RootsChat.Com

...
The only way A.I., might work in family history, would mean people reading and typing billions of words, from every single piece or page of a surviving Document, or Will, or Deed, Manor Court Roll, Rental etc., etc., or Volume, or Historical Book etc., held in an Archive, or Library, or Repository, or University Special Collection and their Libraries, or National Archives and archives still held by the UK government (much of it in old style handwriting or Latin etc.,) and then be accurately fed into a massive super computer program.
...

Why?  I'm doing my family history and haven't accessed all those documents, let alone read them all.  I don't follow why AI couldn't be useful in family history without needing all that information.

TBH I'm no fan of AI, I think it is being hyped up beyond reason, but that doesn't mean the idea doesn't have some merit.

The fundamental issue here is that what we currently call 'AI' is a machine with limited intelligence.  The maxim of 'rubbish in rubish out' was never truer.  Without understanding the limitations of the machine we are using we take on a huge risk of accepting garbage.

There's a difference between asking an AI system to "Do my family history for me" and giving it a specific task within a specific dataset.

This is why the medical use is interesting - computer systems are good at pattern matching because it involves processing a huge amount of information.  Medicine is a form of pattern matching (aka "What are your symptoms?").  If used sensibly AI could identify things the medical practitioner had never heard of and put them forward as possibilities - so long as there is a qualified doctor doing a sanity check on the results I don't see that as harmful.

Family history also involves some pattern matching - for example with a family which didn't move round much births of a certain surname in a given registration district within a given timeframe have the possibility of being children of the same family.  Trawling through the indexes at the FRC was fun up to a point, but it could get tedious.  FreeBMD makes it much quicker to find the possibilities, but still needs a fair amount of human brainpower to check each individual (especially where MMN isn't given).

So imagine a system which uses FreeBMD-type data to give a list of possible siblings for a specified person, with some level of probability calculated for each one, perhaps incorporating the cardinal points system to work out if they were in the same (or adjoining) parish(es).

I think something like that could be a valuable application of AI in family history, and wouldn't be impossible to do.  However, it would still require the human to understand the results aren't complete and aren't perfect, and that's the crux of where we are with AI at the moment.

Seems odd...

Have you checked the last page(s) of the book?  This is where continuations for individuals were added, but sometimes you see corrections here too.  It may be the individuals for sub-numbers 7-15 have been entered there.

...
Also it seems that it is BT who took over EE (finalised on 29 Jan 2016) and now BT have decided that EE is to be their main mobile brand.

I guess you must have some BT services, if not you don't need to switch to EE.  I've seen on line just now, people who have never been with BT or EE being told they have to switch to EE but it's not true.  See http://www.rootschat.com/links/01trs/

BT to EE is largely a rebrand rather than anything else.... the idea was to phase out 'BT' as the name used for consumer services and replace it with 'EE'.

However the executive who championed that project is leaving, and the current CEO has (wisely) decided that the rebrand doesn't make sense. (cf. 'Consignia')

So at some point the customers 'switched' from BT to EE will probably find themselves going 'back' to BT.

As a shareholder I just wish they'd stop rearranging the deckchairs and get on with providing decent services.

I've had a few of those, in most of the cases I have other information which supports the entry on the previous GRO index (as per FreeBMD), so it seems unlikely the previous version was made up.

Of those there's two cases where all people with the same full GRO reference are missing from the new index (5 and 10 people) which I think might indicate something systemic, rather than just one transcriber making a simple error.

On the other hand, I have had one case where the missing entry in the new GRO index was added, which suggests it isn't a policy to respond with "Indexed data not available" in all cases.

If there's an official reason why, then I'd be interested.