I have been watching artificial intelligence develop since I read I Robot for the first time back when I was probably about ten years old. Of course, I didn't really know anything about AI until I was well into my studies of linguistics during my years at the University of Utah. Later, as I became totally involved with computers, I discovered programs that were supposed to try to pass the Touring test for machine intelligence. Then genealogy came along in about 1982 and I realized that AI could help with many of the aspects of genealogy, one major area being optical character recognition and speech transcription. I had seen a demonstration of speech transcription at the IBM exhibit at the World's Fair in Seattle in 1962. See “IBM100 - Pioneering Speech Recognition.” 2012. CTB14. IBM Corporation. March 7, 2012. http://www-03.ibm.com/ibm/history/ibm100/us/en/icons/speechreco/breakthroughs/. I have written quite a few blogs about using speech recognition mostly when I was using the Apple version of Dragon Naturally Speaking by Nuance Software which was the commercial implementation of IBM's program. Unfortunately, Nuance discontinued development and support for the Mac version, and I lost interest in using speech recognition because none of the Mac compatible programs seemed to work well enough to compete with my typing. The editing capability of the Google and Microsoft dictation programs is rudimentary.
As the years passed, I became involved with learning about handwriting recognition. Optical character recognition has reached an extremely high level of competence and for documents that are in type, accuracy is sufficient for all levels of genealogical research with the one qualification that the original document be available. The current challenge of OCR is not recognizing the characters, it is formatting the output. This turns out to be the larger challenge of handwriting transcription also. It is one thing to transcribe a document written by hand such as a letter and it is another challenge to transcribe a parish register with columns and keep the written document such as this Swedish Household Examination document.
The arrangement of the information is just as important as being able to read the handwritten text.
Now, what has happened in the past few years is a constant improvement in handwriting recognition and document transcription that now allows billions of records to be transcribed in a relatively short time. We saw this in 2022 with the release of the 1950 U.S. Census and the rapid transcription and index of the census by Ancestry.com with the support of FamilySearch.org and a lot of volunteers. 2023 will see the implementation of this technology to transcribe a flood of new documents and indexes to unlock many more.
Will the human element be replaced by computers? That is still an unanswered question. But the recent news about ChatGPT is an example of what will undoubtedly affect almost every aspect of the research process for genealogy and a lot of other academic disciplines. Here is an AI attempt at the subject from ChatGPT.openai.com
Artificial intelligence (AI) is already starting to impact genealogical research in a number of ways, and it is likely to continue to do so in the future. Some of the ways in which AI could potentially affect genealogical research include:
Handwriting recognition: AI-powered handwriting recognition software can help to transcribe and digitize handwritten records, making them more easily searchable and accessible to researchers. This can be especially useful for older documents that may be difficult to read due to faded or illegible handwriting.
Image analysis: AI algorithms can be used to analyze images and extract information from them. For example, AI could potentially be used to identify people or places in photographs or to transcribe information from census records or other documents.
Data analysis: AI algorithms can be used to analyze large datasets, such as census records or military records, and extract useful information for genealogical research. For example, AI could potentially be used to identify patterns and connections between different individuals or families.
Predictive modeling: AI algorithms can be used to make predictions based on historical data. For example, AI could potentially be used to predict an individual's likely occupation or place of residence based on their age, education, and other factors.
It's important to note that AI is still a rapidly developing field, and its impact on genealogical research will depend on a number of factors. Some limitations of AI in genealogical research include the need for high-quality data and the potential for biased or inaccurate results if the algorithms are not properly designed and trained.
I am not worried about being replaced by this or any other program but I appreciate the advances that make doing research possible. You are going to see a lot more AI involved in genealogy in 2023 and beyond and how we do genealogical research will continue to change just as rapidly.
This is good and bad. I've gotten more false positives, but it also has uncovered some that were misspelled. Double edged swordReplyDelete