Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    DJ Tinkers
    @TinkersDj_twitter

    Process of elimination.

    We need to think of ways allowing the computer to produce code from the tests. Or else it will loop till the end of time.

    JomoPipi
    @JomoPipi
    It can just hard code the answer in a memo
    in that case
    DJ Tinkers
    @TinkersDj_twitter

    Speaking of hardcoding,

    if there were small number of outputs,

    then a hardcoded hashtable would be the most optimized solution.

    The npm module "memoize" does that.
    Also caches recent inputs
    JomoPipi
    @JomoPipi
    So I guess we did it hooray :D
    DJ Tinkers
    @TinkersDj_twitter
    No. 1 simple feature of a million.
    JomoPipi
    @JomoPipi
    so, what defines an AGI? What are the minimum requirements?
    DJ Tinkers
    @TinkersDj_twitter

    To be honest. Don't know.
    My contribution is writing software that optimizes software.
    Current CPU forecasts for AGI demands deep pockets, HUGE teams spreadout, closed door company secrets, ETC.

    That's bull donkey. Solve the software problems first. Coordinating AI's development is only yielding loss in translations, waisted debates, profressionalls accidently reinventing the wheel.

    Softwares problem is it's writers scalability. Does any one person no what's happening at all angles/stages anymore?

    DJ Tinkers
    @TinkersDj_twitter

    Our unnecessary complexities outgrown us. Many software fields like Applications development creates frameworks rather a narrow AI automating the field. Web, Mobile, & Desktop Applications development are architecturally split up... why? Why ? Why?

    Everything something in common, yet those fields treat each screen like a project. Terrible.

    I have a project hidden away. It's called autotemplater. It's a semi ML solution for automating content display. All one must do to the wield that tool is fill out the data such as a paragraph, image, title. No need to tell it positions, colors, sizing, etc. It runs through the permutations.

    In short on that note. There are a few fundamentals which desperately need automation to save everyones mind.

    Logic,
    Security ... Memory Inspection,
    Interfacing,
    Performance Optimization.

    Interfacing is like a UI, or an API, or how anything reacts with another.

    Voice is an interface. Vibrations are an interface.

    Sorry for the lecture, I think the software field spreadout to quickly.
    Why do profressionals hold the title "Software Engineer" yet never converse in metrics?
    Why do profressionals hold the title Software Architect? Yet the UML diagrams are illustrated on a white board rather a digital canvas that translates it into templated code?
    JomoPipi
    @JomoPipi
    No idea lol

    My contribution is writing software that optimizes software.

    damn

    I actually think you're on to something, then.
    optimization, in my mind, seems to be the ROOT of the issue.
    Because AGI can easily be brute forced without any effort or code
    Think about what you would like to accomplish with the aid of AGI.
    JomoPipi
    @JomoPipi
    Now, guess what? If that stuff is possible, it will (definitely) eventually happen, given time is infinite.
    So, there you go. Everything you ever wanted is already going to happen, and you didn't even need to do anything.
    But that's useless. We want to be alive while it happens. So that's why optimization is key, we want to keep shortening the path from A to B.
    An auto-memo like the one you want to make sounds like it would be a decent idea. I would like to see how it progresses. Maybe we could make one?
    It can try solving questions on https://codewars.com
    A lot of the problems there have random tests, BTW. How do you plan to account for that?
    DJ Tinkers
    @TinkersDj_twitter

    Lost context, account for what?

    The tests into specDev?

    DJ Tinkers
    @TinkersDj_twitter

    Heres an example of 1 input & output someone gave me.
    Quickly figured out there algorithm.

    @James4Deutschland
    Input: ["group","groupChar"]

    Output:
    [[True, False, False, False, False], [False, True, False, False, False], [False, False, True, False, False], [False, False, False, True, False], [False, False, False, False, True], [False, False, False, False, False], [False, False, False, False, False], [False, False, False, False, False], [False, True, False, False, False]]

    You provided 1 input, & 1 output.
    That input & output was a perfect example of IO set required.
    What I saw
    What went through my head.
    Was there the input contains two strings.
    1 had 5 characters.
    1 had 9 characters.
    The output contained 1 array.
    The output array contained 9 arrays within the output array.
    Each inner array was 5 characters long.
    Hmmm. I wonder what my first assumption would be????
    DJ Tinkers
    @TinkersDj_twitter
    JomoPipi
    @JomoPipi
    However, as larger programs were built, the flaws of the approach became more apparent. It turns out that, given a set of constraints defining a particular problem, deriving an efficient algorithm to solve it is a very difficult problem in itself. This crucial step cannot yet be automated and still requires the insight of a human programmer.
    DJ Tinkers
    @TinkersDj_twitter
    Understood. Thats why I programming a "common context"
    Not using ML to create this.
    DJ Tinkers
    @TinkersDj_twitter

    specDev references existing codebases.

    The idea is to program the applications interfaces first.
    Then write an end to end test.
    Go down the pyramid and keep on writing tests.

    JomoPipi
    @JomoPipi
    I like that pyramid idea!
    I know which one you're talking about
    DJ Tinkers
    @TinkersDj_twitter
    The more SR you are, the more you don't dig in the trenches.
    Best to see what's happening.
    JomoPipi
    @JomoPipi
    SR? Whats that
    DJ Tinkers
    @TinkersDj_twitter
    Senior
    JomoPipi
    @JomoPipi
    ohh'
    DJ Tinkers
    @TinkersDj_twitter

    This module might help us with finding constraints faster

    https://github.com/fmfn/BayesianOptimization

    JomoPipi
    @JomoPipi
    OOOO
    DJ Tinkers
    @TinkersDj_twitter
    Sorry for the slow progress. Work got in the way.
    JomoPipi
    @JomoPipi
    no worries
    welcome back
    Neur0plasticity
    @Neur0plasticity

    this project is put on hold. All my software inventions require one thing to be created. A numbercruncher. The number cruncher is an ML based math learning machine. We should not program the computer mathematics we already know, rather it needs to identify these concepts itself through pattern recognition.

    An example is identifying what are prime numbers. Instead of writing an algorithm for isPrime, it should be able to come to a conclusion from calculations saying, "realized these numbers are not nicely dividable"

    JomoPipi
    @JomoPipi
    How would it know what concepts to identify?
    @Neur0plasticity
    How would it know what kind of math to try to learn?
    Neur0plasticity
    @Neur0plasticity

    Everything in math relies on number theory. Understanding numbers will be the guide to all rationalities. A computer cannot produce an equation on it's own until it becomes aware of one numbers relationship to other numbers.

    We could start off with something simple.

    See how long it takes for the machine to realize the relationships of operators outputs of counting by one and counting by 2.

    To us we know that counting by 1 will produce every number between m->n.
    And count by 2 is half of m->n.
    Also the predictable calculations you could perform.

    if you count by 1 and divide all numbers from m->n by 5 or count by 1 and multiply by 2. will the machine realize that the digits are equivalent 2,4,6,8,0 placeholders as equal to just counting by 2

    JomoPipi
    @JomoPipi

    @Neur0plasticity

    and divide all numbers

    the machine will have a hard time, there.

    Everything in math relies on number theory.

    Why do you think so? Natural numbers can be defined in terms of sets or functions