Shuichiro Ogawa
日本語

Notes · updated 2026-10-11

What Should Juniors Do When Generative AI Takes Away the Entry-Level Work? (Industry Sources)

This note examines, through industry sources, the problem that generative AI takes over simple tasks so that juniors are asked to handle difficult work without the entry-level practice that used to precede it. Sources include primary documents from the UK government, the New York and St.

Contents (6)
  1. The question and the scope of this note
  2. Have junior postings fallen and senior postings risen?
  3. Where did the entry-level work go?
  4. What do people inside the industry see as the problem, and what are they doing?
  5. What juniors can do
  6. Where this account does not apply

The question and the scope of this note

When generative AI takes over simple tasks, what remains for people is the difficult work that AI cannot handle. Simple tasks were also where newcomers built their skills, so newcomers are now expected to deliver results on difficult work without that groundwork. In sports terms, this is like being sent into a real match without the practice swings or ball drills that normally come first. This note checks this view against industry sources and then sets out, within what those sources support, how juniors might act. The focus is creative work (design, advertising, illustration, writing, video), with examples from software development, where sources are more plentiful, used as a supplement. An account from learning science and expertise research is in the academic counterpart, What Should Juniors Do When Generative AI Takes Away the Entry-Level Work? (Academic Research).

Three kinds of sources were used: primary documents from public bodies and trade associations; published figures from research firms and job sites whose methods are stated; and statements by people inside creative organizations and AI companies. Figures from research firms and job sites come from their own data, and only the factual parts with stated methods were used.

Have junior postings fallen and senior postings risen?

The decline itself is observed in several data sources with different definitions. On the job site Indeed, entry-level postings (roughly 0 to 1 year of experience) fell 7.5% and senior postings (5 years or more) rose 14.7% year over year as of May 2026 (Indeed Hiring Lab 2026a). An analysis by the UK government using LinkedIn data found that UK hiring fell 14% year over year and that entry-level hiring of graphic designers fell 28% (DSIT 2026). In the 2025 census by IPA, the trade body for UK advertising agencies, the number of staff aged 25 and under fell 19.2% from the previous year, and the share of agencies employing graduate trainees or apprentices fell from 56% to 43.4% (IPA 2026). A Stanford analysis of US payroll records found that employment of 22- to 25-year-olds in AI-exposed occupations was 19% lower than it would have been had it followed the path of the same age group in less exposed occupations, with no comparable gap for experienced workers (Brynjolfsson et al. 2026).

That generative AI caused this decline, however, has not been established. Indeed’s entry-level postings have been declining since their 2022 peak, and the author, citing among others New York Fed research that points to the spread of remote work, writes that several factors are at least partially at play (Indeed Hiring Lab 2026a). Researchers at the New York Fed analyzed job postings and reported that the decline in postings for AI-exposed occupations began before the release of ChatGPT and shows no clear additional break after 2022 (Audoly et al. 2026). The same article notes that related work using job postings found that demand for junior and senior roles in these occupations declined at roughly the same time. The UK government analysis shows that the decline began in late 2022 and states explicitly that it is not causal evidence of AI’s impact (DSIT 2026). The ECB likewise concludes that the current evidence does not allow changes in euro-area youth employment to be attributed to generative AI or to remote work (Chaloupka and Dias da Silva 2026). The Stanford authors also describe their analysis as descriptive rather than causal (Brynjolfsson et al. 2026).

Some sources do assign AI a role. A St. Louis Fed analysis lists AI-related job openings as one of five factors that help explain changes in young workers’ outcomes, and reports that in states such as Kentucky, Mississippi, and Tennessee, demand for AI-related jobs accounted for nearly half of the rise in young workers’ unemployment rate (Rodgers and Kassens 2026). This is a decomposition of factors and does not show that AI eliminated jobs. In a Gartner survey of 110 heads of HR, 22% said that at least one business leader in their organization had stopped entry-level hiring because of AI automation (reported by People Matters 2026). This is a share of organizations, not a 22% reduction in entry-level roles.

Whether experience requirements have risen depends on the field. Among Indeed’s tech postings, the share requiring at least 5 years of experience rose from 37% to 42% between 2022 and 2025, but outside tech it fell from 16% to 11%, including in marketing (Indeed Hiring Lab 2025).

In short, the first half of the premise, that junior postings have fallen and senior postings have risen, is observed in several data sources, but the second half, that generative AI caused it, has not been established. Some creative entry points, such as UK graphic design and advertising agencies, show large declines. For the drop in UK agency staff, however, some in the industry point to client budget cuts rather than AI, and the IPA census itself does not state a cause.

Where did the entry-level work go?

Even without an established cause, public documents on the creative industries record that AI has begun to take over simple tasks. Skills England writes that generative AI is shifting creative roles from “makers” to “directors” of AI output (Skills England 2026). The UK government’s AI adoption plan, citing a Creative UK report, warns that some generative AI tools can perform tasks traditionally assigned to junior staff, and that automating them without a plan for training and progression weakens existing entry routes (AI Champion 2026). According to the same document, 51% of creative businesses in the UK use AI, compared with 33% of all businesses, and 53% in design.

At the freelance entry point, the loss of work is hard to see. The London AI and jobs taskforce summarized reports of falling demand and fees for some freelancers in graphic design, writing, and photography, and noted that work is being reshaped in gradual, uneven, and hard-to-observe ways that aggregate statistics may miss (GLA Economics 2026). In a member survey by the UK Society of Authors (787 responses), 26% of illustrators said they had already lost work to generative AI, and 37% said the income from their work had decreased in value (Society of Authors 2024). The survey relies on self-selected responses and does not break results down by career stage.

What do people inside the industry see as the problem, and what are they doing?

Those who train juniors describe the loss of entry-level work less as a matter of hiring numbers and more as a loss of time for development. In a discussion among people from design and advertising, participants said that commercial pressure demands immediate results from juniors and leaves less time to develop them (Design Bridge and Partners 2026). In the same discussion, participants said the first six months after joining are critical and require mentoring, buddy schemes, and clear expectations, and they described a 16-week training program and buddy schemes organized by department.

The same change is reported inside an AI company. In Anthropic’s internal study, many questions that used to go to colleagues now go to AI first, and a senior engineer said that more junior people did not come to them with questions as often (Anthropic 2025). The same study reports concern that delegating more work lets one’s own skills erode, and that some people deliberately do work themselves even when AI could help, to keep their sense of the work. The study is mostly self-reported, and its participants are engineers and researchers.

Some executives argue that hiring juniors should continue. The CEO of AWS said that juniors are the cheapest staff and the most eager users of AI tools, and that without developing them there will be no experienced people ten years from now (The Stack 2025). In an open letter published by D&AD, the head of a UK agency argued that what is halting hiring is less the technology itself than the expectation that it will have an impact (How 2026). These are the views of executives, and no hiring record is presented for their own firms.

What juniors can do

From the sources above, five behaviors for juniors have some support. All of them rest on first-hand accounts or small studies, and no industry study was found that compared whether differences in juniors’ behavior affect hiring or promotion. The strength of the evidence is noted for each.

The first is to use AI in a way that keeps one’s own understanding. In an experiment by Anthropic researchers that randomly assigned 52 mostly junior software engineers, the AI group averaged 50% on a comprehension quiz after a task using a new library, against 67% for the hand-coding group (Shen and Tamkin 2026). The largest gap was on debugging questions. Within the AI group, scores were higher for those who generated code and then worked to understand it, who asked for code together with explanations, or who limited their questions to concepts, and lower for those who delegated entirely or handed debugging to AI. Because this is a coding experiment that measured understanding right after the task, it cannot be applied directly to creative work. Still, keeping the effort of checking why an output works, rather than stopping at receiving it, is consistent with this result.

The second is to seek questions and critique actively. As more situations can be settled by asking AI, juniors have fewer occasions to ask senior colleagues (Anthropic 2025). People in design and advertising emphasize mentoring and buddy schemes in the first months after joining (Design Bridge and Partners 2026), and the D&AD letter asks organizations that cannot hire juniors to review portfolios instead (How 2026). Because such occasions shrink if one waits, there is value in juniors asking for them. No source, however, measured whether juniors who sought critique developed better.

The third is to show judgment and thinking process, not only finished work. The design school Shillington writes that studios value junior designers for their judgment, conceptual thinking, cultural awareness, and collaboration, that generic AI visuals without reasoning are unlikely to impress, and recommends showing thinking process in portfolios (Shillington Education 2026). This is the view of a school, not the result of a survey of employers.

The fourth is to become the people who teach the organization what they learn about AI. At the UK agency HOW+HOW, juniors and mid-level staff present what they have learned about using AI to the team every 8 to 10 weeks, which also gives senior staff a way to learn AI (How 2026). The AWS CEO also sees juniors as the most eager users of AI tools (The Stack 2025). Juniors’ early familiarity with AI can serve as a foothold for moving beyond the position of only being taught. This too is one firm’s practice, and its effect has not been measured.

The fifth is to choose entry points that offer training or apprenticeships. The UK creative industries have 71 apprenticeship standards linked to priority occupations (Skills England 2026). However, only 5% of creative employers currently have apprentices, against 11% nationally, and creative employers report that 21% of the graduates they recruit are not prepared for their role, against 14% across all sectors (Creative PEC 2025, based on 2022 data from before generative AI). Entry points of this kind are few, but in organizations that offer training or apprenticeships, the groundwork survives as an institution.

Not skipping foundational knowledge underlies all five. A GitHub researcher writes that even when AI handles implementation, foundational knowledge is needed to evaluate its output (GitHub 2025). A case study by Japan’s JILPT likewise notes that using generative AI without basic learning of one’s own tasks may cause problems at work, and that workplaces need people who understand the tasks well enough to teach them (JILPT 2026).

Where this account does not apply

Figures on junior postings and employment differ in definition across job sites, payroll records, and trade-body censuses, and they cannot be lined up as measures of the same phenomenon. The cause of the decline has not been established, and many sources point to factors other than generative AI (interest rates, remote work, an oversupply of graduates). The five behaviors rest on first-hand accounts, one firm’s practice, a school’s view, and a small coding experiment; no industry study comparing the effects of juniors’ behavior in creative work was found. No public statistics showing trends in junior hiring by occupation for creative jobs in Japan were found either.

Unverified items

No unverified statements remain in the main claims. Sources that could not be reached (WEF Future of Jobs 2025, OECD, PwC, McKinsey, Gartner’s original release, the full Burning Glass Institute report, Japan’s METI strategy for entertainment and creative industries, and the D&AD AI & Creativity Report 2026) are recorded in the ledger with the routes attempted and are not used in this note.

References

Author: Shuichiro Ogawa (Design Researcher / Consultant) About me →