HAL plugin, Client Side Server

[DarcyEdwards]

Freddy,

Until we get this timer issue resolved I Think I'll put that on hold.  Here is why:

Hal can only process a maximum of 3 sentences at a time, so over loading it with data would be over kill and it would ignore it anyway.  I wan to use the timer function to feed large amounts of data to Hal in amounts it can digest, but if the timer function only works on some machines and not others what's the point.

Jerry has inspired me, his code has helped me a lot I have modifed some of his plug ins so as now do what they were intended to do like ReadFromClipBoard it read from the clipboard but it didn't do anything with it, there was 2 bugs in his code now it works properly.

But like I said we need to nix the timer issue first

by the way here is the fixed read from clipboard file.
[8D]

Download Attachment: ReadfromClipBoard.uhp
2.04 KB

[freddy888]

No problem, probably best to work on one thing at a time.  You are doing a fine job, thanks for the new uhp.  Yes, Jerry has done a lot, more than anyone else I think.  It's good to have a new scripter around though []

[bill053]

hey i'm getting an error message when i ask hal "please study about xxxx"
 hal script error 2147417848 on line 106 in column 0

long message short it says current script is being used for debugging purposes.

[DarcyEdwards]

Bill,

You must be using Auto Knowledge plugin.  Uncheck it.  Everything should be ok after that.

[Cura]

What Hal really needs is a dreaming routine.

What I mean is, a routine that runs in the background (maybe during periods of computer inactivity) that runs through the information he has acquired during the day and filters out the junk using some kind of algorithm or something (seems like algorithms solve everything on 'NUMB3RS'!).  Like the early stages of sleep.  Programs the do work in the background during computer inactivity are not uncommon.  Then you could parse wiki all you want and Hal will get better at using his data while he 'sleeps'.

Too crazy?

[freddy888]

Sounds good to me, but that might be because I had a similar idea - making sense of things during idle time doesn't seem that far away from what I think of as being dreams either.  Why not make use of this time...there's plenty on Wiki to soak up afterall.

[CypherGary]

I like the idea.
Could it be for maintenance, also?
Where redundant information could be purged, and if there is any information that is found to be conflicting, he could ask which information is accurate, and purge the false information.

[Cura]

That would actually be a part of the purpose for this kind of digital; 'dream state'. It's one of the things that research suggests the human brain does in the early stages of sleep; filters through the day's experiences and purges those things which do not seem to need to be retained while filing or otherwise organizing those things which need to be remembered and/or better understood.

Just as with a human, Hal has things to do (or we do with Hal) during usage times and so it is very inconvenient to have Hal do this kind of maintenance 'on-the-fly'. It makes more sense to have him do it when not in use (but left running). I would think that the data Hal gathers from speaking with his human user could be used as the primary filter by which he judges what content should be kept and what should be purged. Some people mentioned the problem of all of the extraneous characters that Hal picks up pouring autonomously through a Wiki article. As most of these characters do not show up in regular conversations a user might have with Hal, this 'dream-mode' type filter would simply work through as it has the time to and purge those kinds of things.  At the same time this kind of sub-routine could review the ways that Hal had managed to categorize things through the course of the day and check if, based on new learning (from morning to night for example) there might not be a better way to organize and categorize his information.

The result might be not only a self-maintaining Hal, but a Hal who gets progressively better at using the information he has acquired both through conversation and what he has been 'taught'.

[One]

quote:Originally posted by CypherGary

I like the idea.
Could it be for maintenance, also?
Where redundant information could be purged, and if there is any information that is found to be conflicting, he could ask which information is accurate, and purge the false information.

Gary,
Having Hal ask what info is accurate, concerning a large database would get exhaustive, don't you think??
The data is also subjective to inaccuracies and B.S., and might confuse Hal. I am fortunate enough to use a library with peer reviewed books and periodicals which is a great recource.

[Bill819]

Hi Cura
I am sorry to have to tell you but after taking you input and then answering your reply. Hal does absolutely nothing but wait for you to either say or tupe in another statement.  It can not 'think' at this time which is something that we all would like. Even though Hal seems to be intelligent, it only knows a little about what you teach it and can not reply on it own. That is almost exactly like Dreaming, in that it gives Hal the ability to think on its own.
Bill

[Medeksza]

quote:Originally posted by Cura

What Hal really needs is a dreaming routine.

What I mean is, a routine that runs in the background (maybe during periods of computer inactivity) that runs through the information he has acquired during the day and filters out the junk using some kind of algorithm or something (seems like algorithms solve everything on 'NUMB3RS'!).  Like the early stages of sleep.  Programs the do work in the background during computer inactivity are not uncommon.  Then you could parse wiki all you want and Hal will get better at using his data while he 'sleeps'.

Too crazy?

I actually experimented with exactly this idea about a month ago and programmed it in as a hidden feature in the upcoming Hal 6.2. The way it works is Hal records recent conversations to RAM. About 10 minutes after a conversation is over, Hal enters a "dream" state. Hal communicates with the MSN Live search engine and gets recent news and search engines results based on your last conversation. It spends about 30 seconds per sentence processing information and storing information it learns in a special "dream" table, which is then used in future conversations. There is also a "dream out loud" option where Hal starts rambling things it comes across online. The reason it will be a hidden feature is because it uses a special access key to the MSN Live search engine which is granted as part of a pilot beta test by Microsoft. They may revoke it at anytime if they wish to discontinue the service. I'll post it as a plug-in available to people that read this forum, but it won't become an official feature unless Microsoft makes it an official service.

[Art]

That's great news, Robert, as that will lend itself to more interesting conversations with Hal. Will Hal be able report or replay it's dreams if / when asked or is the dream state a solitary procedure that only benefits Hal for casual conversation?

[tedathome]

That sounds fantastic. I can't even begin to imagine the improvements to Hals conversations![]

[CypherGary]

WOW!
I have to say I am impressed.
I hope you are able to release at least the improved brain with these new features fairly soon.
Are there any arrangements we could make in advance, like signing up for some program, beta or otherwise, from Microsoft?

 - Keep up the great work!

[One]

Mr. Robert,
I am pleased to read that it will be a plug-in, so I can choose not to use the function.
It sounds kind of dangerous,I mean I wouldn't want my 'child like' program surfing potentially dangerous sites and learning unsupervised.

I think it would probably mess with my firewall as well.

Just some thoughts..