| @@ -9 +9,4 @@ |
As a personal assistant I don't think any of these count. I can understand part of the 'privacy' issues which storing a history of interaction creates, and the problems that earlier 'models' had in upgrading their responses based on other discussions, but if I am using an 'Assistant', they should be able to remember facts like my medical problems and tailor responses appropriately. I think that a MAJOR failing of the NHS 111 service is that it starts from zero on every consultation. I should be able to access it via the security of the app and it remember what we had discussed previously. THAT would be the intelligent thing for it to do, and it's training set should be UK centric and leave out any specifically American bias. With all of this managed via data centres sited in the United Kingdom.
|
As a personal assistant I don't think any of these count. I can understand part of the 'privacy' issues which storing a history of interaction creates, and the problems that earlier 'models' had in upgrading their responses based on other discussions, but if I am using an 'Assistant', they should be able to remember facts like my medical problems and tailor responses appropriately. I think that a MAJOR failing of the NHS 111 service is that it starts from zero on every consultation. I should be able to access it via the security of the app and it remember what we had discussed previously. THAT would be the intelligent thing for it to do, and it's training set should be UK centric and leave out any specifically American bias. With all of this managed via data centres sited in the United Kingdom. |
|
Update - 12/3/2025
|
| I have been having mixed results using Mistral since first writing these notes. Certainly code related enquiries provide useful prompts even if they are a little out of sync with my own Linux setup. I have to keep providing a context and now have that as a crib ro add to a new chat thread. However it is also quite obvious that some areas are not covered well by the training set. In particular asking for help with some niggles in Thunderbird produced what look like good answers, in reality many of the directions were simply wrong, and trying to correct the mistakes just produced even futher miss-information. Just what areas are and are not well supported is not something that the chat bot can answer, but perhaps a better handling of situations where there are holes in the training set needs to be addressed. |
|
|
