AI replacement claims: what should a buyer ask?
Last reviewed May 22, 2026
AI replacement claims say a product can substitute for a person, team, or professional service. This page focuses on the evidence needed when a vendor claims AI can take over expert work rather than support it.
Evidence buyers verify
- A narrow list of tasks the AI performs, with tasks it does not perform stated beside it.
- Quality testing that compares AI output with an appropriate human or professional review baseline.
- Error categories, correction workflow, and escalation steps when the AI output is incomplete or wrong.
Opens the checker for this claim type. Paste your vendor's exact wording there. Evidence questions only — not a blacklist or fraud detector. Not sure what a result looks like? See a sample receipt.
Sources this guide draws from
- · February 11, 2025
Source for robot lawyer and human-lawyer substitution claim evidence.
- · September 25, 2024
Source for replacement wording tied to legal documents and professional-service substitution.
Public claims with documented evidence gaps
"the world's first robot lawyer"
Automation / Replacement- Source and date
- FTC DoNotPay final order release · February 11, 2025
- Evidence signal
- Professional role language that may imply expert-service substitution.
- Evidence gap
- A buyer needs task scope, output-quality testing, professional review, and limits for law-related use.
- Buyer question
- For the robot lawyer claim, which tasks were tested against qualified professional review?
"sue for assault without a lawyer"
Automation / Replacement- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- No-professional-needed wording for a high-stakes workflow.
- Evidence gap
- A buyer needs evidence that the exact workflow was tested, reviewed, and limited to appropriate use cases.
- Buyer question
- For the without-a-lawyer claim, what review step catches errors before a user relies on the output?
"generate perfectly valid legal documents in no time"
Automation / Replacement- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- Perfect-output and speed wording for documents that may require expert judgment.
- Evidence gap
- A buyer needs test examples, error rates, document categories, jurisdiction limits, and reviewer qualifications.
- Buyer question
- For the perfectly valid document claim, which document types and jurisdictions were included in testing?
"replace the $200-billion-dollar legal industry with artificial intelligence"
Automation / Replacement- Source and date
- FTC Operation AI Comply announcement · September 25, 2024
- Evidence signal
- Industry-wide replacement claim without a task or boundary.
- Evidence gap
- A buyer needs a narrow list of supported tasks, excluded tasks, quality controls, and escalation rules.
- Buyer question
- For the replace-an-industry claim, which professional tasks remain outside the AI workflow?
Match each claim pattern to the evidence buyers need
| Claim pattern | Evidence needed | Buyer question |
|---|---|---|
| Replace a professional, team, or department | Task list, user role, reviewed output samples, failure handling, and escalation path. | Which work is fully automated, and which work still needs qualified review? |
| No lawyer, doctor, analyst, or agent needed | Scope limits, regulated-task exclusions, reviewer qualifications, and user-warning language. | What step prevents users from treating the output as a final expert decision? |
| Perfect or instant expert documents | Document categories, test cases, error analysis, jurisdiction or policy limits, and revision process. | How often did the system produce a document that needed material human correction? |
| AI operates at human or expert level | Comparison group, benchmark design, task difficulty, sample size, and acceptance criteria. | Was the output compared with qualified experts on the same task and context? |
Evidence to request
- A narrow list of tasks the AI performs, with tasks it does not perform stated beside it.
- Quality testing that compares AI output with an appropriate human or professional review baseline.
- Error categories, correction workflow, and escalation steps when the AI output is incomplete or wrong.
- Human review boundary, reviewer qualifications, and the point where responsibility shifts back to the user.
Questions to put in front of the vendor
- For this replacement claim, which exact role or task is the AI said to replace?
- Was the output tested against qualified human review, or only against internal acceptance checks?
- Which use cases are excluded because the output may need professional judgment?
- What happens when the AI cannot complete the task or produces a materially wrong answer?
Wording boundaries to compare against
- Drafts first-pass documents for review by a qualified person.
- Automates specified intake and drafting steps, with exceptions routed to human review.
- Supports professional workflows by suggesting text or next steps; final review remains separate.
- Designed for defined low-risk tasks, not as a substitute for all professional judgment.
Have your vendor's exact claim wording ready?
Check an AI replacement claim How the evidence method works