Study issues clear signal that a meaningful portion of the work behind every file is now technically automatable with today's tools
For Canada’s mortgage professionals, artificial intelligence is no longer just something happening on Wall Street or in Silicon Valley. It is arriving in brokerages, credit unions and bank mortgage centres — and new research suggests it could already perform a significant share of the work that goes into every deal.
MIT’s Project Iceberg, a large‑scale simulation of the U.S. labour market, finds that current AI tools are technically capable of performing tasks worth 11.7% of total wage value, or about US$1.2 trillion annually. The model represents roughly 151 million workers across 923 occupations and more than 32,000 skills.
READ MORE: Yes, AI is smarter than you sometimes, but can it help clients with mortgages?
The study is American, but the underlying work it measures — financial analysis, document handling, administrative coordination — looks very familiar to anyone in the Canadian mortgage business.
What Project Iceberg actually measures
Project Iceberg doesn’t just speculate about “jobs of the future.” It compares, skill by skill, what workers do today with what existing AI systems can already do.
Each worker is modelled as an “agent” with a bundle of skills and tasks, matched against thousands of AI tools.
The core metric, the Iceberg Index, measures technical exposure: the share of an occupation’s wage bill tied to skills where AI has already shown it can perform occupational tasks. The authors emphasise that it “captures technical exposure, where AI can perform occupational tasks, not displacement outcomes or adoption timelines.”
For Canadian mortgage brokers and lenders, that means 11.7% is not a prediction of layoffs. It is a signal that a meaningful portion of the work behind every file — especially the repetitive, rules‑based parts — is now technically automatable with today’s tools.
The big exposure is in back‑office, not just in “tech jobs”
Most of the public conversation has focused on AI writing code and replacing some technology roles. In Iceberg’s language, that is just the visible “surface.”
Looking only at current AI adoption in computing and technology occupations, the researchers find exposure of about 2.2% of wage value, or roughly US$211 billion — the “Surface Index.” They stress this is “only the tip of the iceberg.”
Beneath the surface, technical capability “extends far below the surface through cognitive automation spanning administrative, financial, and professional services,” bringing total exposure to 11.7% and about US$1.2 trillion in wages.
One summary points out that high Iceberg scores are driven by “cognitive work—financial analysis, administrative coordination, and professional services.” If you think about the typical Canadian mortgage file — incomes to verify, debts to tally, policy rules to check, documents to chase — that description comes very close to home.
Where AI can already handle mortgage work
The report doesn’t talk specifically about Canada or mortgages, but the task patterns it highlights line up cleanly with what happens every day in broker offices, branch networks and underwriting centres.
The highest near‑term automation potential is in structured, repeatable, digital tasks such as:
Document and data processing
- Reading payslips, T4s, Notices of Assessment, bank statements and employment letters
- Extracting income, liabilities and key flags into LOS and underwriting systems
- Classifying documents and flagging missing or inconsistent information
The authors note that financial institutions already use AI for “document processing and analytical support,” while healthcare systems are automating “administrative tasks.” Mortgage workflows combine both patterns: financial calculations anchored in heavy documentation.
Rules‑based checks and case triage
- Eligibility screening against product matrices and risk policies
- Standard debt‑service and stress‑test calculations
- Simple case routing (prime vs alternative, straightforward vs complex, broker vs retention)
Standard communication and reporting
- Generating pre‑approvals, conditions lists and compliance disclosures
- Producing pipeline, pull‑through and performance reports from defined templates
In these areas, AI is not guessing; it is applying logic to structured data, which is exactly what Project Iceberg measures.
This doesn’t mean entire roles disappear. It means:
- The most exposed work sits in high‑volume processing roles — file setup, document review, basic underwriting and post‑funding administration.
- Human effort will be pushed towards edge cases, complex income structures, self‑employed borrowers, multi‑property investors, vulnerable customers, and relationship work with borrowers and referral partners.
What this means for the Canadian mortgage profession
Project Iceberg was designed so governments can “identify exposure hotspots, prioritize training and infrastructure investments, and test interventions before committing billions to implementation.” For Canadian mortgage employers and intermediaries, it offers a similar opportunity: use the data to rethink your operating model before cost pressure forces blunt automation.
Three questions stand out.
1. Where in your pipeline is the “11.7%” hiding?
Because Iceberg is skills‑based, exposure varies inside roles. For a typical Canadian lender or large brokerage:
- What share of staff time is spent on pure data entry and document chasing?
- How many decisions really require human judgment versus applying published rules?
- How much capacity is tied up in rework and manual checks that a well‑designed AI system could streamline?
You don’t need MIT’s supercomputer to answer those questions — just a frank process review grounded in the kinds of tasks the report highlights.
2. Will you use AI mainly for cost‑cutting, or to compete on speed and advice?
With margins under pressure and compliance costs rising, there will be strong temptation to use AI primarily to strip out cost — especially in centralised underwriting and servicing.
The alternative is to:
- Let AI absorb growth in routine volume, reducing cycle times and error rates
- Redeploy people into higher‑value work: complex deals, retention, proactive advice on renewals and refinances, and outreach to underserved communities
In a market where borrowers are frustrated by delays and opaque decisions, speed and clarity may be as important a competitive edge as a slightly lower rate.
3. How will you keep regulators and customers onside?
Canada’s regulators are watching AI adoption closely. Although Project Iceberg focuses on capability, not policy, its message is clear: technical exposure is already large, and traditional metrics “such as GDP, income, and unemployment explain less than 5% of this skills-based variation.”
That should prompt mortgage leaders to think hard about:
- Explainability: if AI tools influence credit decisions, can you clearly explain those decisions to borrowers and to OSFI / provincial regulators?
- Fairness: are models tested against bias across regions, income types and protected groups?
- Resilience: what happens to service and risk management if an AI‑heavy workflow fails or behaves unexpectedly?
A practical roadmap for Canadian mortgage businesses
To turn Iceberg’s findings into an advantage — rather than a future headline — Canadian mortgage employers can take four practical steps.
1. Map your own exposure by process and role
For brokers, lenders and monolines alike:
- Decompose roles into tasks: gathering docs, inputting data, checking policy, talking to clients, chasing conditions, liaising with lawyers and insurers.
- Mark each task as high / medium / low AI suitability, using Iceberg’s exposed areas (document processing, administrative work, standard financial analysis) as your guide.
Even a rough estimate of how much of your payroll sits in high‑suitability tasks will clarify where AI investments will bite first — and where people risk is highest.
2. Run scenarios before you “commit billions” to platforms
Borrowing Iceberg’s spirit, model 3–5 year scenarios:
- Conservative: AI mainly as a copilot, reshaping workflows but not headcount
- Targeted automation: a defined share of high‑suitability tasks moves to AI, with staff redeployed towards advice and complex cases
- Aggressive automation: extensive straight‑through processing for standard deals, with corresponding role redesign
For each scenario, consider impacts on:
- Turnaround times and borrower / Realtor / broker satisfaction
- Cost per file and overall profitability
- Compliance and conduct risk
- Talent: which roles shrink, which grow, and what reskilling is required
3. Decide your people and brand position now
Technology choices are only half the story. The other half is what kind of employer and market player you want to be as AI rolls through the profession:
- Will you commit to retraining existing staff for higher‑value roles where possible?
- Will you position your brand around human advice powered by AI, or around ultra‑fast, low‑touch digital processing?
- How will you communicate the changes to brokers, agents and customers to avoid the perception that “the computer says no”?
4. Put AI exposure on the agenda — not just in IT, but in the C‑suite
Finally, follow the lead of Iceberg’s authors and treat this as a leadership‑level planning issue. Executives and boards should regularly see:
- An estimate of the share of labour cost tied to high‑exposure tasks today
- The planned trajectory of that share under different AI adoption paths
- The associated talent, regulatory and reputational risks — and how you intend to manage them
The MIT team behind Iceberg is blunt: “The window to treat AI as a distant future issue is closing.”
For the Canadian mortgage profession, that window is narrowing fast. AI can already shoulder a meaningful portion of the work behind every approval and decline. Whether that turns into faster, fairer, more transparent lending — or a wave of rushed automation and unintended consequences — will depend on the choices lenders and brokers make now, while there is still time to steer the market in the right direction.
