[Don Patrick]

In truth there are no advancements in conversational AI ..... to tell the truth they are all using keyword and response! or hard-coded conversational trees.

Ah, I was afraid of that since that's all I've come across too, but it's good to have that confirmed from someone who studied. Intent classifying systems have fascinating learning mechanisms, I suppose they could be used for conversational intents instead of for virtual assisting, but either way that's not the kind of NLP I have going on.

[spydaz]

Your completely right!

The achievements and capabilities of my AI is much higher and closer to the dream!
The bots of this world can be recreated - in so many ways that each time they are actually the same!
Businesses are not really interested in AI as its self learning - The are however interested in BOTS....

They are where we were 20 years ago! even ultrahal is more advanced! and it is stagnant!
I remember the XTF & VONSMITH Auto topic creating brains.... they were very advanced for the time... and yet i have not seen any thing that matches these advancements.... I have only just included my own topic creating module but it just makes sure that there is some form of response. usually i disable it to let other higher function act.... But the auto topic crating brain enables for mass information gathering;

I feel that concentrating on What gets uploaded to the data warehouse or long-term storage should only be HIGH CONFIDENT information; in the transactional databases information with low confidence can be stored... obviously working memory is the current conversation....

I actually designed my memory based on "atkinson and shiffrin model"... https://en.wikipedia.org/wiki/Atkinson%E2%80%93Shiffrin_memory_model

to cover "THINK LIKE A HUMAN".....

this is why i feel guided by psychology & philosophy...

[infurl]

In truth there are no advancements in conversational AI ..... to tell the truth they are all using keyword and response! or hard-coded conversational trees.

Ah, I was afraid of that since that's all I've come across too, but it's good to have that confirmed from someone who studied. Intent classifying systems have fascinating learning mechanisms, I suppose they could be used for conversational intents instead of for virtual assisting, but either way that's not the kind of NLP I have going on.

Google harder guys. Hint: different universities have different specialties.

[Don Patrick]

What a terribly unhelpful thing to say. I hope you're not referring to sequence-to-sequence learning as an advancement in conversational AI. Otherwise you might as well name the methods.

[infurl]

What a terribly unhelpful thing to say. I hope you're not referring to sequence-to-sequence learning as an advancement in conversational AI.

Well I'm sorry Don but I wasted a lot of time a few years ago trying to point you guys in the right direction and largely being ignored. I guess it was all going over your heads. But now to hear you stating categorically that such research doesn't exist is a bit much. I'm not posting the links again (and the work I indicated back then has advanced considerably in the meantime) so you'll just have to find the research for yourselves.

[Don Patrick]

Very well. If I've ignored it in the past, I will have had my reasons, and your reluctance is then understandable.

[infurl]

Very well. If I've ignored it in the past, I will have had my reasons, and your reluctance is then understandable.

You've been working as hard as anybody and I have no doubt that they were good reasons if you didn't have time to pay attention then. I don't mean to be an ass about it now either. I figure that if I make you go and look harder there's a strong chance you'll find something else that I missed. Then we can share and we'll both be better off. But trust me, there is some extremely advanced dialog research out there and it's ongoing and published in full by the people concerned.

[LOCKSUIT]

Hehe I made this thread...

[Don Patrick]

Sorry if we've gone off track too much, Locksuit, but I did try to keep it on the topic of learning AI, and the later topic of collaborating as you brought up.

Infurl, I'll have another look around, can't make any promises, of course.

[LOCKSUIT]

Don't worry it's only human to leave ideas wherever we happen to be lol.

[LOCKSUIT]

You got more of this stuff infurl?:

http://mindconstruct.com/

if u do tten my mouth is watering riight noww....

[infurl]

You got more of this stuff infurl?

More of what stuff? I don't have anything to do with that website.

[Don Patrick]

I posted that link, Locksuit. You could find some more about it if you search for posts by the author on chatbots.org (he's no longer active there):
https://www.chatbots.org/ai_zone/viewthread/364/
https://www.chatbots.org/ai_zone/viewthread/2125/
However, not much concrete information is known about it out here in the open. The entire project went silent between 2015 and recently. And just so there are no misunderstandings: I only posted this because it is relevant to your tastes, I'm not a fan of it myself.

[infurl]

It always seemed like a bit of a boondoggle to me.

I wouldn't have wanted to have anything to do with it.

[LOCKSUIT]

to do with.....AGI?