, writing, design principles or engineering basics. You can't check what you don't understand.
- Build one small project that shows you can use a tool responsibly. Our guide to an AI-ready portfolio walks through it.