October 3, 2026 - The Chicago Journal

The Junior Job Question in the Age of Agentic AI

The Chicago Journal Staff on October 3, 2026

The first draft used to be a junior’s job. Someone two years out of school got the memo assignment, the research summary, the intake questionnaire, the boilerplate contract, the first pass at the code. It was tedious work, and it was also how people learned the trade. Partners and managers marked it up, sent it back, and slowly the junior turned into someone who could be trusted with the second draft, then the whole matter.

Agentic AI now does a competent version of that first draft in minutes. That is a productivity story for the firm and an awkward one for the person who used to hold the pen. The question is not whether the work still needs doing. The question is what happens to the on-ramp when the on-ramp is the part the software is best at.

The First Draft Was More Than a Draft

Entry-level knowledge work has typically done two jobs at once. It produced a deliverable the firm could bill for, and it trained the person producing it. A first-year analyst building a model learned the business by building the model. A junior associate reviewing discovery learned what mattered by seeing what showed up.

When an agent handles the mechanical layer, only the first of those jobs is obviously covered. The deliverable arrives, often faster and cleaner than a tired 24-year-old would have produced it. Most of this only works, though, if the agent is embedded in the firm’s own systems, retrieval on the firm’s documents, connectors into the systems of record, a review interface that shows the reasoning. Teams that lack the engineering bench for that work often bring in outside help, whether an internal platform group or Chicago custom AI developers who have built these plumbing layers before.

The Data Is Starting to Show It

This is not a hunch anymore. Research from the Stanford Digital Economy Lab found that employment among workers ages 22 to 25 in highly AI-exposed occupations sits roughly 19% below where it would be if it had tracked employment among similarly aged workers in less-exposed jobs. Older workers in the same occupations are not seeing the same drop. Whatever is happening is concentrated at the entry point.

The pattern lines up with how firms actually adopt these tools. When agentic systems substitute for human tasks, headcount for those tasks softens. When they complement the person doing the work, employment holds up or grows. The trouble for juniors is that the tasks a two-year-old career can handle are, almost by definition, the tasks most amenable to substitution.

Adoption Is Faster Than the Org Chart

The speed matters. Anthropic’s own tracking of how its models are used in the economy shows automation-style usage now edging past augmentation-style usage for the first time, meaning more prompts look like “do this task” than “help me do this task.” Read the Anthropic Economic Index for the underlying breakdown; the shift is not enormous yet, but the direction is clear and it is recent.

Corporate adoption is moving on a similar clock. Agentic pilots that were slideware in 2024 are running in production in 2026, and the org charts have not caught up. Most companies still hire, promote, and train on the assumption that the bottom of the pyramid does the first pass. If the pyramid loses its bottom row, the shape has to change.

What Firms Should Actually Do About It

The wrong response is to freeze junior hiring and declare the problem solved by attrition. That saves money for a year and starves the firm of mid-level talent for a decade. The better response is to redesign what a junior does so the role survives contact with the agent. A few practical moves are worth naming:

Hire juniors to supervise agents, not compete with them. Treat the first-year role as a reviewer, editor, and prompt designer for agent output. That is a real skill, it produces real judgment, and it maps onto how the work will actually get done.
Rebuild the feedback loop. When the agent writes the first draft, someone senior still has to explain why the second draft is different. Bake that critique into the workflow instead of assuming juniors will absorb it by osmosis.
Widen the aperture on backgrounds. If the technical first draft is commoditized, the differentiator is judgment, communication, and domain feel.
Instrument what the agent does. Log the prompts, the outputs, and the human edits. That log is the training material the next cohort of juniors will actually learn from.

The Path Up Has to Be Redrawn

The uncomfortable truth is that the traditional apprenticeship model was subsidized by work that was mostly rote. Firms tolerated the cost of a first-year because the first-year did things a partner did not want to do. Agents now do those things faster and cheaper, and the subsidy is gone. If firms want the next generation of seniors, they may have to pay for the apprenticeship directly instead of hiding it inside billable work.

That is a harder conversation than a headcount decision. It is also the one worth having now, before a cohort of would-be juniors quietly routes around the industries that no longer have a first rung.

Illinois Data Center Study Projects 121,000 Jobs, but Only 2,800 Would Be Permanent

The Chicago Journal Staff 2 on October 3, 2026

A study released October 1 by the Illinois Economic Policy Institute and the University of Illinois Project for Middle Class Renewal projects that $57 billion in new data center investment through 2035 would support 121,000 jobs across the state. Only about 2,800 of those would be direct, permanent positions. Residential electric bills could rise about $12 a month.

Key Takeaways

Most of the 121,000 projected jobs are temporary construction roles or jobs created indirectly. Consumer spending alone accounts for 41,751 of them.
Operating data centers could generate nearly $300 million a year in property tax revenue, which could lower local residential property taxes by 3% to 10% if no abatements are granted.
The study’s mid-level scenario adds 6.5 gigawatts of electricity demand to the Illinois grid by 2035 and raises overall electric system costs by almost 10%.
Grid operator PJM forecasts 10.7 gigawatts of new data center demand by 2035 in ComEd’s northern Illinois territory alone, above the study’s statewide estimate.
Illinois already has more than 240 planned or active data centers, about 19 per million residents.
The report puts forward 10 policy proposals, including tighter qualifications for state tax incentives and a ban on local property tax breaks.

Study Separates Construction Jobs From Permanent Payrolls

The 121,000 figure needs some context. Data centers employ large crews while they are being built and very few people once they are running. The Illinois Economic Policy Institute (ILEPI) study shows that clearly: of the projected jobs, about 2,800 are direct, permanent positions. Divided across the full $57 billion in projected investment, that comes to roughly $20.2 million in developer spending for each long-term job, a calculation Capitol News Illinois noted in its coverage.

The construction work is not minor, though. The report found that 13 data center projects in 2020 and 2021 created 8,000 construction jobs paying more than $100,000. For the Chicago region’s building trades, a decade of data center construction would mean years of steady, well-paid work, even if the finished buildings need few people to run them.

The 41,751 consumer-driven jobs come from that spending moving through the wider economy, from restaurant workers near job sites to factories making construction materials. ILEPI is a labor-backed research organization, and its economist Frank Manzo has argued that a moratorium would push this investment to neighboring states. Critics at Good Jobs First, a national group that opposes corporate subsidies, counter that public incentives should be tied to lasting employment rather than temporary construction booms.

Property Tax Revenue Depends on Local Abatement Decisions

The study’s property tax findings may matter most for Chicago-area suburbs. Once operating, data centers could add nearly $300 million a year to local tax bases across Illinois. Researchers estimate that could reduce residential property taxes by 3% to 10%, but only where local governments do not offer abatements.

The report uses DeKalb County as an example. In 2024, a large data center there paid $31 million in property taxes. That year, average property taxes in the county rose about 2%, compared with about 6% statewide. Robert Bruno, the University of Illinois professor who co-authored the study, described data center tax revenue as a sizable and reliable source of school funding. He also said each community should decide for itself whether to pursue it.

Abatements change the math. Because Illinois limits how much local governments can raise property taxes in a single year, tax breaks for data centers can wipe out much of the relief homeowners would otherwise see. The report recommends banning local property tax breaks for data centers so that more of the revenue reaches residential taxpayers.

Northern Illinois Ratepayers Face the Sharpest Energy Questions

The report’s energy findings matter most in northern Illinois. Its mid-level scenario estimates that average residential electric bills would rise about $12 a month, or about $150 a year, as data centers add 6.5 gigawatts of demand statewide by 2035.

Other forecasts put demand higher. PJM, the regional grid operator that manages power in northeastern Illinois, projects data center demand in ComEd’s service territory alone will grow by 10.7 gigawatts by 2035, though not every proposed project will be built. The rest of Illinois is served by the MISO grid, where peak demand across 15 states is expected to rise about 35% by 2035. The Citizens Utility Board has warned that higher electric bills take money out of household budgets that would otherwise be spent at local businesses.

That gap between forecasts is the main uncertainty for Chicago households. The study’s $12 monthly figure is a midpoint estimate. Whether actual bills end up near it or well above it depends on how much of the proposed data center pipeline gets built and how the costs of grid upgrades are divided.

Report Lands as Springfield Weighs Data Center Rules

The study comes at a decision point. Lawmakers did not pass data center regulations in the spring session, and the state’s data center tax incentive program is currently paused. The Illinois Department of Commerce and Economic Opportunity estimates incentives given to data centers had exceeded $660 million in value as of 2025.

The report’s 10 proposals include raising the investment and job thresholds needed to qualify for state tax breaks, shortening how long incentives last, and banning nondisclosure agreements between developers and local governments. Several proposals line up with the POWER Act, which would require data centers to secure their own renewable energy, obtain water-use permits, and file transparency reports. Supporters hope the bill will move during the fall veto session. Environmental groups such as the Sierra Club of Illinois have argued for mandatory standards rather than incentives alone.

For Chicago, the question is not whether data centers arrive. More than 240 are already planned or operating statewide. The study’s value is in laying out what the region would gain and what it would pay.

FAQs

How many jobs will Illinois data centers create?

The ILEPI and University of Illinois study projects 121,000 jobs from $57 billion in data center investment through 2035. About 2,800 of those would be direct, permanent positions. Most of the rest are temporary construction jobs or jobs created indirectly.

Will data centers raise electric bills in Illinois?

The study’s mid-level scenario estimates an increase of about $12 a month, or $150 a year, for average residential customers. Other grid forecasts suggest the impact could be higher depending on how many projects are built.

Do data centers lower property taxes?

The study estimates operating data centers could generate nearly $300 million a year in property tax revenue statewide. That could reduce residential property taxes by 3% to 10% where local governments do not grant tax abatements.

How many data centers are in Illinois?

Illinois has more than 240 planned or active data centers, about 19 for every million residents.

Who produced the Illinois data center report?

The Illinois Economic Policy Institute and the University of Illinois Project for Middle Class Renewal co-authored the study, which was released October 1, 2026.