Stephen Little, the National Genealogical Society's Artificial Intelligence Program Director, has created another useful saved ChatGPT prompt -- "Steve's Speculator" -- to You can see the complete prompt on Facebook in the Genealogy and Artificial Intelligence Group here.
Steve's description of the prompt is:
Fully aware that we are venturing into the deep end of the pool, today we explore possible insightfulness and creativity in large language models. Today we PROMPT our chatbots to EXTRAPOLATE, HYPOTHESIZE, and SPECULATE. Obvious caveats abound: already susceptible to hallucinations (factual misstatements), here we are explicitly directing the AI to envision--to imagine--beyond the four corners of our document (or image or file), so be warned: Here Be Dragons.This prompt pushes the model into creative exploration. While large language models excel at pattern recognition and insight generation, TODAY they cannot verify facts in real-time. Treat their outputs as sophisticated research leads requiring independent verification. When LLMs make factual claims, always confirm through authoritative sources (which language models emphatically are not). Unverifiable information should be considered speculative rather than factual.The full-text of the PROMPT is included below. You can also find a ready-made version of this saved prompt as a free tool in the OpenAI GPT Store: https://chatgpt.com/g/g-rMp27yokY-steve-s-speculatorYou can try this prompt at any chatbot, attaching a document, photograph, record, or file (and at Gemini, audio and video). If you get an interesting response, please share your source attachment and generated summary. You are also welcome to use the PROMPT below to create your own Custom GPT at OpenAI's ChatGPT, Project at Anthropic's Claude, or Gem at Google's Gemini (the various names the vendors give to saved prompts).The "Steve's Speculator" custom PROMPT is:What is this, and what does it mean? Answer with this process: transcribe; describe; then, explain and analyze; finally (and central to this task) extrapolate, hypothesize, and speculate; conclude with a pithy 100-word summary of all of the above.And do ALL that this way:
A) State your initial assessment and plan.
B) Review and critique your plan.
C) Revise and improve your plan.
D) Execute your plan.
I wanted to try this prompt out with one of my ancestors records, and have it asses the record, ranscribe, describe, and explain analyze the facts in the record, and analyze the record, and finish with a short summary.
1) Here are the Bible pages for two generations of my Kemp family in Ontario, which I found in an Orange County, California public library:
I uploaded the image and then wrote "Please help me understand this record."
2) Here is the complete response by the free ChatGPT4 AI tool using the "Steve's Speculator" custom prompt:
Initial Assessment and Plan
This document contains the genealogical records of the Kemp family, presented in the format of a "family Bible record," which often includes births, deaths, and marriages. It appears to be a transcription or facsimile of the Kemp Family Bible, printed in 1877 and now archived with the Orange County, California Genealogical Society. My plan is to transcribe key parts, describe the structure and layout, and analyze the document to highlight family relationships and notable historical details. I'll then speculate on the significance of specific entries, such as any rare historical dates or patterns in family names or locations.
Review and Critique of Plan
This plan will work for a structured understanding, but it could be improved by separating different family branches and focusing on relationships (parent-child, marriages). Also, I should clarify which entries belong to different family generations for clearer context.
Revised Plan
I’ll enhance the analysis by:
- Identifying distinct family branches by generation.
- Focusing on specific names, birth and death dates, and relationships that might indicate the family's migration, occupation, or social status.
- Hypothesizing why certain records are emphasized, such as significant age gaps or unusual dates.
Execution
Transcription and Structure:
Title and Bible Source:
- Title: "KEMP FAMILY BIBLE RECORD"
- Source: Bible printed by A. J. Holman Co., Philadelphia, 1877.
- Dedication: Given to John Evans Kemp, born in Canada. Last entry indicates it’s held by the Orange County Genealogical Society.
Marriage Records:
- Peter Evans Kemp married Catharine F. Ryder on April 6, 1875, in Norfolk, officiated by Rev. J. Van Loon. Witnesses: Gideon V. Ryder and Emeline Powel.
- Gideon P. Kemp married Florence Bragg on November 20, 1921.
- Sarah E. Kemp married Oliver C. Baker on June 1922.
Birth Records (First Page):
- Sarah Elizabeth Kemp: Sept 17, 1877
- Mary Waty Kemp: Dec 10, 1880
- John Evans Kemp: Apr 27, 1883
- Peter Gideon Kemp: July 10, 1885
Death Records (First Page):
- Mary W. Kemp: Nov 26, 1922
- Catherine F. Kemp: Apr 30, 1919 (Age: 74 years, 4 months, 6 days)
- Peter E. Kemp: Apr 10, 1922 (Age: 84 years, 4 weeks, 16 days)
- Florence Kemp: Oct 20, 1922 (Age: 32 years, 4 months)
- Sarah Elizabeth Kemp: Mar 28, 1943 (Age: 66 years)
- John Evans Kemp: Oct 14, 1956 (Age: 73 years)
Continuing Entries (Second Page):
- The record also traces back to earlier generations, listing Abraham Kemp (b. Nov 4, 1795) and Sarah Fletcher, married on April 16, 1818.
- Additional marriages and births trace further back, including:
- Peter Evans Kemp married Isabella Eagles Feb 22, 1865
- Ab'm J. Kemp: Feb 12, 1831
- Peter Evans Kemp: Feb 28, 1837
Description and Analysis:
- Generational Structure: This record begins with Abraham Kemp and Sarah Fletcher's marriage in 1818, suggesting they are the earliest ancestors recorded here. This initial generation includes multiple offspring, with birth years spanning from 1795 (likely Abraham) to 1847. Their children include Ab’m J., Peter Evans, and Andrew, who, in turn, form new branches in the record.
- Family Names and Patterns: The names Peter, John, Sarah, and Mary recur, suggesting these were valued family names across generations. Some middle names (e.g., Fletcher, Evans, Ryder) appear to reference either maternal family lines or significant associations.
- Marriage Ages and Locations: Multiple marriages and births imply a migration history, possibly from Canada to the U.S. (specifically, California).
- Longevity and Mortality: Notable ages at death—several individuals reaching their 70s or 80s—may hint at health or lifestyle aspects for this family.
Extrapolation, Hypothesis, and Speculation:
This record may reflect a prominent family with roots in Canada and later migrations to the U.S., typical of 19th-century movements. Their detailed documentation, unusual for the era, could suggest social standing or strong traditions of record-keeping, possibly for legal, religious, or inheritance reasons. The presence of witnesses at marriages and specific attention to ages at death indicates importance placed on these life events, perhaps for religious or social lineage tracking.
The entry of "Orange County, California" as the final repository hints at the family’s eventual settlement in California, perhaps during westward expansion or after moving south from Canada.
Summary (100 Words):
The Kemp Family Bible Record documents a lineage from Abraham Kemp (b. 1795) to descendants in the mid-20th century, emphasizing Canadian roots and eventual U.S. settlement. Marriages and significant life events—tracked with precision—suggest a family valuing heritage, likely tied to migration and settlement patterns of the 19th century. Recurring names and detailed mortality ages reflect social and possibly religious customs, with longevity suggesting a stable, possibly well-off lineage. The meticulous recording hints at a prominent family, ultimately archiving its legacy with the Orange County Genealogical Society, a testament to its historical and genealogical value.
3) I noticed that it did not list every name, birth or death record on the page, so I prompted: "on the second "page" there are additional births of children of Abraham and Sarah, and also birth and death records of children of Peter Evans Kemp. Please include them all in the summary."
The ChatGPT4 response was:
Revised Plan
I’ll now incorporate all the births and deaths from the second page, particularly the additional children of Abraham Kemp and Sarah Fletcher, as well as the children of Peter Evans Kemp and Isabella Eagles. I’ll ensure the final summary includes a more complete picture of the family structure, covering these additional entries.
Execution
Additional Transcription from the Second Page:
Abraham Kemp and Sarah Fletcher's Children:
- Abraham Kemp: Nov 4, 1795 (noted as the likely birth date)
- Sarah Kemp: July 7, 1802 (approximate birth date)
- Waty C. Kemp: Sept 23, 1821
- Mary Ann Kemp: July 1824
- Stephen G. Kemp: Feb 1826
- W. H. Kemp: Mar 12, 1829
- Ab’m J. Kemp: May 22, 1831
- John L. Kemp: Apr 28, 1834 (approximate)
- Peter Evans Kemp: Feb 28, 1837
- Andrew Kemp: Nov 6, 1840
- Sarah J. Kemp: June 2, 1843
- Chas. W. Kemp: Sept 14, 1844
- Wesley Kemp: Nov 16, 1847
Peter Evans Kemp and Isabella Eagles' Children:
- P. E. Kemp: Feb 28, 1837
- Isabella Eagles: July 10, 1839
- Janet Isabella Kemp: July 22, 1866
- Andrew Evans Kemp: Sept 9, 1872
Death Records of Peter Evans Kemp's Children:
- Isabella Kemp: Mar 14, 1873
- Andrew Evans Kemp: Aug 29, 1873
Revised Summary (100 Words):
The Kemp Family Bible Record spans multiple generations, starting with Abraham Kemp (b. 1795) and Sarah Fletcher, who married in 1818. They had 12 children, including Peter Evans Kemp, who married Isabella Eagles in 1865 and had three children, two of whom died young in 1873. With roots in Canada, the family eventually migrated to the U.S., settling in California. Their meticulous record-keeping of births, deaths, and marriages highlights their dedication to lineage. This extensive documentation, ultimately archived with the Orange County Genealogical Society, reflects the family's valued heritage, resilience, and perhaps a respected social standing.
4) How did the ChatGPT4 AI tool do with the transcription of this typed Bible records list? My opinion is that the tool worked pretty well (with the additional prompt), but the user still needs to check that the tool captured and discussed all of the meaningful content. My opinion is that 100 words may not be sufficient for a closing summary.
5) Thank you to Steve Little for creating this custom AI ChatGPT tool. It works easily and well, and should be helpful for item identification and analysis for records or documents that are easily readable by the AI tool.
No comments:
Post a Comment