1. 1

    I’m not talking about Genetic Algorithms here. My thinking here is: What happens if we give computers the ability to reproduce?

    A Genetic algorithm seems better. Evolution took a couple billion years to get to the point we are now. If computers could reproduce a million times faster than cells it would still have taken over 1000 years.

    maybe we can simply teach a computer how to design (and build) better computers, show it the existing designs we came up with, and let it build its own, then through an evolutionary process, we can get them to become better and better.

    You’re confusing evolution with progress. Natural selection does not choose the more advanced or smarter specimen; it selects the one which reproduces better.

    [multi-phased network] Phase 2: Get a snapshot of the neural network that resulted from phase 1, and learn from it, or optimize it, apply learning algorithm on it.

    I’m not really sure what you’re going for here, what distinguishes this from straight training?

    1. 1

      Actually now that you challenged the idea of why not genetic algorithms. I think Genetic algorithms may be the key. The only problem is that you need a genetic algorithm that can build hardware & software, because as I argue in the post GAI is hardware + software, software alone is not enough, especially to create awareness and consciousness.

      For multi phased approach allow me to explain my thinking: Current neural networks are networks that solve a particular problem, or a series of problems, based on training data that you feed it. So a speech neural network will be fed text + speech fragment, one with search engine will be fed search data, etc.. For GAI you need a neural network that learns ‘how to learn any topic’. So if we want to solve this problem you need to feed the neural network the ‘original problem + how you solved that problem ’ which equals original input + ‘ the entire neural network that solved that problem’ and keep feeding that neural network more and more neural networks. And this is at some level what consciousness is. Your mind being aware of the exact problem it is solving, and being able to make judgements and observations about that. This at a programming level is recursion.

      Let me know your thoughts

      1. 1

        Got it. Sorry for the mistake. I will propose adding an AI , Artificial Intelligence, or Machine learning to the community as such tags are not currently available.

      1. 4

        Some quotes for a general idea of the level of the article:

        We can speed up this process (and get more interesting variations) if we get two machines to have machine sex to design a better Machine. Even better let them have an Orgy!

        Awareness will come naturally once we feed all of the data from various organs into the thinking brain.

        An operating system gives the computer the ability to monitor its ‘Organs’.

        After reading this, I did somewhat feel like the judge from Billy Madison:

        …what you’ve just said is one of the most insanely idiotic things I have ever heard. At no point in your rambling, incoherent response were you even close to anything that could be considered a rational thought. Everyone in this room is now dumber for having listened to it. I award you no points, and may God have mercy on your soul.

        1. 1

          …what you’ve just said is one of the most insanely idiotic things I have ever heard. At no point in your rambling, incoherent response were you even close to anything that could be considered a rational thought. Everyone in this room is now dumber for having listened to it. I award you no points, and may God have mercy on your soul.

          Thanks for the feedback. I’m not writing scientific literature here so keeping it fun and interesting while simplifying it for the less technical folks is the objective. Maybe I went overboard with the orgy thing :)

          1. 3

            Well, honestly, the problem was that your submission didn’t really have any information for anyone.

            It didn’t explain clearly any of the concepts you were talking about (like, neural nets don’t work like you’d sketched), it didn’t even link out to more in-depth tutorials, and the links to books on Amazon seemed to be basically an attempt at affiliate bait.

            Your speculation on computers having “awareness” doesn’t mean anything without some brief stab at defining what awareness is, same as with the ideas of emotion, and as such it came across more as pop-sci spew than anything else.

            I checked your submission history before posting/flagging (because hey, we all deserve the benefit of a doubt), but that gives us this.

            I’d strongly suggest looking over this essay a bit more, and maybe posting some articles other people have written that explore in-depth:

            • what a basic neural network is, with code and math
            • what people consider awareness to be
            • what principle is used for genetic algorithms
            • what problems people have running AI on distributed systems

            If nothing else, go find some old papers, check them for info value, and post those. That should help you at least get your karma above 0.

        1. 2

          unless I’m missing something YaCy comes close to what you’re describing

          1. 1

            I looked into YaCy and Faroo before starting to think about this. Faroo is not even close. YaCy on the other hand is close enough but comes with a number drawback such as that it is very slow (due to fraud and spam protection) and does not rely on AI/ Learning but instead on conventional ranking methods. With that said it might be a good codebase or concept to start from.

            YaCy also works conventionally by building an index and then traversing that index to respond to queries but what I have in mind (and have yet to start experimenting with) is No Index. Just a large scale global neural network that holds the information within that network. Now this comes with a million issues but on the other hand since you don’t fully understand the impact of your node on the search result (since is is part of the larger global network) you cannot in theory manipulate the search results..

            1. 1

              PeARS is also worth looking into (recently funded by Mozilla).

              About the NN, I’m not sure what that would do..

              since you don’t fully understand the impact of your node on the search result you cannot in theory manipulate the search results

              Not sure about how that would translate into practice.

              YaCy on the other hand is close enough but comes with a number drawback such as that it is very slow

              Reaching feature-parity with YaCy would require a ton of effort. But, you can always fork YaCy and try out your ideas and see how it goes.

          1. 1

            Bitcoin-style proof-of-work relies on the fact that the compute is burned, wasted for any other purpose. What’s to stop someone putting a lot of compute into indexing zillions of variations on something they want to promote? And how would you force the indexing work to be “genuine” (e.g. what’s to stop a hostile node pretending it did a lot of crawling when in fact it had direct access to the index)?

            1. 1

              great questions :) let me think about this one a little more

            1. 1

              Would the blockchain here be used as a storage database of some sort?

              It seems like the concepts from TAHOE-LAFS where people give up space and some computing power for a “distributed cloud” would work a little better here, especially considering the awesome drama going on in the Bitcoin community about their block sizes.

              Overall it does seem like a pretty cool idea, and i guess having a distributed index that any search engine system can access would make it possible to have multiple implementations of search engines.

              1. 1

                It is not exactly blockchain - but basically imagine a few things: 1- A global tree that is the index of keywords with sites, etc.. We won’t need to retain the actual content of the site only enough data for the indexing and retrieval of data. 2- Whatever additional input that is necessary to support the learning algorithm that runs and serves queries.

              1. 2

                “cloud” seems really vague to me. We have devops, hardware, virtualization, but there isn’t a containers tag. Would adding a containers tag cover the bases?

                1. 3

                  The existing tags cover most of the container-related articles I can think of. If it’s about the low-level tech of containers, in the sense of OS-level virtualization, then virtualization seems fine (I wouldn’t reserve virtualization only for hardware virtualization). And if it’s about the whole orchestrating/images/deploying/etc. side of containers, then devops is a good fit.

                  1. 2

                    here is something that I posted yesterday about Spotify moving to Google Cloud: https://lobste.rs/s/n8lqyn/announcing_spotify_infrastructure_s_googley_future

                    This does not necessarily qualify as virtualization or containers. It is mostly infrastructure / Cloud how about we add an ‘Infrastructure’ tag or a IaaS and PaaS tags?

                1. 1

                  That is pretty interesting. You never think of Google as a PaaS or anything like that, in my experience.

                  I just checked the prices for Google Cloud Platform offerings and it seems, at a glance, that my company would actually save a little bit if they switched from AWS. Doesn’t mean we’ll switch or anything, but it’s a good alternative to have in mind to prevent vendor lock-in.

                  1. 1

                    Well there is always risk of vendor lock-in especially if you start relying on Google specific technologies Google BigQuery , Cloud BigTable, etc.. I think being mindful of whether you are using a vendor standard or industry standard, and whether a vendor solution gives you an edge, and the cost of a vendor’s lock in will help you make the right decisions along the way.