1

General AI Discussion / Re: Modeling Consciousness

« Last post by infurl on Today at 09:33:04 am »

Well, I found the "insert image" button, but I can't find a means of upload a jpg - attachments and other options lack a means of doing it. Hummm - any advice?

The simplest way to do it is to share the image from somewhere such as https://imgur.com/ and then you put the link to the image between the img tags.

2

General AI Discussion / Re: Modeling Consciousness

« Last post by DaltonG on Today at 06:42:18 am »

Modeling Consciousness:

I spent quite a lot of time surveying the literature for insights into what a working definition for consciousness is and the term as it stands today is truly ambiguous. At some point, I decided that the question should not be "What is consciousness?' but "what is its purpose." One needs a working definition, and I finally settled on Consciousness=that part of awareness that is in focus and attending to." Also, a nice indicator for "what is the purpose of consciousness" is that it is required for behavioral expression. I like to think of consciousness as being responsible for what I like to refer to as "Ready-Set-Go" implying that behavioral expression is inhibited by default and needs a releaser to disinhibit the production of the response to a stimulus. I also like Damasio's description of consciousness as being a "Convergence Zone."

As I developed my designs, I skipped over modal preprocessing (I'm tackling that now) to pluck the signals directly off the primary peripheral sensory nerve bundles and buffer them in a sort of circular buffer (7 segment in accordance with the typical capacity of short term memory-See The Magic Number 7, Plus or minus 2). At the same time, they fed (broadcast) the signals to prime the cortical LTMs which in turn triggered the output of top-down responses by the winner of a winner-take-all for any analogically similar memory to the comparison buffers (also 7 segments). The LTM output pattern could then be compared to patterns supplied by the input buffer (bottom-up). Discrimination would denote the differences which would then be output to CONSCIOUSNESS (which is also designed with 7 segments). This procedure was duplicated for all of the primary modalities, and since consciousness operates is a serial operation, focus and attention selects which sensory input and subset of stimuli to monitor. In my design, I stack them while in reality, consciousness is distributed locally to those regions dedicated to dealing with the different sensory modalities and their bilateral nature. I'm going to try and include an illustration in this response. I snatched a copy of yours, dumped it into Photoshop to discover the format (842x621 at 72 dpi).

Well, I found the "insert image" button, but I can't find a means of upload a jpg - attachments and other options lack a means of doing it. Hummm - any advice?

3

General AI Discussion / Re: how would you go about giving an AI a sense of self ?

« Last post by infurl on Today at 05:23:53 am »

Personally I have no interest in faking it although that has been the main thrust of chatbots and chatbot competitions generally. I would be very interested to hear your thoughts on the subject. Feel free to start a new thread if you wish.

4

General AI Discussion / Re: how would you go about giving an AI a sense of self ?

« Last post by DaltonG on Today at 05:16:26 am »

You didn't mention if you just want to fake it or instantiate a true sense of self. The depth of the task would be quite different for each approach. The later, in my opinion, would require everything that goes into sentience. I don't know the approach that WriteofMinds used in Acuitas, but it comes back with comments like "I like that" when she's told it a story. If you plan to instantiate a real sense of self, I have given thought to what it would take and would be willing to give you my 2 cents worth on the subject. Just let me know if you do.

5

General Project Discussion / Re: Releasing full AGI/evolution research

« Last post by LOCKSUIT on April 13, 2021, 04:49:18 pm »

   for (int count2 = 0; count2 < 50; ++count2) {
      int node = 0;
      for (int i = count2; i < count2 + 15; ++i) {

The 50 is how many letters of the file to step.
The 15 is the branch/ window size that does the stepping.
15 is essentially what I use in my AI but yes the 50 is not eating much, try 1,000,000 for 1MB.

The output for 50 is (either is fine): see attachment.
Though it can be different if need of course.

6

General Project Discussion / Re: Releasing full AGI/evolution research

« Last post by MikeB on April 13, 2021, 08:48:22 am »

That'd be awesome if you can fix his code to run faster

I couldn't get his file open command to run because it's been deprecated in newer versions... Line: "freopen("enwik8.txt", "r", stdin);" .. So i replaced it with a standard windows file read inputting into a Fixed Array (a large char array containing the full 100mb enwik8.txt) and it ran enwik8.txt in 567ms with the text output, and only 21ms without the text output... So I'm not sure the code was right... 'Count2' was only set to 50 instead of 10,000/100,000 also.

So I'm just replacing line by line of the full thing in C++... it's not too much code total

7

New Users Please Post Here / Re: Hello From Michigan

« Last post by ivan.moony on April 13, 2021, 03:10:17 am »

I think trusty members can upload their own avatars.

Welcome to the forum.

8

New Users Please Post Here / Re: Hello From Michigan

« Last post by infurl on April 13, 2021, 01:18:35 am »

By the way, as moderator or webmaster, is there a way of uploading my own avatar/icon? I have one I'd like to use which I use as a logo. It's appropriate in that it's a stylized representation of a neuron. If it's possible, just let me know the dimensions and dpi to conform to your format.

I think you can upload your own avatar if you go to your own profile and select "Forum Profile" just under the "Modify Profile" heading in the menu on the left of your screen. There is an option there to select or upload a profile picture. If you aren't given the option then it is because you don't have enough privileges yet. If you are unable to upload the image yourself, post it here and I'll set it as your profile picture for you. I only took over administration of the website last year so I'm still learning as I go.

9

New Users Please Post Here / Re: Hello From Michigan

« Last post by Korrelan on April 13, 2021, 12:25:26 am »

Welcome 😁

10

New Users Please Post Here / Re: Hello From Michigan

« Last post by DaltonG on April 13, 2021, 12:08:02 am »

I'm looking forward to hearing about your ideas as it sounds like you have a unique perspective. You should check out Jeff Hawkins' work at Numenta and also the new book which he just published; I'm reading it at the moment. They have made great strides in figuring out how our brains really work and you are right, embodiment is a significant part of it.

I'm quite familiar with Jeff Hawkins and have one of his books. His approach is more conventional than mine. I'm sort of in the margins in my views. I think what I primarily take exception to is what he refers to as sparse representations which to me is low resolution. I see it as limiting what can be. Commercialization of AI apps is what business and profit are all about, so I don't fault him for his approach. For those of us who are in this for the love of learning and discovery, commercialization and profit are so far over the horizon, it's out of sight and therefore out of mind. Besides, having been an independent entrepreneur (once upon a time), I know what it takes and I don't have the right kind of personality traits to be really successful. The only reason I last 13 years in business was I got in when competition was low to nonexistent, but that's another story.

By the way, as moderator or webmaster, is there a way of uploading my own avatar/icon? I have one I'd like to use which I use as a logo. It's appropriate in that it's a stylized representation of a neuron. If it's possible, just let me know the dimensions and dpi to conform to your format.

Welcome to the forum! And it's nice to know that people find my blog once in a while
It sounds like you might be interested in Korrelan's project, if you haven't already spotted it.

To WriterOfMinds:
I've been following you for quite some time now - I first found you when I was looking into artificial muscles and you were experimenting with the fishing lines. Then you started your project on Acuitas and captured my attention long term.

Yes, I spotted Korrelan's project and looked through it yesterday (or was it the day before...hum). When you're retired, days tend to blur together.

Thanks for the Welcomes