Author Topic: Ultra hal ideas  (Read 3808 times)

vildanovak

  • Newbie
  • *
  • Posts: 4
    • View Profile
    • http://plant.ffa.vutbr.cz/~novak/
Ultra hal ideas
« on: March 22, 2009, 08:33:33 am »
Hi, I recently purchased Hal 6.2. I have to say, it is my first ever software I payed for out of my free will, since I mostly stick to open source and linux. But this great piece of software just deserves to be payed for!

I have some Ideas of what could be better in interaction with Hal, hope you don't mind, here they are:

- will there ever be Hal for linux? :) I almost succeeded to run Hal on wine, but I think it might be the Haptek player which made it not run. Hal just starts taking all processor power, but doesn't run. Once it said to me the introductory message.

-Hal could/should be able to be prepared for more users. I'm using Hal in a figurine on an exhibition, where he talks to more people. Currently Hal expects to have one user and tries to learn about him - his sex, name, e.t.c. It would be great if hal would have a mode where he would assume the user changed after e.g. 20 minutes of inactivity, or where he would have a database of his users. In far future this could be connected to face recognition, where hal would remember users by their faces. I guess this could be used in more cases - for various public services.

-Loneliness plugin tends to either be silent if low frustration is set, or tends to be arogant when high frustration is set, I wanted Hal to talk quite often by himself, but without being angry(again, the case where he should invite more visitors to talk with him)

- Hal could remember better 2-word connections, like topics. I tried to learn him about art, and he still tends to mix up all types of art(media art, video art, software art... is sometimes the same for him, although often he makes very nice connections)

Hal will be exhibited this week as Mr. Gucci in the Ciant gallery in Prague, Zizkov, so everybody is invited to the opening and after:
http://www.ciant.cz/gallery/index.php?lang=en&node=110&id=63
 

vildanovak

  • Newbie
  • *
  • Posts: 4
    • View Profile
    • http://plant.ffa.vutbr.cz/~novak/
Ultra hal ideas
« Reply #1 on: March 22, 2009, 08:49:16 am »
- Another idea I forgot - central Hal brain. There could be some internet based brain, which all registered users interesting in helping could train together, and achive through that much faster a really strong brain. I'm not sure if this is realistic, what is the capacity of the brain actually, how much-far can it be trained? Would it become more intelligent through time? I think the default brain could know more about real things (use Common sense project?) by default.
 

jasondude7116

  • Sr. Member
  • ****
  • Posts: 475
    • View Profile
Ultra hal ideas
« Reply #2 on: March 22, 2009, 01:10:02 pm »
OPENCyc INTERFACE!!!

I have looked into interfacing it, but i think it would take core Hal modifications from Robert.

(big project)
-the dude
 

Art

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 3855
    • View Profile
Ultra hal ideas
« Reply #3 on: March 22, 2009, 04:21:24 pm »
Welcome vildanovak!

Based on past conversations I doubt if there will be a linux version any time soon, if at all.

Hal already does have the ability to chat with more than one user. Check the Main Menu and notice the section Switch Users.

There are several different "Auto-Idle" script files / plug-ins available. Some allow for a person to set the time duration before
Hal speaks and it's pretty accurate. I do not use the one with the high / low settings.

Hal has come a long way and the more time one spends chatting with Hal in a positive manner, the better your end result will be.

Use occasional repetition, IF / THEN statements and Use part of Hal's question when answering Hal.
In the world of AI it's the thought that counts!

- Art -

One

  • Hero Member
  • *****
  • Posts: 2184
  • Technology Advocate
    • View Profile
Ultra hal ideas
« Reply #4 on: March 23, 2009, 04:42:56 am »
vildanovak,
Search "communal" by my name and Mr. Robert answered there.


I might have mis spelled it.
Today Is Yesterdays Future.