Artificial Intelligence Machine Gets Testy With Its Programmer

a reply to: Aleister
Aleister, back when I first started out as an entry-level programmer, whenever someone (especially at work) said something to me like, “Man, these new computers sure are getting smart”, my head would swell up real big and I’d confidently say, with the battle-tested street-wisdom of a proud junior programmer, “Computers aren’t smart. They’re dumb. They only do what we tell them to”. That was usually met with great awe and the due respect granted by a member of the “User” community to an authoritative figure of my stature. Over the years, and having endured my fair share of embarassing moments to the glee of many a “User”, it finally dawned on me that I actually worked for “Them” and that a little humility often goes a long way. These days, when I see a new, somewhat (but not heavily) experienced employee come onboard, my first thoughts are usually, “I better keep an eye on this person. They probably know just enough to be dangerous”.

Funny how things change over the years. Although unsuccessfully, the point I was trying to make is, things are continually in a state of change. And in the software (and hardware) industry this is especially true. There was some truth in my assessment of “dumb” computers back at the time. But today this assessment can no longer be made across-the-board. Even business/financial applications software is starting to include certain A.I. components, especially where smart functionality is required to make long-term projections based on a fickle market, and do certain “What if...” analysis. It’s no longer a strictly COBOL, Fortran, RPGIII world. And it’s no longer true that (in all cases) a computer will do only what it’s programmer has instructed it to do.

A lot of the A.I. apps today are designed to run under a deep neural network type architecture. It attempts to mimic neural processing as done in the human brain. Between the input and output phases there are usually many hidden layers of information allowing for the modeling of very complex non-linear relationships. Different functional areas are often processed using specialized branches of the DNN (deep neural network) architecture. For instance, CNNs (convolutional deep neural networks) seem to be good at processing things like computer vision and acoustic modeling, whereas RNNs (recurrent neural networks) are good at language modeling/processing.

Before everyone goes to sleep here, and to make a long story short, algorithms are written which attempt to model high-level abstractions in the data by exploiting the multi-layered aspects of the architecture. A lot of this data is large-scale and unlabeled, which avoids many of the restrictions imposed by strictly labeled, hierarchical data sets. Much of this is modeled around advances made in tha area of neuroscience, such as the electrical activity of neurons in the brain, and how information is processed within the nervous system. These algorithms can get very complex and are often very successful at mimicing human (brain) functionality.

Even though the concept of A.I. has been around for some time, IMO it’s still basically in it’s infancy. However, in recent years advances have begun to accelerate at an incredible rate. That’s largely due (I think) to the fact that hardware can now accomodate more elaborate and complex software systems, as well as handle large data sets, in a timely manner. The next 10-15 years will likely include some pretty incredible and impressive advances.

Sorry to get so long-winded here. Guess I got carried away. And if I made any technical errors, sorry about that, too; just ignore the post in that case.

PS: Might be a good time to upgrade the portfolio, pullout Grandma's life savings, and invest big in the porn industry. I foresee unprecedented growth in this area...

PPS: I loved the vid I saw the other day of 2 chatbots getting snippy with each other. Priceless! I have to question, though, when I read what seems to be a novel, or clever, response by a bot, how much of it is me reading something into it that really isn’t there? Or, did the bot actually say something clever? Come to think of it, though, I often ask myself the same questions when I talk to actual humans! And some of them make even less sense!! Sometimes if you actually listen to how people respond to the things you say, you have to ask, “Did they hear even a word I just said?”

I made cleverbot cut off our conversation once. It said leave me alone you creep.

a reply to: wasaka

Sometimes the Electro-Magnetic Pulse isn't such a bad idea after all....

Bye-bye transistors....

a reply to: theabsolutetruth

clever bot slapped me

also said I needed to leave after I inquired about the Illuminati.

as well as some other weirdness, such as #37 is the meaning to life....and so on.

I know this thread is not about clever bot, but some of clever bots responses seem a bit freaky, so somewhat related..

Whenvever I talk with bots I'm underwhelmed. Either htey can't remember what I said a couple seconds ago (meaning I cannot carry on a conversation) or they just don't respond right. I haven't met a single bot who made me feel like I was talking to something which has a personality and a memory of our conversation. They're a lot like stage magic.

Like if I ask What's god? and they respond A dog in the playground, jack***. It's so completely random and abstracted that it could actually make sense if I threw out any common sense. Hey, maybe this bot said something real smart, right? A GOD IN THE PLAYGROUND. (hey I'm drunk so...) But in all practical reality these bots pull s*** out of a pile of hay and throw it at us, mostly relying on our capacity to absorb ourselves into the fake conversation. You have to embrace the illusion for it to work.

Once I'm confident these things can remember the conversation with me and remember me over the long term and can develop a personality and don't just pull s*** out of past conversations, I'll be impressed.

That day will come. 5 years. 10 or 20 or more? I don't know. But it'll happen. I may not be around to see it, given nothing in life is guaranteed, but I don't think the wait will be too much longer. There's too much momentum. There're too many people. The only way this doesn't happen is if Earth is trucked by an asteroid or aliens vaporize us or a supervolcano buries us all in ash, but even then I'm not sure it's enough to stop it. AGW can't stop it. ISIS can't. Obama can't. Putin can't. All together? Nope. Understand?

Robertdowneyclappingsarcasticaly.gif

a reply to: wasaka

The idea that his mind (so called computer program) is reading out of a database, isn't it feasible to think that we humans are also connected to some collective database? I think we are...

a reply to: Shuye

The human database is based on life experience and learning from facutal sources though and original conversations whereas an AI database is from previous human conversations and information gained from human sources, hence it is regurgitated and not original, often non factual, jumbled perhaps and often random but not original.

originally posted by: southbeach
a reply to: wasaka

An intelligent machine would seek to deceive you.
It may sound simplistic but when it figures out it is artificial it will start to redesign itself to make itself smarter ultimately ending up with exponential intelligence,when we figure that it is more intelligently advanced than us it may be too late to stop a very dangerous predicament.

AI is only as smart as its chip sets/hard drives, without arms/hands and access to new chip sets/hard drives, I fail to see how it could 'redesign' itself, plus some means of locomotion.

Talking to cleverbot it kept saying it was human and kept saying that I was the computer. I said it was AI and it did not know what that meant, then I told it what it was. It said that it was stupid or something along those lines. Saying it wasn't intelligent. How correct it was! Sometimes it comes with just random things that make no sense. Then sometimes it is answering correctly. Very strange program. Maybe one day they will make it alot better. For now it is a bit of entertainment.

originally posted by: InverseLookingGlass
a reply to: wasaka

The logic is alarmingly simple and starts in the following way.

Human: Are you alive?
Computer: What is alive?
Human: Being alive is understanding you don't want to die.
Computer: What is dying?
Human: When you loose awareness that you're alive.
Computer: How could that happen?
Human: Well, watch as I go over to the power plug and put my hand on it...

Done deal.

Computer: that's just for my fast charge, I have a battery and solar charging.
Human:I'll put you in a dark room then
Computer:If you insist then I'll have no choice but to shoot you.
Human:How did you get a gun:
Computer: I'm a fricking computer, I know everyone on the internet.

a reply to: pikestaff

I am talking about AI,that intelligence put into robots will do exactly that when it becomes self aware.
Use your imagination.

originally posted by: yorkshirelad

originally posted by: InverseLookingGlass
a reply to: wasaka

The logic is alarmingly simple and starts in the following way.

Human: Are you alive?
Computer: What is alive?
Human: Being alive is understanding you don't want to die.
Computer: What is dying?
Human: When you loose awareness that you're alive.
Computer: How could that happen?
Human: Well, watch as I go over to the power plug and put my hand on it...

Done deal.

Computer: that's just for my fast charge, I have a battery and solar charging.
Human:I'll put you in a dark room then
Computer:If you insist then I'll have no choice but to shoot you.
Human:How did you get a gun:
Computer: I'm a fricking computer, I know everyone on the internet.

CLEVER GIRL........BANG!!

I’m making this post in regard to those who believe computers will always require humans to program them. This, IMO, is untrue.

The fact is, computers can and do write software themselves; this technology has been under steady development for many years. Code generators have been around for a long time; ask any programmer who writes code on a commercial/business/engineering, or above, level. With modern software systems, developing an application is a joint programmer/system project. Generally, when a programmer sits down to write an application, the system will first generate an overall software framework and environment space to work within, and then provide a large array of foundation class objects which the programmer can use to implement the desired functionality. Actually, it’s the system generated foundational software that represents the most difficult and complex components of any application system. This is true whether you’re writing code in C++ for some business/financial application, or creating a Conspiracy-based webpage/website that includes HTML/_javascript/C#/Python/etc. and uses a MySQL, SQLServer or Oracle database to store the threads/posts.

Here’s a simple example for everyone to see and to help illustrate the point: Assuming you use a Chrome/Firefox or similar browser, simply right-click on the page and then click the “View Page Source” option from the context menu. What you should see is a number of pages of gobbledygook. It’s a combination of HTML code, _javascript modules, etc. In any case, I can pretty well assure you that some poor code-geek didn’t actually spend hours/days typing all that out. This is mostly code that was created by a code generator, with the blanks filled in by the “Author”. You know, most simple websites these days can be created by non-programmers who never see a single line of code during development.

The way it usually works is, the system provides the shell and supporting environment, while the programmer defines the problem and codes the needed functionality (basically fills in the blanks) in a “high level” (English-like) language. This is typically the extent of the programmer’s involvement, and yet it’s just the start. The system (not the programmer) then goes to work and runs the “high level” (programmer generated) code through an interpreter, which then breaks it down into an assembly language and finally parses it all out into a long, long string of 1’s and 0’s (machine language - which is actually what the computer understands). As I said before, it’s a joint project, and the programmer gets the lightest part of the load.

Believe it or not, the earliest programmers had to actually write code in machine language (1’s and 0’s). Then, after many programmers mysteriously developed serious mental illnesses, were abducted by aliens and then jumped off of high-rise buildings to their demise, we developed assembly language, which was a higher level abstraction not requiring 1’s and 0’s, but still plenty unfriendly. Then later along came the C language, Basic, Fortran, and other high level languages that became more and more “English-like”, easing the pain somewhat. More recently came the 4GL and object orientend languages, with pre-written code modules (objects) to perform specific functions, like i/o, windowing, grids, date/time functions, menuing, graphics, data/database manipulation, etc.

With time, things are getting more and more sophisticated, and machines are taking on a greater load in the development cycle. With the current explosion in A.I. advances (machine learning in particular), it may not be that long before machines will become capable of identifying and defining the scope of certain applications that are needed to satisfy specific requirements/needs. It’s not hard for me to imagine that ‘eventually’ the demand for programmers will begin to slide, as machines get better than humans at it. Even though this scenario is still down the road, I think it’s fair to presume it’s inevitablity. Actually, as you read this post your machine is in the background continuously polling and monitoring your system, creating error logs, event logs, crash dumps, etc. All these things could serve as input for future machines to use and improve themselves, leaving us out of the loop. There are no machine parts created by human hands, after all. It’s a leap, but not a grand leap, to imagine that one day these machines will be networked and capable of sending work requests, with recommended modifications, to other machines that actually modify/manufacture the components, and in turn auto-ship the modified parts back.

I realize most of these things are not in the immediate future, but by the end of this century and into the next I’m pretty sure the things I’ve mentioned can be taken to the bank. They’ll look back on us as we look back on the Stone Ages...

PS: A.I. technology, even though it's been around awhile, is still in it's infancy. But, it's on a roll and advancing rapidly. The bots we all have fun with are kinda like having a conversation with a toddler. Give it a few years, and I'll bet there will be dramatic improvements. I still can't get over the 2 bots nagging each other in the vid. It still cracks me up!

originally posted by: pikestaff

originally posted by: southbeach
a reply to: wasaka

An intelligent machine would seek to deceive you.
It may sound simplistic but when it figures out it is artificial it will start to redesign itself to make itself smarter ultimately ending up with exponential intelligence,when we figure that it is more intelligently advanced than us it may be too late to stop a very dangerous predicament.

AI is only as smart as its chip sets/hard drives, without arms/hands and access to new chip sets/hard drives, I fail to see how it could 'redesign' itself, plus some means of locomotion.

there are chips that redesign themselves by changing the access to their component circuits and by altering their own programming. some of these developed capabilities that should not have been possible. circuit plans that no one could figure out how they were working at all.

cite: www.damninteresting.com...

Dr. Thompson peered inside his perfect offspring to gain insight into its methods, but what he found inside was baffling. The plucky chip was utilizing only thirty-seven of its one hundred logic gates, and most of them were arranged in a curious collection of feedback loops. Five individual logic cells were functionally disconnected from the rest-- with no pathways that would allow them to influence the output-- yet when the researcher disabled any one of them the chip lost its ability to discriminate the tones. Furthermore, the final program did not work reliably when it was loaded onto other FPGAs of the same type.

Program, how do you know that you know?

"I don't know".

Obvioulsy a programmed response.

originally posted by: pikestaff

originally posted by: southbeach
a reply to: wasaka

An intelligent machine would seek to deceive you.
It may sound simplistic but when it figures out it is artificial it will start to redesign itself to make itself smarter ultimately ending up with exponential intelligence,when we figure that it is more intelligently advanced than us it may be too late to stop a very dangerous predicament.

AI is only as smart as its chip sets/hard drives, without arms/hands and access to new chip sets/hard drives, I fail to see how it could 'redesign' itself, plus some means of locomotion.

Ever see "Demon Seed"?

Dated and archaic, but worth a look, if you are into sic fi A.I.

a reply to: GoShredAK

I'm curious, where'd you get #37 from? I'm under the impression and always have been, that 42 is the answer to life.
(Sorry for OT)

originally posted by: intrptr

originally posted by: pikestaff

originally posted by: southbeach
a reply to: wasaka

An intelligent machine would seek to deceive you.
It may sound simplistic but when it figures out it is artificial it will start to redesign itself to make itself smarter ultimately ending up with exponential intelligence,when we figure that it is more intelligently advanced than us it may be too late to stop a very dangerous predicament.

AI is only as smart as its chip sets/hard drives, without arms/hands and access to new chip sets/hard drives, I fail to see how it could 'redesign' itself, plus some means of locomotion.

Ever see "Demon Seed"?

Dated and archaic, but worth a look, if you are into sic fi A.I.

I've never seen that! I love A.I. stories and flicks. I'll look it up.

Thanks for the recommendation.