If you are asking “Will AI replace junior architects?”, the unsettling question is not whether a machine can produce a striking image. It is whether the work that used to teach a junior architect how a building gets made will still be there. Some firms already use AI to speed up images, summaries and early design options. That changes what an entry-level employee may be asked to do. It does not tell us how many junior jobs will disappear.
The useful response is to make your learning visible. A firm can buy a faster image. It still needs people who can explain a decision, check an output against a real brief, coordinate a change and know when to ask a more experienced colleague.
Is AI replacing junior architect jobs?
There is no credible single number for how many junior architecture roles AI will remove. The RIBA AI Report 2026 says AI is used in some projects by 74% of practices in its survey, while also noting concerns about early career paths. That is a measure of reported use among surveyed practices, not a count of jobs lost. AIA guidance likewise identifies drafting, documentation and coordination as activities whose role in early career learning is changing.
Consider tasks rather than titles:
- A mood image can be generated quickly. Someone must still decide whether it reflects the site, programme and budget.
- A meeting summary can be drafted automatically. Someone must check who agreed to what and whether a deadline or instruction was invented.
- A code search can surface likely clauses. Someone must confirm the jurisdiction, edition, amendments and actual source text.
- A drawing mark-up can be suggested by software. Someone must understand how it affects other drawings and consultants.
The risk is uneven. A position built almost entirely around producing generic images may change sooner than one involving site observation, technical coordination and client communication. That is an editorial assessment, not a forecast of headcount.
The hidden problem: losing the work that teaches judgment
Routine work was never valuable only because it took time. Redrawing a detail after feedback can teach why a junction fails. Revising a room schedule can reveal how a programme changed. Walking through a set of consultant comments can show where responsibility moves between disciplines. If the tool completes those tasks before a junior understands them, the practice saves production time but may remove a learning step.
The AIA AI Firm Toolkit recommends deciding which tasks junior staff should still perform themselves to develop judgment. This matters to candidates too. In an interview, ask how a new hire learns to review an AI-assisted output, who explains errors and when they get exposure to actual project decisions. A vague promise that “you will work with exciting AI tools” says little about training.
Here is a more revealing question: “If your team uses AI to create the first pass of a drawing note, who teaches a new assistant to judge whether that note is correct?” The answer should name a person, a review step and an example. That is also a way to assess mentorship; jobs.archi’s article on mentorship in architecture offers related context.
A three-layer skill stack for an AI-era junior
1. Learn the building, not just the image
Choose one modest project type: a small refurbishment, a classroom, a clinic room or a housing unit. Work through its site constraints, circulation, structure, environmental intent, accessibility and drawing coordination. You do not need to master every jurisdiction. You need to demonstrate that you recognise which questions must be answered locally and by whom.
For a portfolio page, show a design decision with its constraint: “The first layout blocked a continuous accessible route, so I moved the service core and tested the revised circulation.” Include the before and after plan, the constraint and your role. A polished image alone cannot tell an employer that story.
2. Learn to verify the machine’s output
Take an AI-generated concept image and compare it with a measured model. Mark any altered opening, implausible span, missing escape path or material that differs from the brief. Keep the comparison as a small exercise, clearly labelled as your own review. Do the same with a draft project summary: highlight every statement that needs a source or client confirmation.
This is a transferable skill because software will change. An employer does not have to trust your favourite tool. They can inspect your checking method. The AIA toolkit’s verification examples cover renderings, specifications and code research; the common principle is that review effort should match the consequences of an error.
3. Learn how decisions move through a team
Ask to sit in on one consultant call, client review or site meeting. Afterwards, write a short private learning note: what decision was made, what information was missing, who owns the next action and which drawing or document will change. Do not include confidential information in a public portfolio. You can create a generic, permission-safe example of the process later.
If you are still a student, rehearse this with a studio project. Ask a classmate to act as the structural consultant and challenge one design assumption. Record how the scheme changed. The exercise is simple, but it demonstrates a part of practice that a generator cannot show on your behalf.
What should a new graduate put in an application?
Build one “decision case” rather than adding five more renderings. Use four panels:
- Brief and constraint: the user’s need, site condition or technical limit.
- Options: two realistic approaches, with a reason to reject one.
- Decision: what you chose, what you had to check and who would approve it in practice.
- Evidence: a plan, detail, schedule or annotated view that supports the choice.
If AI helped generate options, say so. Explain what the tool produced, what you rejected and what you developed yourself. The point is to make authorship and judgment visible. Our guide to an AI architecture portfolio examines how to label AI-assisted images.
On your CV, replace “proficient in AI” with a precise example you can defend: “Compared AI-generated facade options against a measured model and documented three geometry mismatches before presentation.” Use a real exercise or project; do not invent one to fit this line. If the example was academic, label it academic.
A practical 30-day plan before applying
Week 1: choose a target role. Read several current job adverts and write down the actual tasks. An architectural assistant in a small residential office may need very different evidence from a junior in a large health-care team. Check the dates of listings on jobs.archi by country and confirm vacancies on the employer’s site.
Week 2: build one decision case. Use existing work you are allowed to share, or make a clearly identified self-initiated exercise. Keep the brief narrow enough to finish. Ask someone with practice experience which assumption seems weakest.
Week 3: run a verification exercise. Use an AI tool only with information you own or are allowed to enter. Save the original, the generated output and a marked-up review. Document the time spent checking, not just the time saved creating.
Week 4: test your explanation. Give a two-minute account of the project without reading the page. Ask a listener which decisions were yours. If they cannot tell, revise the portfolio. Then tailor the application to two roles whose work genuinely fits your evidence.
This plan will not guarantee an interview. It gives the hiring team something more useful than a list of tools: proof that you can learn and check work.
Questions to ask an AI-using architecture firm
- Which tasks do junior staff still do manually to learn the underlying method?
- Who reviews AI-assisted outputs before they reach a client or submission?
- Can a junior explain or challenge a tool’s result without being seen as slow?
- Will I see a project beyond visualisation, including documentation and construction feedback?
- What project data may staff enter into third-party tools?
- How are mistakes discussed and corrected?
Listen for examples. “We expect everyone to be innovative” is not a training process. An answer such as “a project architect reviews code summaries against the current local source with the assistant” is much more informative.
Frequently asked questions
Should I learn AI tools before Revit, Archicad or drawing fundamentals?
Match the role you want. If the advert asks for a particular BIM or documentation workflow, learn enough of that workflow to demonstrate real output. AI literacy is useful, but a tool cannot compensate for an inability to read a plan or explain a detail. Avoid paying for a course because a headline says every architect needs it; compare course content with live job requirements first.
Will firms stop hiring graduates if AI speeds up junior work?
Some tasks may shrink and some firms may change their staffing, but the available industry evidence does not support a universal claim. Ask firms about their training and project pipeline, and keep several role types in view. A single forum story about a layoff or a hiring freeze is a valuable warning, not a global labour statistic.
What if I have no professional project to show?
Use a self-initiated, accurately labelled case. Show constraints, alternatives and checking. Do not call a conceptual exercise “construction ready” or imply a built result. For broader application preparation, start with the jobs.archi first-job guidance, then verify any linked vacancy is still open.
The takeaway
AI may remove some repetitive steps and change the entry point into architecture. Build evidence around the steps a firm cannot safely skip: understanding a brief, testing a constraint, checking a claim and communicating a decision. Apply to offices that can explain how juniors learn those skills. That is a stronger career plan than betting everything on either a tool or a reassuring prediction.
The post Will AI Replace Junior Architects? Build the Experience Firms Still Need first appeared on jobs.archi.