Time to ask this question

I picked closet one (“I’m working on a serious chatbot, but nothing like strong-AI”)  .. .  I’ll come in around perhaps “upper medium strength AI”   (for now)
But I will have learning in mine.

Perhaps another poll.  . list of features you will have in your bot.

Need a poll that allows multiselect

The poll is designed as much as to say “My work is based on research, thus I am right, and you are wrong” .. .. and implying that others are not doing any research.

O The problem is, even if you claim to have actually accomplished it. . .it will always be a matter of personal opinion if your creation is truly intelligent.  There will always be someone that will say “Well, it works this way, or that way, thus it is not intelligent”...  People will even argue as to the terms of the tests for consciousness.  “Oh, you used THAT test for consciousness, well, I don’t agree with it because it uses XXXXX or doesn’t use YYYYY”.  It will always come down to what people simply want to believe (or not believe).

There could be another option…

( x ) I am building a serious chatbot build from the ground up, which will first be a very useful system.  Which then can be added to later, as a foundation for a higher level AI system.

anyway, its obvious which you picked Hans (“I’m developing strong-AI based on academic research in many related areas”) ...  I wish you luck . .there have been MANY over the last 60 years that have claimed they’d accomplish this…. good luck   !!!!

Victor Shulist - Mar 11, 2011:

The poll is designed as much as to say “My work is based on research, thus I am right, and you are wrong” .. .. and implying that others are not doing any research.

No it’s not, the poll is designed to find out how many are coding stuff to see what works and how many are taking the academic route by doing (a lot) of research to substantiate their views.

I started the poll because I feel many here are coding NLP-stuff, are doing research but limit their research to NLP-stuff. On the opposite site of the scale I think there are only a very few who are actually working their way through strong-AI research that has been done so far. For example I never saw David Chalmers been mentioned here, besides by me, while he is one of the well known opponents of John Searle. Same goes for all the opponents of Chomsky. Those are just examples, I have already over a hundred research papers on strong-AI and related stuff, covering psychology, philosophy and even theology, in addition to more technology related papers.

So I really like to know how many here are in the different categories.

I see

yes, my scope is to a chatbot for useful purpose such as knowledge management, I.R. and or entertainment.  Probably learning and tutoring.

Yours is a more general purpose wider objective.

thanks for clarification!

I would say I am using a chatbot to research and prototype AI systems. If my AI is strong or not is left for others to determine. Right now it is strong (broad?) but not deep.

Ultimately, I would at least like the AI to be an assistant that can automate some of my daily tasks for me.

Merlin - Mar 11, 2011:

If my AI is strong or not is left for others to determine. Right now it is strong (broad?) but not deep.

Yes, exactly.  People believe what they want to believe.

It will ALWAYS be purely subjective whether we consider a given system to be intelligent or not.  This is why I only care about what my bot will do, not what different people want to consider it to be.

As Mack the Knife said in the Dreigroschenoper to the people trying and fumbling to hang him on the gallows:
“Dilletanten!”

1. Opportunities—check at outset of Web session.
http://www.ai-forum.org/forum.asp?forum_id=1
http://slashdot.org

2. Priority—monitor the status quo.

http://code.google.com/p/mindforth
http://cyborg.blogspot.com
http://mind.sourceforge.net/cpp.html
http://www.chatbots.org/ai_zone/viewthread/240/

3. Usenet

3.1. Check as source for off-site commentary posts
http://hplusmagazine.com
http://www.ddj.com/hpc-high-performance-computing

3.2. Check German Usenet newsgroups.
http://groups.google.com/group/de.comp.lang.forth
http://groups.google.com/group/de.comp.lang.javascript
http://groups.google.com/group/de.sci.informatik.ki

4. Flickr meme insertion
http://www.flickr.com/photos/tags/smartphone/

5. Yahoo Answers
http://answers.yahoo.com/activity?show=kL6PMwtwaa (Mentifex profile)
http://answers.yahoo.com/question/index?qid=20110205162029AAfb4g2
http://answers.yahoo.com/question/index?qid=20110211090723AAnEayI

I’ve gone with the second choice (“I’m working on a serious chatbot, but nothing like strong-AI”), because that’s where I am in my journey. Over time, that will change, as my skills and experience “catch up” with my ideas and dreams. then I’ll have to change my answer, I suspect.

The question of course becomes, what do you regard as ‘valid’ research. Do papers on grammars, compiler-, database-, algorithms-, UI-design, parallel programming (unrelated to AI),... count.
What I find so strange about AI researchers is this: don’t people realize the first thing you need to learn is the craft of programming itself? How do you expect to make a beautiful statue, if you barely know how to carve wood!!

Jan Bogaerts - Mar 12, 2011:

... How do you expect to make a beautiful statue, if you barely know how to carve wood!!

Well, you could start with a slab of granite, a hammer and a chisel. Then simply remove whatever doesn’t look like what you’re envisioning.

Seriously, though. The vast majority of the research I’m doing all starts here in these threads, where I gather insights and ideas, and then search the net (or follow provided links) to learn more. I’ve taken “reams” of notes in my head, and if I were to ever actually write them down, I’d fill a CDROM (not quite enough data to fill a DVDROM, yet. ).

Anyway, I once again want to express my gratitude for everyone’s contributions here. I may not agree with everything said here, but I think it’s all valuable and helpful. Keep up the great work.

Jan Bogaerts - Mar 12, 2011:

The question of course becomes, what do you regard as ‘valid’ research.

Research that validates your views an beliefs.

Jan Bogaerts - Mar 12, 2011:

Do papers on grammars, compiler-, database-, algorithms-, UI-design, parallel programming (unrelated to AI),... count.

Of course they count…. when you talk about implementation. They don’t have any bearing on how (for example) consciousness should be constructed in AI-minds, because that is a design-issue.

Let’s not forget the engineering principle: describe the problem -> research solutions -> design the implementation -> implement it -> test it -> deliver it.

Jan Bogaerts - Mar 12, 2011:

What I find so strange about AI researchers is this: don’t people realize the first thing you need to learn is the craft of programming itself? How do you expect to make a beautiful statue, if you barely know how to carve wood!!

That again goes to implementation and not designing. An architect knows (in most cases) very little about structural integrity design of a building. Many architects design things that can not be build the way they where designed. So during building many compromises are being made towards the initial design.

To put it in other words: the one that designs the system does not have to take into account ‘how’ it is going to be implemented. I’m a software developer by trade and have been for the last 30 years, so while I do actually design the system, I’m also taking into account that it has to be implemented in some way. So I’m a researcher who actually does know how to ‘carve wood’.

Hans Peter Willems - Mar 12, 2011:

Jan Bogaerts - Mar 12, 2011:

The question of course becomes, what do you regard as ‘valid’ research.

Research that validates your views an beliefs.

and stay away from those that are against it?

I still think that the poll is setup such that the more of your own ideas you put in, and also the less of other people’s ideas (such as research papers) you put into your project, the more you are implying it will fail.  It also says ‘my approach is based on the most research, THUS I am right and will succeed, the rest of you are wasting your time.’

There was a time when most people believed the world was flat ... even if the entire planet believed it, it did not make the world flat, even if there was 1 million ‘research papers’ that say it.

A lot of these research papers have been around so long, why in h——do we not have a TT passable machine yet ? I know your getting tired of hearing it… but the proof of the pudding is in the eating

Victor, somehow you keep being strung up about my views being ‘against’ yours. You really should let that go.

Research that INVALIDATES your beliefs should be used to tune your beliefs towards something that CAN be validated (it’s called ‘learning’). The point is that almost all papers I’ve read lately, in the realm of strong AI, are in fact validating most of my model so far. Yet, research on NLP does NOT invalidate my model simply because NLP is not concerned by modeling consciousness in AI, it is focused on implementing language handling algorithms in a way that human-like language handling can be achieved (which is of course an impressive achievement if it gets worked out, by you or by someone else). NLP is important, we just differ on the implementation. The need for NLP does not invalidate my model either, as eventually my model will do NLP, but not based on loads of grammatical algorithms.

Now here is the thing; I do in fact research other’s views to see if they are maybe right and I might learn from that. And from what I’ve found, you really need to read up on Noam Chomsky and especially his opponents, because that debate gives quite some insight to the (in)probability of ‘language’ being the defining property for ‘intelligence’.

Excellent, thanks for the clarification.

Hans Peter Willems - Mar 12, 2011:

The need for NLP does not invalidate my model either, as eventually my model will do NLP, but not based on loads of grammatical algorithms.

Haha.. you’re saying my system uses “loads of grammatical algorithms” ?  It uses a single unifying algorithm.  There are several rules, both those are the same rules that you and I know about grammar.  I give it atomic rules, and it figures out for itself the combination of those rules to determine the structure of a sentence, for itself…..  loads of grammatical algorithms… puh!!  Single algorithm my friend… the rules are simply information, atomic peices of information that it assembles on its own.

I like the way you think you know how everybody elses bot works and say things like “but not based on loads of grammatical algorithms” . .. you’re so sure of yourself.

Victor Shulist - Mar 13, 2011:

Haha.. you’re saying my system uses “loads of grammatical algorithms” ?

Why do you think I’m talking to you. You really have to stop this.