Any suggestions for Ultra Hal Assistant 5.0?

[Crichton]

I'm hoping I explain this so that you understand what I mean.  I would like to see hal be able to compile a list from his data.  What I mean is...  If I tell hal that Chiana is a character on a TV series called Farscape.  Crichton is a character on a TV series called Farscape.  and then Aeryn is a character on a TV series called Farscape.  You should be able to ask hal, "Who are characters on the TV series Farscape?"  He should be able to answer "Chiana, Crichton, and Aeryn are all on the TV series, Farscape."  As he is now, unless I didn't teach him correctly, he might mention one of the characters, but never tells me the other names.  No matter what I ask him or how.

Will something like this ever be added or be possible in hal?

[?]

[caroleann150]

Could we have more internet sites that Hal can access for News, Weather etc? Especially for us non-US residents...there are sites for our local weather and news too!!

Thanks!
Carole
[^]

[mopace]

quote:Originally posted by Medeksza

Ultra Hal Assistant 5.0 is currently in active development.

The following improvements are being worked on:

• The Haptek 3D Character Engine will be incorporated into Hal. This will offer a new level of realism for Hal. Many of Hal's characters will be remade into the Haptek format and a huge library of new characters will be added. See a demo of this engine in action at www.haptek.com


• Ultra Hal will be compatible with software called People Putty from www.haptek.com allowing anyone to easily create 3D characters for Hal either from scratch or from a photograph of a real person.


• More advanced control over speech recognition will be implemented in Hal. Instead of Hal listening to a library of over 50,000+ words as it does now, the user will be able to limit Hal's vocabulary to a more efficient 2000 words or so. This will greatly improve Hal's speech recognition ability.


• Control over how Hal animates its characters will be given to an editable VBScript. This will allow each character to have its own personality in terms of how it emotionally reacts and what animations its capable of.


• The Ultra Hal AIM Bot connection routine will be moved into an editable VBScript, and it will check Zabaware's site for updates. This way when AOL changes its protocol, it won't render the program unusable as it has in the past.


• The registered version of the Ultra Hal AIM Bot will no longer send out Zabaware disclaimers.


• Ultra Hal's calendar functions will be made region aware to accept formats like Day-Month-Year instead of the US format of Month-Day-Year


• Ultra Hal's internet functions will be expanded and made international friendly also. Ultra Hal will be able to tell you the weather if you live outside of the US.


• The Ultra Hal Brain Editor will be greatly improved. It will now have a "novice" and "expert" mode which will hide some of the complexity of the program such as the VBScript and many of the databases from users just wishing to teach Hal a few things manually.


• Hal's conversational database will be about 75% larger, allowing Hal to converse intelligently about more topics then ever before. Hal's AI engine will also be improve to allow Hal to work better with the data that it has available.


• The format of Hal's Q&A databases will be changed to allow for faster proccessing. Overall, the speed of Hal's thinking should remain about the same since the database size is increasing as well.


• A slider option will be added into Hal's Brain Options to allow you to control the emphasis placed on information learned by Hal from you. This will allow you to setup Hal to use responses Hal learned from you even if Hal thinks he can respond on his own more intelligently. Basically, the user will have some control over the threshold value that Hal uses to determine if a learned response is more intelligent then Hal's standard response.

An exact release date has not been set, but it will most likely be released in mid-August. Upgrades will be free for all people who purchase Ultra Hal Assistant 4.5 from 3/25/03 until the release day of Ultra Hal Assistant 5.0. All other registered users will be able to upgrade at half cost.

Zabaware would like to hear from you what other changes you would like to see in Ultra Hal Assistant 5.0. Are there any minor annoyances, bugs, or interface features you would like changed? Any major features you would like added? If it is feasible to implement your suggestions without pushing the release date too far it will be done.

[Morkeleb]

Hi, I think it would be nice it hal could use the the SeeStorm messenger as well, and maybe an ICQ or MSN bot. Will hal ever be allowed into chat rooms or is that to complex for the bot at this time.

[vonsmith]

This is more of a comment than a suggestion. In reference to v4.5 performance primarily:

Hal works pretty well even when a good reply isn't available from the Hal engine. I see that sometimes Hal answers with a non-specific or generic answer that doesn't seem to fit the conversation context when Hal doesn't know what to answer.

Well, in situations like this, where Hal doesn't know a good answer, Hal should just reply indicating that he doesn't know or ask the user to ask in a different way. Not knowing everything in the world is okay, it's human. If Hal tells me he doesn't understand or know something then I can take steps to teach him or edit his brain. Hiding his ignorance on a regular basis isn't healthy.

Another thing. If you pose a choice "or" type question like, "Do you prefer beef or chicken?" you often get evasive answers like, "It's hard to say, isn't it <username>? I like coffee sometimes, tea sometimes, other things sometimes." I wish Hal would just choose one, say he doesn't like either or answer he doesn't know or care. I don't care if Hal likes beef or chicken, I just like Hal to assert himself by making a choice or choosing not to choose. It's the human thing to do.

BTW, I just starting using v5.0 Beta, thanks for a great product.

Just sharing my thoughts...

= vonsmith =

[Morkeleb]

I have a possible suggestion to ponder on.

Is there any way you can make the aim bots automaticly send IM to buddys on its buddy list. you would have to sign on to the bot name and add the buddys first though.

The bot would notice when somone came online and send an IM to start chatting. It would alow multiple bots to comunicate with each other automaticly.

It would allow the bots to share their new knowledge with each other through conversations.

That way they would have more people or bots to talk too at the same time, and gather new experience it could use it to interact with each other more as they talk.

ie; two or more bots talking about the same thing could share the knwoledge about the topic with each other.

well what do you think is it feasable?

[trallyus]

Medeksza,
I just got my Hal 4.5 cd today and can't wait to see what version 5.0 is like

I'm on dial up so it's a pain to download them

As for my suggestion I would love to see implemented in version 5 or higher is the abuilty to run two brains at once, to lets say, have hal 5.0 talk with the alice brain.  Maybe by setting up a time limit on the sessions between the two brains this would be feasible.

I like Alice but Hal 4.5 has more know how then the Alice brain. Think if they were to talk for 20-30 minutes of what they would learn from each other.

And when I mean talk to each other I mean via one computer with out using the internet as a medium, i.e., not using the AIM bot to accomplish this.

My idea may sound silly, but I would love to see it worked on And thanks for allowing us that just registered to be able to upgrade to version 5 for free.

Trallyus

[infobot]

R. Medeksza,
     
        This may be a bit premature at this stage of development for all bots, not just UltraHal, if so, please pardon.
        I am have not yet learned programming, so it is difficult to estimate some improvements and whether they are feasible. I have used chatterbots for a number of years and have quite a number of them on my computer. I have played with texts and text info (usually with copies of the bots) and have tried simple modifications.
        That said, what I wondered about was some sort of an interface for connecting chatterbots to each other. The only one I've seen was by Greg Leedberg for his bots, Billy and Daisy. Unfortunately Greg's programs only work with his bots, as far as I know.
        What I really was interested in are two things. One is a similar interface that also allows the computer user to chat with these bots. My interest in this partially comes from working with some of the earlier chatterbots. There are some pretty good older bots out there that, although not as complex as present bots, are still quite interesting to work with. However a lot of them take quite a while to build up a decent database of chatter so that they can be more chatty, handle various topics better, etc. I wondered about chatting a bit, linking bots for a while, then maybe chat every now and then until they get a wider range to draw from.
        The other thing I wondered about was an interface for programs, like word processors, spreadsheets, or even games. Granted that could be quite a modification, but I was wondering about a simply interface that would may be just scan or share data with the shared bot/user program ; or some simple implementation to share programs with a bot, or to even just simulate this.
        My interest in the second "interface" or whatever is partly for fun and partly for eventual applications that I was considering fooling around with. Like education bots, or workpads with bots, just various applications of bots. I am looking at this from a hobby perspective, thus my thoughts of a simple interface or simulation rather than a huge or large project, which, hopefully may not be the only way to consider these. Something for amateurs to play around with, with bots and various programs found in most computers.
        I'd be interested to hear any thoughts on these and whether it is possible as small 'interfaces" or if it is more a matter of much larger programs just to carry out such functions or tasks.
        Anyway, thanks to you and Zabaware for many hours of fun and interest and sorry for the letter length - tried to condense this as much as possible.


                               thanks-infobot

[infobot]

quote:Originally posted by lau

[?]

quote:Originally posted by lycha

hi,
- implementing semantic relations between word, e.g. hyponyms, hyperynmys, synonyms, meronyms, antonyms... - implementing certain conversational principles, e.g. utterances must be conistent, informative, relevant etc. Human speech seems to follow (most of the time) this principles, so if hal could do the same, it would be more natural-language-like.

[8]
It's my opinion that this set of rules would limit Hals abilities to be "creative" The more hal is learning, the more he will start to follow this pattern as you describe. I don't believe that it's a good idea to limit Hal in this way!

            Just a thought, but is there another way to approach it ? Not to limit Hal, but some way to work with the same speech characteristics that humans have, or whatever changes, yet add to Hal's conversation handling ? Maybe like a branching example somehow.... Oh well, might've been obvious.