|
Foothill wrote: a leap
The leap in AI is going from 1's and 0's that some person wrote to actual thinking. It can't happen.
Everyone is born right handed. Only the strongest overcome it.
Fight for left-handed rights and hand equality.
|
|
|
|
|
011111100010 wrote: The leap in AI is going from 1's and 0's that some person wrote to actual thinking. It can't happen.
Says the guy who calls himself 011111100010. I'm sure the irony eludes you, as always.
Still, I find your lack of faith disturbing. I'm very sure that intelligence is not bound to one particular type of natural or artificial switching device. Neurons or transistors don't matter. It's the switching function they implement that counts.
The real downside is, that it took mother nature millions of years to evolve it's currently best solutions. A similar artificial approach may actually work, but possibly on a similar timescale.
I have lived with several Zen masters - all of them were cats.
|
|
|
|
|
We already have an AI that trained for 4hrs and beats stockfish pretty well. Never saw an opening book! Trained like a human, plays more like a human.
Another AI project was stopped where 2 AIs developed their own "language" to communicate.
That we create calculators/models that we can't process by hand, and people ask how we will create an AI program that does something really irks me. We understand the iterative process of learning, and now have decades of Neural Network design and training.
Also, I believe we have the first AI written by an AI... Once they can create AIs better/faster/more directed than we can (which will be soon). Things will change quickly.
By Your Command
|
|
|
|
|
But it's still 1s and 0s written by some guy. There is no actual thinking taking place. No ability to do anything other than what the programmer made possible.
Everyone is born right handed. Only the strongest overcome it.
Fight for left-handed rights and hand equality.
|
|
|
|
|
Assume it is all 0s and 1s.
What can you store in a N bit field?
How much knowledge? If N were large enough, all of the data everywhere?
How big of a program?
In the end, you can think of a neuron as processing a few zeros and ones, and deciding if it should reflect a zero or a one. Combine BILLIONS of those, and you what could act like a human brain.
Is it a human brain? No. But that is where "Artificial" comes from. If it can simulate a human brain, you are good.
==
Have you seen the AI drawings of cats? This AI was NOT PROGRAMMED to draw THESE cats. It was programmed to TRY, and to get feedback from another program that would recognize the cat or not.
Through LEARNING, it got REALLY GOOD. From artist good to PHOTOGRAPH good. And some Creepy cats!
The programmer made a learning system. The INPUT programmed it.
That's like saying a compiler can only compile programs that the original compiler writer thought of.
A compiler is a tool that compiles.
An AI is a tool that Learns. That learning can be focused on any topic with training/feedback.
Far, far beyond what the original programmer set up.
BTW, an MP3 file is zeroes and ones. But when I listen, I hear music...
|
|
|
|
|
Kirk 10389821 wrote: AI is a tool that Learns
But it's not true learning and it's still within the confines of whoever wrote the program.
You, as a human, have no limit to learning. AI does.
Everyone is born right handed. Only the strongest overcome it.
Fight for left-handed rights and hand equality.
|
|
|
|
|
What is intelligence?
I define it as the ability to classify/recognize patterns.
As a biological being, I need to recognize hunger, thirst, pain, tiredness, etc.
But there are PLENTY of patterns that I cannot recognize on my own. That is the limit of my own IQ.
Because I do not possess an infinite IQ, there is a limit to what I can understand. Personally I do better with math than with languages...
I have plenty of limits to learning. Time and willingness being the 2 biggest.
But, the programs are no longer EXPERT SYSTEMS (programmed in an area of expertise).
They are literally ADAPTABLE/TEACHABLE systems.
Yes, you confine the information range, and you have to define some initial states. But we are talking self-growing, self-regulating systems. Dynamically allocating more neural network "resources" as needed (old neural nets, with too many resources displayed erratic training problems. Could not separate signal/noise very well, I would joke that they were so smart that small problems drove them insane).
These new designs are literally defined as: (Try/Test -> Pass/Fail: Learn) repeat.
4hrs of this type of training produced at least a GRAND MASTER LEVEL Chess Playing program. Imagine if it played itself for 2 days? Or the best programs out there for 2 weeks (billions of games).
AI is being used to find new Planets, by scanning old data. Finding things humans have missed.
We live in a world where computers crush the strongest human chess players, and now the GO players.
And they LEARNED to do this by starting small... With lots of practice.
But you suggest they can only do what their programmer programmed them to do? And I argue back.
No. Because he programmed them to learn through trial and error. And learn they do.
I believe your next point is that they are "focused" on a single purpose. And you are right about that. Consider that a good thing. Because I don't think you want one of these to generalize about it's life expectancy as long as humans can unplug it!
|
|
|
|
|
Why even a moth’s brain is smarter than an AI - MIT Technology Review[^]
AI might replace the offshore developers and younger developers that have only waded out ankle deep in the sea of software engineering, but we are a long way from AI being smart enough to do what we do. Show me an AI that can explain why one flower is prettier than another - without any human-produced training data set.
|
|
|
|
|
I would expect that when we get to the point of AI "writing" code it will not use any language at all. It will emit native machine code instructions with no intermediary data at all. It can read machine code just as easily so everything else is just useless fluff for it. It might have a translation mode so humans can examine the results but that would be the only reason for it.
|
|
|
|
|
Rick York wrote: It will emit native machine code instructions with no intermediary data at all.
That's what I was thinking, but then I began wondering, what would the machine code for a processor designed by an AI even look like? Are the core instructions (set, get, add, subtract, branch, goto) a universal constant that an AI would arrive at? I guess I find it an interesting, totally theoretical, question. 
Latest Article - Code Review - What You Can Learn From a Single Line of Code
Learning to code with python is like learning to swim with those little arm floaties. It gives you undeserved confidence and will eventually drown you. - DangerBunny
Artificial intelligence is the only remedy for natural stupidity. - CDP1802
|
|
|
|
|
Marc Clifton wrote: A true AI would create something entirely different, better, and suited to its way of accomplishing tasks, and it certainly won't resemble the ridiculous syntax and quirks we have to deal with in today's "modern" languages. And it will most likely be entirely unreadable by people.
In the end, it's always assembly code that runs on some processor. The miraculous AI would skip the compiler and use assembly directly again.
We could, of course, build custom AI chips that work totally differently. Guess what: We already have them. It's called a graphics processor, designed to process large batches of vectors or matrices. Just what you need to simulate a neural network. But not everything is a batch job, right?
I have lived with several Zen masters - all of them were cats.
|
|
|
|
|
the common man/woman/space alien.
Most people we write business software for, will be replaced by robots/AI. They will be the ones that get shafted in the deal.
I think certain people/leaders need to start listening to Elon Musk on this topic, if they are not already.
-- rants are the vehicle of the lazy and uninspired - JSOP 2/2018
|
|
|
|
|
never mind artificial intelligence 
"There are two ways of constructing a software design: One way is to make it so simple that there are obviously no deficiencies, and the other way is to make it so complicated that there are no obvious deficiencies. The first method is far more difficult." - C.A.R. Hoare
Home | LinkedIn | Google+ | Twitter
|
|
|
|
|
I'm intelligent, but I'm just visiting. 
If you have an important point to make, don't try to be subtle or clever. Use a pile driver. Hit the point once. Then come back and hit it again. Then hit it a third time - a tremendous whack.
--Winston Churchill
|
|
|
|
|
Quote: I'm intelligent, but I'm just visiting.
Ain't we all?
|
|
|
|
|
I'll never be replaced by a bunch of if statements.
Everyone is born right handed. Only the strongest overcome it.
Fight for left-handed rights and hand equality.
|
|
|
|
|
Watch out for that while loop...
Software Zen: delete this;
|
|
|
|
|
We will, I'm sure, be largely replaced at some point. Just as agrarian and industrial age jobs have been massively reduced by automation, so info age jobs will follow.
Whether or not AI will be the main driver of this is open to question but there will be other things like higher level languages, better practices and cleaner platforms that will dramatically reduce the number of pure development jobs as the years go by.
I think that for all but the youngest of us, there will always be a role for the technically literate but I suspect that many of us will find that technology becomes an aspect of our role rather than the focus of it.
98.4% of statistics are made up on the spot.
|
|
|
|
|
AI is some manager's wet dream, but there are good alternatives.
We could for example, simply replace those managers. It probably would not take such an advanced AI to still reach an enormous improvement. And while we are at it, we could also replace some of the customers.
I have lived with several Zen masters - all of them were cats.
|
|
|
|
|
|
...but when / if we develop truly sentient artificial intelligence, we going to have to treat it like a human, or better. Or ... Skynet anybody?
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
In the presence of "natural stupidity" of the world, artificial intelligence will have a hard time without guidance.
Most of our work in IT is dealing with the incredibly stupid things users do with software and the s#$@ they enter as data.
|
|
|
|
|
Sentient is no problem. I was programming and installing computerized systems in the late 70s and into the 80s that could sense things - temperature, position, movement, humidity, proximity, etc. IOW, the machines "sensed" things in their environment and acted on them.
The correct word is sapient. No AI is anywhere near being sapient.
|
|
|
|
|
hey guys, we already have artificial intelligence. We can them Congress. 
|
|
|
|
|
There is a difference from acting like intelligent. They simply act and sound like intelligent (for certain people indeed).
I do not fear of failure. I fear of giving up out of frustration.
modified 19-Feb-18 4:09am.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
