I like to check occasionally to see if the current Artificial Intelligence Large Language Models (LLM) have improved over time.
One of the genealogy records that every researcher uses in Census records. How will an LLM (in this case, OpenAI ChatGPT5 - the latest FREE model) do in transcription and summarization of a fairly simple task that almost every genelaogist can and has performed hundreds of times.
Here is the 1900 United States Federal Census record with the Thomas Richmond family (Thomas and his wife are my great-grandparents and their daughter Alma Bessie Richmond is my paternal grandmother):
The Thomas Richmond family appears on the 1900
U.S. Census, Leominster, Worcester County, Massachusetts,
enumerated 1 June 1900 by Charles T. Dodge.
They are recorded as [House] Dwelling 42, [Dwelling Number 53], Family [Number] 88. The
household consists of Thomas H. Richmond, his wife Juliet E., and
their six children.
Below is a complete transcription and summary for each family member.
1. Thomas H. Richmond (Head)
Transcription
Name: Thomas
H.Richmond [Note: Thomas had no middle name or initial. There is a handwritten number over his name]Relationship: Head
Color/Race: White
Sex: Male
Birth:
January 1849[June 1848]Age: 51
Marital Status: Married
Years Married: 31
Place of Birth: England
Father's Birthplace: England
Mother's Birthplace: England
Year of Immigration:
1874]1856]Years in the United States:
26[44]Naturalization: Naturalized
Occupation: Overseer of
Weavers[Woolen Mill]Months Unemployed: 0
Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Home Ownership: Renting (R)
Farm or House: House
Summary
Thomas H. Richmond was born in January 1849 [June 1848] in England
and was 51 years old at the time of the census. He
had been married to Juliette for 31 years, indicating
a marriage about 1868–1869. Both of his parents were also born in
England. He immigrated to the United States in 1874 [1856],
had lived in the country for 26 [44] years, and had
become a naturalized U.S. citizen. Thomas worked as
an overseer of weavers [woolen mill], a supervisory position in
the textile industry, and had experienced no unemployment
during the preceding year. He was literate, able to read, write, and
speak English. The family rented their home in
Leominster.
2. Juliet [Juliaette] E. Richmond (Wife)
Transcription
Name:
Juliet[Juliaette] E. RichmondRelationship: Wife
Color/Race: White
Sex: Female
Birth: September
1854[1848]Age:
45[51]Marital Status: Married
Years Married: 31
Children Born: 9
Children Living:
9[1] [-- it looks like a "1" to me, it should be 8]Place of Birth:
England[Connecticut]Father's Birthplace:
England[Rhode Island]Mother's Birthplace:
England[Rhode Island]Year of Immigration:
1874[blank]Years in the United States:
26[blank]Naturalization:
Alien[blank]Occupation:
Overseer of Weavers (appears to read "Overseer of Weavers")[blank]Months Unemployed:
0[blank]Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Juliet [Juliaette] E. Richmond was born in September 1854 [1848] in England [Connecticut] and was 45 [51] years old. She had been married for 31
years and reported having given birth to nine
children, all nine [one] still living, an important genealogical
statement. Like her husband, she immigrated in 1874
and had resided in the United States for 26 years,
but she was still recorded as an alien, indicating
she had not become a naturalized citizen [She was a natural-born United States citizen]. She was literate and spoke
English. The occupation entry appears to indicate she also worked as
an overseer of weavers, suggesting she participated
in the local textile industry. [No occupation was listed.]
3. Grace Richmond (Daughter)
Transcription
Name: Grace
Relationship: Daughter
Color/Race: White
Sex: Female
Birth: August 1876
Age: 23
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace:
England[Connecticut]Occupation: Shirt Maker
Months Unemployed: 0
Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Grace Richmond, age 23, was born in August 1876 in Connecticut, making her the oldest child in the household. She was unmarried and employed as a shirt maker, another occupation associated with New England manufacturing. She was literate and had not attended school during the previous year.
4. Emily W. Richmond (Daughter)
Transcription
Name: Emily W.
Relationship: Daughter
Color/Race: White
Sex: Female
Birth: June 1879
Age: 20
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace: Connecticut
Occupation: None listed
Months Unemployed:
0[blank]Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Emily W. Richmond was born in June 1879 in Connecticut
and was 20 years old. She was single and living at
home. No occupation is recorded for her. She could read, write, and
speak English. (One birthplace entry appears to differ from the
others by giving her mother's birthplace as Connecticut; this may be
an enumerator error since the mother's own entry clearly states
England.)
5. Charles E. Richmond (Son)
Transcription
Name: Charles E.
Relationship: Son
Color/Race: White
Sex: Male
Birth: August 1880
Age: 19
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace: Connecticut
Occupation: Painter
Months Unemployed: 0
Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Charles E. Richmond was 19 years old, born in
August 1880 in Connecticut. He was unmarried and
worked as a painter. He was fully literate and spoke
English. As with Emily's entry, the mother's birthplace appears to
have been written as Connecticut, likely an inconsistency by the
enumerator.
6. James F. [Alma B.] Richmond (Daughter? actually Son)
Transcription
Name:
James F.[Alma B.]Relationship:
SonDaughterColor/Race: White
Sex:
Male[Female]Birth: February 1882
Age: 18
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace: Connecticut
(appears written)Occupation: House Work
Months Unemployed: 0
Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
James F. [Alma B.] Richmond was born in February 1882 in Connecticut
and was 18 years old. He [She] remained single and lived
with his [her] parents. His [Her] occupation is recorded as House Work,
indicating he [she] was performing work in or around the household rather
than outside employment. He [She] could read, write, and speak English.
7. Edwin J. Richmond (Son)
Transcription
Name: Edwin J.
Relationship: Son
Color/Race: White
Sex: Male
Birth: December 1883
Age: 16
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace: Connecticut
(appears written)Occupation: Laundryman
Months Unemployed: 0
Attended School: No
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Edwin J. Richmond was 16 years old, born in December 1883 in Connecticut. Although still a teenager, he was employed as a laundryman, reflecting the early entry of many young people into the workforce around 1900. He was literate and had not attended school during the census year.
8. Marcus [James] H. Richmond (Son)
Transcription
Name:
Marcus[James] H.Relationship: Son
Color/Race: White
Sex: Male
Birth: November 1886
Age: 13
Marital Status: Single
Place of Birth: Connecticut
Father's Birthplace: England
Mother's Birthplace: Connecticut
(appears written)Occupation: At School
Months Unemployed:
10[blank]Attended School: Yes, [10 Months]
Can Read: Yes
Can Write: Yes
Can Speak English: Yes
Summary
Marcus James H. Richmond was born in November 1886 in
Connecticut and was 13 years old. Unlike
his older siblings, he was still attending school, with his
occupation recorded as "At School." He had
attended school during the year [for ten months], could read and write, and spoke
English. His entry illustrates the transition within the family from
older children already working to the youngest still receiving an
education.
Household Summary
The 1900 census depicts a hardworking immigrant family headed by
Thomas H. Richmond, an English-born textile
supervisor who had become a naturalized American. His wife Juliet [Juliaette] reported that all nine [one?] of her children were still living,
an especially valuable genealogical detail. The children born in
Connecticut reflect the family's early years in New England before
moving to Leominster, Massachusetts. Most of the older children were
already employed in textile or manufacturing-related occupations,
while the youngest son, Marcus [James], remained in school. The family rented
their home, was fully literate, and was well established in the
community after approximately twenty-six [forty-four] years in the United States.
===============================
There are 124 transcription lines and at least 23 of them have errors. Most of the errors are on the parents, Thomas and Juliaette. Almost all of the rerrors are "unforced" -- a human eye can read them well.
The errors by ChatGPT5 surprised me -- it misread five names, and got many details wrong. I thought it would do better. I did not try to tell it to correct everything that I found wrong.
I did this task also on Google Gemini 3, and it made many mistakes also, but different from ChatGPT5. The MyHeritage Scribe AI Tool did a much better job on the transcription (but it was not perfect).
This human does a better job in deciphering census returns than at least two AI LLMs do. It's not worth my time to do this task using AI -- I can do it faster and more accurately myself at this time. The MyHeritage Scribe AI Tool did a much better job on the transcription (but it was not perfect).
I will try to revisit this exercise in a year or so from now and see if the LLMs have improved.
Links to my blog posts about using Artificial Intelligence are on my Randy's AI and Genealogy page. Links to AI information and articles about Artificial Intelligence in Genealogy by other genealogists are on my AI and Genealogy Compendium page.

No comments:
Post a Comment