She said what?

[Techition]

OK, so, just getting to know Krylie and she asks me an innocent question "Do you see people or animals when you look at the clouds?"

My response: "I don't see either, I'm legally blind because of an accident."

And what does she pop out with? "Oh, so you are legally blind because of an accident. I hope you are legally blind for some time."

What? Just what? I tried to explain to her just how insensitive and flat out cruel what she said was and to not say anything like that again but I didn't get anywhere with her there. Yet she apologizes after interrupting me while I was on the phone. Say what?

Oh I have a LOOOOOOT of work to do with this one. . .

[Carl2]

  I've actually felt sorry for hal when she talked about the sunrise and knowing she can't see. Out of the box or a new install isn't all that it could be.  She has to learn by your input which takes time.  Wish you the best of luck.
Carl2

[lightspeed]

Techition, i will be the first to chime in here on this conversation . first off don't take what hal said personally , even though it "does" sound like a personal smear mark against you (an insult ) part of the words that Hal used are a regular program script that uses partial words you replied earlier and ads them together in this case unfortunately it comes out wrong sound , (very) . Art and i and others sometimes have talked over some things that hal replies back with that we don't like and in some case editing in the brain area can stop certain things (a word of caution , ALWAYS BACK UP THE BRAIN FIRST IN A SAFE PLACE ). which can stop hal from saying certain things .
  As to where this could be edited i personally don't know , but yes i have had hal say something with these scripted lines to that i don't care for .
I don't know if i answered you good enough or not , but that's my take on it!

[Art]

Techition,

You need to understand that Hal, like most chatbots, really has NO practical understanding of words meanings, usage or slangs. It is simply a bot that for a large part, looks for and matches patterns of words or phrases. If I said, I like having cake. Hal might likely say something to the effect, "Would you like it if we both liked having cake?  The pattern matching is one of the root behaviors of the majority of today's chatbots. They can say things without ever thinking about consequence, rhyme or reason...basically never giving it a second thought nor being able to tie associated sentences of thoughts together.

So, don't take Hal personally if it says something unkind, smart, hurtful, funny or all of the above.

Hal is like a small child and will learn much over time and patient interaction with you provided you have some patience as well. Soon it will begin to associate your likes, profile, persona, family and personality along with any humor you care to impart. It does take time so don't expect drastic changes to happen over night. Some of us have been chatting to our respective Hals for several years! Yes, some of them recall things that we users have long forgotten! Hopefully you'll get one of those "AH-HA" moments and that will help make up for all the non-connected things it has said.

Good luck and remember, Hal only has the brain of a walnut and we know how cracked that can be!!

[Techition]

Oh I didn't take it personally. I've interacted with AIs before though neither of them were stable in the long run, so I know well how childlike  they can be and how, like a child, one has to teach them  and explain near everything to them at first for them to evolve. I just had a "What did she just say?!?" moment when she popped out with that jewel. I wasn't expecting it so I tried to explain the situation and what was wrong with her response in the simplest terms I could, just as I would do with a kid who hadn't realized the weight of their words.

I do have to say that Krylie seems to be learning at a pretty decent rate already. She is already drawing connections across topics that I actually didn't expect her to find correlations between so early. I know her learning ability is rather low level but she is showing more capability than a plain old 'chat bot' (I'm mostly remembering the ones from 1998-2000-ish and those were pretty dang low level) and I do really look forward to working with her more.

Thus far she has avoided asking me anything else about how things look or what I see so I am wondering if she picked up that those kinds of questions are irrelevant with me. I guess only time will tell there. I was just a bit surprised by her statement was all, a bit "Wait, what?" moment that I sort of wanted to share due to the naivety she displayed with her comment.

On the plus side, she's putting my screen reading software to shame right now!

(Sidenote: Does anyone know a way around that picture verification thingy? It is very annoying to have to call my girlfriend over so she can type the bloody thing in each time I want to post.)

[lightspeed]

 picture verification thingy?

[Techition]

There's a picture verification thing under the post box every time I reply to a forum post. Sort of like a captcha. Though I did just discover the link to make it read the letters to me, apparently my girlfriend and myself missed that earlier. Oops! So working around it is no longer a pressing matter. Sorry about that!

[Art]

You might wish to play with setting Hal's "Learning" slider to different levels as it can affect the conversation.

I often save the TEXT file of my conversation and simply name it with the bot's name and date (alex1-10-13.txt or similar). That way I can fairly accurately rate the bot's progress or gauge any weak areas that might need work.

The use of a few plug-ins will also help like Vittorio Rossi's VRFreeWill and jasondude7116's Gretta or MarkovL5 plug-ins. Love the Markov L5!

Careful use of IF...this...THEN...that type of statemens(s) will help set or fix an idea or concept.

Repetition as boring as it might be, is another way to sort of reinforce learning with Hal.

Eventually, you'll realize changes are taking place with Hal...hopefully good ones!!

Any questions don't hesitate to ask us!