• Only registered members can see all the forums - if you've received an invitation to join (it'll be on your My Summary page) please register NOW!

  • If you're looking for the LostCousins site please click the logo in the top left corner - these forums are for existing LostCousins members only.
  • This is the LostCousins Forum. If you were looking for the LostCousins website simply click the logo at the top left.
  • It's easier than ever before to check your entries from the 1881 Census - more details here

Transcribing old writing

Victoria11

New Member
In the latest newsletter there is an article about using ChatGPT to transcribe an old will.

I have experimented with ChatGPT, but have also used Transkribus which is much better for this purpose. I have mainly been working on old house research and it is very good at old hand written legal stuff.

Transkribus is a non profit multi national organisation set up by various universities and other research organisations. It is free to use for simple transcriptions.

They also run free webinars on how to use it from time to time.

It is not perfect but it does not invent things like ChatGPT and if you choose the right 'template' for your old writing it can do a near perfect job a lot of the time.
 
In the latest newsletter there is an article about using ChatGPT to transcribe an old will.

This is incorrect. The article speaks about using MyHeritage's Scribe AI and NOT ChatGPT. Sadly whenever people speak about AI they always mention the least capbable of the AIs out there, namely ChatGPT, as typically that's all people have heard of. However there are hundreds of different AI models available almost all are way better than ChatGPT. The simplest to use that most people will have used without realising it is Gemini from Google. Most of the time these days when you do a Google Search the answer you get is from Gemini.

You can specifically try out Gemini for free by visiting this link the key to getting good results is treating it like you were having a conversation with someone and explain exactly what you want in detail. If you just give it a few key words it's no better than a quick search. If you explain things you get some really good results, and unlike ChatGPT it doesn't make things up on almost every response.

To prove that you don't need to pay for Scribe AI from My Heritage I asked Gemini to transcribe the image from the newsletter and I just pasted the image into it's chat with a simple prompt - can you transcribe this document- it took about 3 seconds to come back with
Here is a full transcription of the 1769 last will and testament of Morris Britton from image_b0862d.jpg.
Transcription of image_b0862d.jpg
In the Name of God Amen I Morris Britton of the Hamlet of Oldland in the Parish of Bitton in the County of Gloucester Coalminer Considering with my self the Certainty of Death and the uncertainty of the Time thereof do make this my last Will and Testament in Manner following

Imprimis I Give devise and bequeath unto my Dearly beloved Wife Elizabeth Britton all my Mes = = suages Cottages or Tenements Outhouses Edifices Buildings Gardens Orchards Inclosures of Ground Together with the Appurtenances thereunto Adjoyning and belonging Situate lying and being in the Hamlet of Oldland in the Parish of Bitton and County of Gloucester Aforesaid Together with all my Goods and Chattels both within Doors and without AND also all my Wearing Apparel both Linen and Woollen Presently after my Decease during her Natural Life Item I Give and bequeath unto my Loving Son Thomas Britton One Shilling to be Paid him by my Executor in a Twelve Month Time after my Decease Item I Give and bequeath unto Thomas Britton Son of my Said Son Thomas Britton Ten Pounds of Good and Lawfull Money of Great Britain to be Paid him by my Executor in a Twelve Months Time after my Wifes Decease out of my said Cottages or Tenements Gardens Orchards and Inclosures of Ground and Premisses thereunto belonging or in any wise Appertaining but if the said Thomas Britton shall Happen to die before the Time Limitted for Payment of the said Ten Pou = = nds before Mentioned then the said Ten Pounds shall fall to my said Children herein after Named to be equally divided between them share and share alike Item I Give and bequeath unto Hannah Hiscocks Daughter of Richard Hiscocks of the Hamlet of Oldland in the Parish of Bitton in the County of Gloucester Coalminer five Pounds of Lawfull Money of Great Britton to be Paid to her by my Executor in a Twelve Months Time after my Wifes Decease to be Paid out of my Cottages or Tenements Gardens Orchards Inclosures of Ground and Premisses thereunto belong = = ing or in any wise Appertaining but if the said Hannah Hiscocks shall happen to die before the Time be Expired for Payment thereof then the said five Pounds before Mentioned shall fall between my said Children herein after Named to be equally Divided between them share and share alike Item I Give and bequeath unto my Loving son Itzkel Britton my Cottage or Ten = = ement Garden Orchard and Inclosure of Ground and Premisses thereunto belonging or in any wise Appertaining Presently after my Wifes Decease [interlined: now in the occupation of Isaac Green] and after my said Son Itzkel Decease the said Cottage or Tenement Garden Orchard and Inclosure of Ground shall Immediately fall to his son Morris Britton Item I Give and bequeath unto my Loving son Sampson Britton the Cottage or Tenement wherein he now Dwells together with the Garden and Inclosure of Ground there = = unto Adjoyning and belonging with the Appurtenances thereunto belonging or in any wise Appert = = aining Presently after my Wifes Decease and Immediately after my son Sampson Britton Decea = = se the said Cottage or Tenement Garden and Inclosure of Ground and Premisses shall fall to his Son Samuel Britton Item I Give and bequeath unto my Son Samuel Britton my Cottage or Tenement Garden and Inclosure of Ground thereunto belonging or in any wise Appertaining Immediately after my Wifes Decease all which said Premisses are Situated lying and being in the hamlet of Oldland Aforesaid Item if my said Son Samuel Britton shall happen to die without Issue then the said Cottage or Tenement Garden and Inclosure and Premisses shall fall between the rest of my Children to be Divided equally between them share and share alike Item my Sons Itzkel Britton Samuel Britton and Sampson Britton shall Pay unto my Daug = = hter Mary Cook Wife of Francis Cook equivalent according as their Places is Valued by Two Prope = = r Men to Pay to equal share and Share alike Presently after my Wifes Decease Item I Do make my Loving Son Itzkel Britton my whole and Sole Executor of this my last Will and Testament In Witness whereof I have to this my last Will and Testament Set my Hand and Seal this 7 Day of May 1769

Signed Sealed Published and declared by the Testator as his last Will and Testament in the Presence of us who at his request and in his Presence Subscribed our Names as Witnesses

  • Jos. Owen

  • John Radford

  • The Mark of William Pear [X]
The Mark of Morris Britton [X]
 
Once you have the transcription it is easy to follow up by asking it - please extract the key information from this transcripton highlighting individuals and family relationships. Which resulted in
Here is the key information extracted from the transcription of the 1769 will of Morris Britton (image_b0862d.jpg), focusing on the individuals mentioned and their family relationships.

The Testator
  • Morris Britton: The individual making the will; a coalminer residing in the Hamlet of Oldland, Parish of Bitton, Gloucestershire.
Family Relationships & Beneficiaries
Immediate Family
  • Elizabeth Britton: Wife of Morris. She is granted a life interest in all real estate (messuages, cottages, tenements, land) and all goods, chattels, and wearing apparel.

  • Itzkel Britton: Son. Named as a primary beneficiary of real estate (a cottage/tenement and land after his mother's death) and appointed as the whole and sole Executor of the will.

  • Sampson Britton: Son. Bequeathed the cottage and land where he was already dwelling, to take effect after his mother's death.

  • Samuel Britton: Son. Bequeathed a cottage and land after his mother's death, with a stipulation that if he dies without issue, the property is divided among the surviving siblings.

  • Thomas Britton: Son. Bequeathed a nominal token of one shilling, to be paid twelve months after Morris's death.

  • Mary Cook (née Britton): Daughter. Wife of Francis Cook (son-in-law). She is to receive an equivalent monetary share paid by her brothers (Itzkel, Samuel, and Sampson) based on a two-man valuation of their inherited properties after their mother's death.
Grandchildren
  • Thomas Britton: Grandson, identified explicitly as the son of Morris's son Thomas. Bequeathed £10, to be paid twelve months after the widow Elizabeth's death.

  • Morris Britton: Grandson, identified as the son of Itzkel Britton. He is the reversionary heir to the cottage and land left to Itzkel.

  • Samuel Britton: Grandson, identified as the son of Sampson Britton. He is the reversionary heir to the cottage and land left to Sampson.
Other Individuals Mentioned
  • Hannah Hiscocks: Beneficiary of a £5 legacy (to be paid twelve months after Elizabeth's death). She is identified as the daughter of Richard Hiscocks, who, like the testator, is a coalminer in the Hamlet of Oldland.

  • Isaac Green: Mentioned as the current tenant/occupier of the cottage bequeathed to Itzkel Britton.

  • Jos. Owen: Witness to the signing of the will.

  • John Radford: Witness to the signing of the will.

  • William Pear: Witness to the signing of the will (signed with his mark).
  • another couple of seconds later, and all entirely for free.
 
This is incorrect. The article speaks about using MyHeritage's Scribe AI and NOT ChatGPT
Alexander, the original post was written on 2nd June about an article in the June newsletter which was about ChatGPT.
Thanks for explaining how to fully use Gemini.
I agree that ChatGPT makes things up, even it's references.
It's great to see you back on the forum.
 
Ah my confusion was that the July newsletter also had AI analysis. It's a bit of a annoyance with IT professionals that everyone goes on about ChatGPT when it's probably the least capable LLM AI tools out there, challenged only by Microsoft CoPilot in how bad it is. Note ALL LLM AI models will potentially summarise and leave things out, the degree to which they do is very strongly influenced by how much you prompt it to be diligent.

Perhaps a quick getting started guide:

Visit the Gemini page and login (this requires a free google account which many people will already have especially if they use Chrome and are signed into Chrome so their bookmarks are shared across devices.

Once logged in start by giving Gemini some context of how you want to interact with it. eg:

"Please remember this context for all future conversations. I am retired with a keen interest in family history and in X, Y and Z. I live in Anytown, My County, UK so have a particular interest in local references and I prefer clear less technical answers with links to relevant articles so I can read up on the background in my own. Please take time to check your sources of all information ideally by finding at least two sources for anything you report back to me. Please try to avoid "reading between the lines" and guessing, if you are unsure or have been unable to find reliable sources, then tell me you are unsure. I'd rather be told you cannot find proof than be given inaccurate information. The tone of our converstaions should be coffee table chat, rather than corporate style, or too formal."

Then start a conversation.

Now hopefully that initial context gives you a flavour of how you can "set the scene" for all future interactions. Obviously you can tweak that to your own preferences, eg: you may prefer a more cold solid fact based discussion without the "coffee table chat" vibe, or you prefer some other style. A friend of mine has his talk in a local dialect, if you instruct it to remmeber to do something it will adapt the style of reply to suit. Note it is defintitely NOT too late to give it context if you are already using Gemini. The key is asking it to remember something for later (which is why you need to be logged in). The more you set the scene and carve out a mode of interaction that doesnt annoy you the better the results you will get. Plus if the answers you get don't sit right, then tell it. "Actually I'd prefer you to be less chatty and more professional."

The key thing is because you are logged in it will remember these directions and use them for all future chats.

So for instance for a first "google search" conversation instead of the old fashioned key word search you have done in the past "Lost Cousins history". Try the conversational "Tell me about the history of Lost Cousins and the key features of what makes it stand out as a unique family history website." It will come back with answers read them and then question it on what it replied, ask it for links for further reading or follow the links already provided.

The key is ENGAGE IN CONVERSTATION not just old fashioned keyword searching. You'll likely be shocked at just how useful, AND accurate this new method of searching is.
 
I've not tried to use any sort of AI to help with my family history endeavours as yet, but this is really encouraging. I do have a few documents that I've set aside to look at 'later' (aka perhaps never) because they are not easy to read and not central to my current research. I'd certainly consider giving this a go at deciphering them because it sounds much easier than spending countless hours puzzling over the text!
 
I'd certainly consider giving this a go at deciphering them because it sounds much easier than spending countless hours puzzling over the text!

The key is it can give you what it thinks then you can use that to compare with the text and see if you think it matches, which is easier to deal with than trying to work it out from scratch yourself. It might have got it wrong, and you can mark bits as questionable but it's a starter for 10 as they say.
 
The key is it can give you what it thinks then you can use that to compare with the text and see if you think it matches, which is easier to deal with than trying to work it out from scratch yourself. It might have got it wrong, and you can mark bits as questionable but it's a starter for 10 as they say.
I just clicked 'agree' before because I was in a rush, but yes that's exactly what I had in mind. I wouldn't expect AI to be as good as a human expert, especially if it was free. However it would be very useful if it could get me started and plough through a document, most of which will likely be non-controversial, leaving me to check the results and only needing to investigate anything that was important to me and that I was not sure was correct.
 
Back
Top