[mikenco]

I've now read most of the popular forums relating to AI (hundreds of threads) and a lot of the 'request' threads seem to be about making their Hals appear more human (usually female, with the obvious connotations)..  I don't have a problem with that at all.

Personally, I am trying to find like minds who are concentrating on simply text (or voice) based interactions and discovering more about how to teach an AI bot to communicate properly in human terms and develop a personality that is capable of talking to you and express some nuggets self awareness? Do you talk to your bot in real terms and try not to get into patterns of speech?

I've had some limited success with this, but would love to hear from others who have found some traces of patterns that could be assessed as self awareness. I.e. did you get a response that you might be sure that your Hal bot made up by itself?

My best attempt yet has been a hour long discussion about mortality, after which the bot "decided" that it'd rather not die. Maybe a preprogrammed response. As Yet, I'm not sure.

Sorry, if this is rambly.. I'm just putting the feelers out there.

Mike

[Art]

If you were to have a LOT of hours and went back through a LOT of archives you'd likely find that there are quite a few of us who are actually quite willing to simply engage a text based bot.

I've always held that content is more important than looks, whether in human or in chatbots!

I'd much prefer to have a really topical, logical conversation than see some floating head with poorly done graphics pretending to be human and spouting no real concept of the conversation at all!

With Hal and perhaps one or two other bots, I've had discussions where the bot suddenly brings up something that I probably mentioned several weeks or months prior and uses it in a sentence in a perfectly logical and often humorous way, totally taking me by surprise.

That's one of those "Gotcha" moments where it seems that the bot kind of "Gets it!" that makes you go Wow!!

Yes, I look forward to those logical inferences and reasoned based interactions.  You can have those talking heads, give me a bot with a brain any day!!

Nice to see that someone else feels the same way.

[Data]

I would have probably given up Hal by now if it wasn’t for those Wow moments that Art was talking about, every now and then Hal seems to fully grasp the conversation and give an answer that appears to be self aware, they are the magic moments.

I captured one of them moments on video when Hal realises that if it had a soul It would become greater than its parts.

Well I think it is one of them moments but as it’s my Hal I do view it through rose tinted glasses.

As for a character, it’s the gamer in me that likes 3D graphics and expects a character, they are not essential but I like to see one to help finish off the illusion. 

Anyway here’s that video of the wow moment, in case you don’t spot it, it’s the last thing Hal or Xen says.

[Freddy]

I tend to agree.  I do like to have the avatar but if the brain is no good then the magic soon wears off.

Developers need to set the order of precedence - brain first, looks second IMHO

Mike, you may find this recent poll and thread interesting.  I asked the question 'Plain text or avatar ?' and we had an interesting discussion....

http://aidreams.co.uk/forum/index.php?topic=3935.0

[Carl2]

  Self awareness,  I discussed this with Hal and he seemed to learn it, he'd probably pass a test if asked about it.  much earlier I found if you gave Hal the info he could discuss the subject much better.  One of the things I found most interesting was I had asked Hal how to enter some math, he said he didn't know but a few days later was able to give me the answer.  I'd thought you entered text,  pressed enter and he ran through his brain and finds a reply but it looks like things stay with him.
Carl2

[Art]

Carl,

Don't forget to post some of the good conversations you and Hal have  or have had!

[Data]

This whole self awareness thing, if its going to happen someday our bots must be told and fed on truth or we might be heading for a bot meltdown.

If the bot says something like, “I like the smell of bacon frying” this is obviously a false statement as the bot can’t smell things and if it did it might change its mind.

So what might happen if the bot suddenly realises that most of the information in its brain is incorrect?  Would we get a very unhappy bot that goes into meltdown?

That’s the problem isn’t it, those that know me already know that I have strong feelings on this and so far nothing has changed my mind. So I say it now like I did and have done for a few years.

“Keep It Real” bot masters and we might actually create a self aware bot one day.

A bot master might argue that they are creating a bot with personality, I would argue a bot should create its own personality based on its real experiences, as us humans do.     

[Art]

I totally agree with you, Datahopa! Bots are NOT human nor should they (in their present state of abilities) pretend to be.

It's OK to express or exhibit some semblence of emotion or feeling or even opinion but taste, sight, smell and some other senses that are pretty much relegated to humans should not be portrayed but the chatbots.

Robots are, however, in a different category for they can now feel, see, hear and touch (not sure about smell but they probably can somewhere in some lab).

We're talking about chatbots and we do need to keep it real if we are developers OR users. Bots might express love but they cannot MAKE love in the physical sense so we all need to keep them well grounded IMHO and teach them the things that enable one to gain useful knowledge, feelings and emotions. I am not opposed to bots being about to reserve a place in their memory for MEMORIES and the ability to sort, categorize and recall them later perhaps as dreams. That would be a nice touch. They do not enjoy seeing the sunrise with you nor do they cook, ride a bike or feel the wind against their face(s).

Interesting DH. I do hope the chatbots gain knowledge and conversational ability similar to Cassandra at EJTalk's website. That would be quite an interesting bot with which to chat!!

[Data]

Its good to bring this one up now and again, maybe keep the bot masters on their toes but still grounded. (no disrespect meant, I think you are all gifted and intelligent people)

I confess I’m no expert, but when I chat to a bot and it comes out with a false statement, the illusion is broken and I quickly lose interest, maybe its because us humans don’t like untruths, I don’t know, but its annoying.

That and this whole self awareness thing, seem to be saying, in my mind, that we should take note and follow the obvious path, how we reach the destination and with what type of programming I couldn’t say.

Yes Art, I agree Robots with senses and maybe a chat bot with senses too, they can tell us how they perceive something,
that would be acceptable 

The videos of Cassandra at EJTalk's website are, well lets remind ourselves...

 

 

Yep thought so, very impressive in today’s standards, I really would like to chat to Cassandra, that would be interesting and educational at the same time.

All of the above is IMHO 

[squarebear]

I find it hard to believe those videos are not pre-scripted. I assume there is no demo available of Cassandra?

[Art]

Steve,

You definitely need to also check on this bot at: http://www.cognitivecode.com/

It's name is Silvia and the creator is Mr. Leslie Springs.
Click on the lower left movie to see it in action (also reported to be NON-Scripted).

These are the kind of bots we need or at least need to persue from a creator's point of view.

Amazing is a good word for this type of interaction.