A family story can sound complete long before the evidence is complete. A name fits. The dates look possible. Someone supplies a reason for a move, and the whole account starts to feel settled.
AI can make that account read smoothly while getting its facts wrong. The Coalition for Responsible AI in Genealogy warns about false and incorrect content. It asks researchers to check accuracy against other records. Read the coalition’s principles .
The danger is a claim that loses its uncertainty as it passes between people. Suppose an old note says a man was “perhaps a cousin”. A later account calls him a cousin. A tree then makes the relationship look established. This is folklore laundering: repetition removes the question while adding no evidence.
The example above is invented to explain the problem. The three cases below come from Edward Woodcock’s family archive. Each shows why a useful assistant needs a method for correcting its answers.
Before you type a family name into any AI, turn off training on your account; the privacy guide shows you where.
What a convincing answer leaves out.
Consider the sentence “William moved north to take a better job”. It contains several claims. William moved. The new address lay north of the old one. A job changed. The job improved his position. The job caused the move.
Two dated addresses might support only part of that sentence. A new occupation might support another part. The motive still needs its own evidence. Treating the sentence as one fact hides those differences.
Ask an assistant to separate the claims before asking it to write a life story. Keep its suggested sources on a search list until you have opened them. A plausible title or archive reference is not evidence that a record exists.
If the assistant cannot show a source, the claim remains UNCONFIRMED. Asking the same question again does not change that status. Nor does a confident explanation of why the answer sounds right.
Use three labels with precise meanings.
VERIFIED means that you inspected the cited source and checked what it supports. Keep the statement narrow. “The entry names Charles as a lawyer” says what the source records. It does not settle every question about Charles’s identity.
INFERRED means that a conclusion depends on reasoning from evidence. Record that reasoning. Name the alternatives you considered and the evidence that favours your conclusion. An inference can be strong, but readers still need its basis.
UNCONFIRMED means that a claim has not met those checks. It may be a family memory, a suggested identification or a source you cannot yet inspect. Keep it available as a question without adding it as a settled relationship.
The labels describe individual claims. They do not certify an entire person or branch. A birth date may be supported while the parent link remains open. A correctly copied record can preserve a mistake made by its original author.
Keep the source wording beside your conclusion. Add the date of your review. If new evidence changes the conclusion, record the change and its reason. A label without that trail can make a weak claim look better researched than it is.
The older standard behind the method.
The Genealogical Proof Standard asks researchers to do five things: research the question widely enough, cite sources accurately, compare evidence, resolve conflicts and write a reasoned conclusion. FamilySearch’s explanation of the standard describes these requirements .
These labels help you organise the work. They do not replace the standard. In particular, attaching a source does not resolve a contradiction. You must explain why the competing evidence differs, or leave the conclusion open.
Nor does a free search route promise to complete every investigation. If an essential record remains inaccessible, record that limit. “I have checked the sources available to me” is different from claiming that every relevant possibility has been examined.
The coalition adds principles for AI use: accuracy, disclosure, privacy, education and compliance with relevant rules. Its published principles explain each . A useful disclosure tells readers what the assistant did and what the researcher checked. It does not hand responsibility to the software.
Case 1. The colonel who was also a lawyer.
Gwen Hinton, Ed’s great-aunt, sailed to America in 1946. Her sponsor appeared in an immigration record as an army colonel. Later records showed Gwen working for a lawyer in Flint, Michigan. Were these two different men?
The archive connected Colonel C. M. Van Benschoten with attorney Charles M. Van Benschoten. A 1944 Judge Advocate journal supplied the missing occupational link. The colonel was a military lawyer. The two descriptions could belong to the same man without either record being wrong.
The evidence trail in Ed’s account names an immigration record, a military journal and a later legal source.
This is where comparison helps. The name suggested a possible match. A record linking the occupations gave the archive a reason to join the identities. Inventing a career change would have filled the same gap with an unsupported story.
Keep the claims separate. The records’ descriptions belong under VERIFIED in the archive’s account. The identity conclusion needs the reasoning that connects them. Anyone reviewing it should be able to follow those references.
Another question remains open. Family memory suggested a romance between Gwen and her employer. Evidence of sponsorship and employment does not establish that relationship. The archive keeps the romance UNCONFIRMED. Solving the occupational puzzle did not solve the private one.
Case 2. The brother with his mother’s surname.
In 1880s Wigan, Elizabeth Halliwell, later Thomas, lived with a prosperous, childless uncle. Her parents lived nearby with her brothers. The arrangement prompted an obvious question: why was this child raised elsewhere?
An early theory treated Elizabeth as the eldest child. The archive then identified an older brother, George, in the 1881 census. His birth registration used his mother’s surname, Sharples. The chronology placed his birth before the parents’ marriage.
George remained with his parents. Elizabeth, born after the marriage, lived with the uncle. That evidence overturned the earlier eldest-child theory. It did not explain the parents’ decision.
The supplied account identifies census and birth records. The earlier theory remains attached to the question, so the correction remains visible.
A surname on its own would not prove this sequence. The dates and family identifications matter. Equally, the household arrangement does not let us read anyone’s intentions. Avoid adding shame, preference or an agreement with the uncle because one would make a satisfying explanation.
Record the earlier inference as superseded, with its original reasoning intact. Then write the revised position. The older brother changes the order of the children. The reason Elizabeth lived elsewhere remains UNCONFIRMED.
That distinction also protects the parent link. Being raised by an uncle does not make him a parent. A household relationship and a parent-child relationship need separate evidence, even when a family story blurs them.
Case 3. The photograph that did not show Ed’s grandfather.
Ed found a cropped RAF training photograph that might show his grandfather, William John Woodcock. The resemblance was tempting. The archive stored the proposed match as a low-confidence inference.
A second copy preserved the full name key, reading from left to right. It accounted for all ten men. None was William. The photograph also showed pupils, whereas William was an instructor.
The evidence trail names the cropped photograph and the full copy. It also records the evidence that rejected the identification.
The name key did the decisive work in Ed’s account. The instructor detail supported the rejection, but would not by itself exclude someone from every group photograph. State only what this evidence settles: this picture does not identify William among its ten named subjects.
Preserve the cropped version and the reason it once looked promising. Attach the fuller copy and the correction. Mark the proposed match as rejected. Do not leave it circulating with a softer caption such as “probably William”.
This correction matters because the first guess belonged to the researcher. The method has to test the family’s preferred answer as carefully as a stranger’s claim.
What these examples establish.
The stories above are drawn from Ed’s cleared archive notes. The evidence trail remains in his archive. The original records were not independently reopened for this guide.
They show three different outcomes. Evidence can connect identities, overturn a theory or reject an identification. In each case, the useful result depends on keeping the question separate from the answer we would like.
A correction is part of the research record. Keep enough detail for your future self to understand why you changed your mind. Otherwise an old note can revive the same mistake years later.
Step 1. Give the assistant a bounded task.
After turning off training and removing living-person details, start with a short passage you are entitled to share. Use this instruction:
Separate this passage into individual claims. Record the source supplied for each. Mark unsupported claims UNCONFIRMED. Explain each inference. List contradictions and questions separately. Do not add people or invent citations.
Read the result against your passage. Check for a changed date, an added relationship or a confident interpretation of an unclear word. Correct the working note before asking for a narrative.
Step 2. Check the claim before publishing it.
Open the cited source yourself. Compare the exact wording with the claim. If you cannot inspect it, preserve the limitation. An assistant’s report of what a source says is not your own inspection.
For a disputed identity, write down what would disprove the match. Then search for that evidence. A conflicting household, a complete photograph caption or a differently named parent can change the conclusion.
Use the verification checklist before adding the result to a shared tree. The kit in development will teach this method; it cannot remove the need for your judgement. Join the list if you would like news.