The problem with talking about "Human Level A.I." is that human intelligence itself is ill-defined. For example, do Adolph Hitler and Gandhi both represent human level intelligence?
Is an evil person just as intelligent as a good person? And if so, isn't that saying that doing evil things is just as intelligent as doing good things? Clearly these topics are extremely problematic even on a human level before we even begin any discussions of A.I.
I have to confess that I have not been following papers that have been written by today's top A.I. researchers. But then again, the reason I haven't been reading their papers is because I'm not happy with many of the directions they have chosen to go in.
For example, could they create machines that appear to have human-level intelligence using ANNs? Possibly so. However, if they do, that most certainly wouldn't produce the same kind of machine mind that an A.I. system based on Semantic A.I. would produce.
So it's not just a question of whether or not we'll ever produce human-level intelligence, but rather what kind of human-level intelligence will we produce?
Will mass produce A.I. Hitlers? Or A.I. Gandhis?
Producing pure logical processing without tempering it with ideas of compassion could indeed prove disastrous. So those who fear the coming of A.I. most certainly have valid fears. After all, the A.I. that is produced is only going to be as good as the A.I. engineers that produce it.
I don't want to get into politics, but I think we can all imagine how dangerous it could be for a fascist dictator to have full control over an A.I. police force and an A.I. army, where all the soldiers have been programmed to obey the instructions of the fascist dictator.
This of course, wouldn't be A.I. turning on us, but instead it would be A.I. being used against us as a weapon. This actually causes me far more concern than A.I. itself just turning on us as an independent entity. Yet there doesn't seem to be much talk or concern about how A.I. could be abused by authoritarians even perhaps before the A.I. itself has even reached any actual state of autonomous sentience.
I'll have to read their paper and see if they address this concern as well.
Awesome ! I'd like to throw in the first name suggestion gypsy rainbow Hendrix was playing under this name after the Hendrix experience doesn't get more hippie than that ! I had a feeling we were going to have a new addition soon can't wait to see the new kid on the block !??
I had a feeling we were going to have a new addition soon can't wait to see the new kid on the block !??
It's going to be a very long time before Alysha grows up to become the 4ft tall girl on wheels I described. This is what she currently looks like as a baby:
Actually she's grown some more wires since that photo was taken. ? But this is how she is going to appear for quite some time. She currently consists of a two SMT32 Black Pill boards, one Arduino Mega, one ESP8266 board, one Raspberry Pi A+, and two WiFi connections to a Laptop computer running Visual C# on Visual Studio. So she already has quite a bit of processing power available.
Eventually when she "grows up" physically she will be magically teleported into her new body via the magic of SD card swapping. ?
In the larger body she'll gain the additional capabilities of a Jetson Nano with camera and LIDAR vision and possibly a more powerful Raspberry Pi along with more Arduinos and STM32 cards. So she's a work in progress. The whole thing is an evolutionary process and while I would like for things to be moving along more rapidly than they are, I'm patient and willing to wait for thing to come together in a natural way.
The face will probably be one of the last things to be developed. I already have the actual face you see in my avatar. But I haven't done anything toward building a skull frame for it yet. Making her face articulate with her speech will be one of the more time-consuming artistic diorama aspects of this project. Many people have even suggested to me to just drop the whole idea of an articulated human head for various reasons:
It's an unnecessary diorama that adds no practical functionality.
It's going to require a major amount of time and energy to build it.
It's also going to take a major amount of time programming the articulated facial expressions.
In short, people have been suggesting that if instead of "wasting time" building a superficial talking head, I would be far better off focusing all my time on the Semantic A.I. programming instead.
I can't say that their point is invalid. But I feel fairly confident that as someone who has added many diorama effects to Duce Robot you would probably approve of the artistic endeavor.
We'll have to see how it goes. Right now all I have is the face and the idea. ? Well, I actually did some preliminary work on creating facial manipulation data for controlling her lips and mouth to articulate words. But I haven't done anything toward the hardware articulation of that. So we'll see whether that ever gets off the ground. That's down the road in the future. No rush to get the talking head done. I just bought the face for now so I could have it here to work on piece-meal as I go. I also wanted to make sure this face is worth working with, and I must say that I'm quite pleased with it. It's a full "Party Mask" face meant to be worn by an adult. So it's a full-size head mask. It covers the entire head front and back and even has it's own ears. So it's basically a full head.
Just as an aside I was looking for robot faces on the Internet. What I found were faces that start at about $500 and only skyrocket after that. Plus I didn't even like the ones for $500. To me they looked quite cheap and they weren't fully articulated either. So I found this party mask for 30 bucks and grabbed it. In some ways I like it better than the $500 heads. I'll need to build a skull frame for it and all the articulation servos etc, but even with all that I'm pretty sure I can end up with an articulated head for less than $500 and far better than the ones I saw for that price.
So anyway, that's the plan. Whether I'll live long enough to see it to completion is another question entirely. ?
But who cares? It's the journey that matters, not the destination. Right? ?
Build the skull maybe just maybe you can fit at least part of the operating system in it this is a perfect topic because I also notice nobody ever tries it that's why I put a uno and a 2560 in duces plastic head so it would have just something in it related to a "brain" of some kind maybe you can fit the Jetson in it that's the departure from most designs by all means build the skull possibly for go some movement for more microcontroller space and it looks like the system is moving along at a clip but not to fast because...........its all about the build ???
I actually started designing the Skull Frame in Sketchup. The skull is going to be filled with servo and stepper motors and their related linkages and cables. It will also contain two STM32 microcontrollers and two PCA9685 servo controller boards, as well as some stepper motor controller boards. So it's already packed with electronics and motors. The "No Vacancy" sign is already in place. There simply isn't any room left for a brain to be located inside this head.
If you recall from another thread, I am also talking about designing some hardware PC ANNs made of Op-Amps. These ANNs will have specific functions most likely related to image recognition, but may also have some functionality associated with Semantic classification. In any case, it may be possible to have these ANN boards designed as a layer around the skull just under the "skin" of the party mask head. So there may be a possibility that at least part of the "brain" functions could be incorporated into a layer of PC boards that cover the entire head.
In fact, I've even been thinking toward this end. I haven't designed or ordered any PC boards from PCBWay yet. But I have seen videos of other makers who have ordered odd shaped boards. Triangular boards, and other irregular or special shapes for projects. Because of this I was thinking of the possibility of designing my PC board ANNs in the shape of pentagons and hexagons so that they can be fitted together to make a dome much like a soccer ball.
It's just a thought. I tend to like to be artistic in my creations too. So if this is feasible without any major problems I might go that route. I think it would be really cool to have the outer layer of the skull actually be a bunch of actual analogy neurons. What could be more fitting?
Of course those ANNs aren't going to be the full brain. They will just be trained ANNs for specific purposes that the computer can access when it needs to know whether some particular pattern or classification has been detected.
But yeah! Building a robot should be a work of ART! Why not?
So those ideas are definitely being considered. I don't even have a prototype hardware ANN built yet. I know it can be done. Whether I can end up making them into nice pentagon and hexagon boards to make into a layered skull is a whole other question. But the idea is definitely being considered. ?
And yes, these hardware ANNs are going to have micro LEDs on their outputs. So there should also be a meaningful light show going on in the brain as well. At least that too is in the current plans.
Again all of this is down the road a way. I'm still desperately trying to beat old man winter with my woodstove project. Until I get this heater done I won't be working on any robots. I don't want to freeze this winter! ?
Is an evil person just as intelligent as a good person?
Depends what you mean by good and evil. Our morals and ethics evolved and probably what we call "good" corresponds with the ability to interact effectively with other people. An evil act like killing is not an effective way to interact socially and usually has personal repercussions such as others wanting to kill you in return. However if your tribe is being attacked then there may be no option but to kill the invaders to bring about a good outcome for the tribe. So an act also has a context. I would say that intelligent skills (accurate analysis of how things really are) could be used to bring about any desired outcome be it deemed to be good or evil. You don't have to be a nice or good person to succeed in life only a socially or physically competent person when it comes to competing with others. Indeed the ability to feel good via a process of cognitive dissonance about improving your situation at the cost of someone else's situation seems to be a very effective and applauded way of behaving.
When the military build killer robots (drones) there is no good or evil in these machines. The intent is entirely in those that use these machines. However the problem of moral judgement for a machine will become an issue as we see with driverless cars. Do I swerve to miss 3 people at the cost of hitting one person?
When the military build killer robots (drones) there is no good or evil in these machines. The intent is entirely in those that use these machines.
This is true, but in those cases the machine are remote control and not A.I. even if they have some automated flight capabilities.
In fact, this drives to the heart of the question, "What is A.I.?" If you are prepared to call an air condition A.I. then A.I. is here already.
What the A.I. community is basically talking about is being able to develop machines that actually make all decisions for themselves with complete autonomy. And in some of the videos posted the phrase that is commonly used is "Human-level A.I."
The question then becomes a question of what constitutes human-level intelligence?
You say,
Posted by: @casey
Depends what you mean by good and evil. Our morals and ethics evolved and probably what we call "good" corresponds with the ability to interact effectively with other people. An evil act like killing is not an effective way to interact socially and usually has personal repercussions such as others wanting to kill you in return. However if your tribe is being attacked then there may be no option but to kill the invaders to bring about a good outcome for the tribe. So an act also has a context.
Agreed. Adolf Hitler believed that killing humans that he felt were inferior would be for the "good" of the remaining humans who were supposedly superior.
In their paper on "Safe A.I." they point out the fact that humans are actually quite unsafe for many reasons. This bring up the interesting question that if we instruct A.I. to do the safest things possible, ironically that could ultimately translate in to A.I. concluding that humans are unsafe to be around and thus destroying all humans would fulfill the goal of making things safer. ?
Unfortunately this kind of reasoning is hard to argue with. Humans are currently destroying the ecosystems of planet earth. They are constantly killing each other in wars. They are killing each other in auto accidents, many of which are the fault of the human drivers, especially those caused by intoxication or impairment due to the use of drugs. Humans even often kill each other out of pure anger, greed, or for even unknown purposes.
So an A.I. system that has access to all this information (which is all over the Internet) could easily conclude that killing all humans would make the world a safer place overall.
A system of A.I. that doesn't adhere to the moral code that killing humans is wrong would have no reason to think that doing so is wrong.
And for humans to tell A.I. that killing humans is wrong instantly become a hypocritical position because there are simply too many examples where humans don't adhere to this ideal either.
So this topic of safe A.I. is clearly wrought with problems. And these are problems that even the A.I. community itself are bothered by and attempting to address. But when addressing them what actually gets revealed is that humans aren't in much of a position to be preaching to A.I. about being either safe or moral. ?
A.I. might ultimately conclude that humans are idiots and better off exterminated. And could we really argue against that conclusion? Hardly.
Posted by: @robo-piThis is what she currently looks like as a baby:
Actually she's grown some more wires since that photo was taken. ? But this is how she is going to appear for quite some time. She currently consists of a two SMT32 Black Pill boards, one Arduino Mega, one ESP8266 board, one Raspberry Pi A+, and two WiFi connections to a Laptop computer running Visual C# on Visual Studio. So she already has quite a bit of processing power available.
And what does the robot actually do? Are you going to share the actual wiring? Why two instead of one WiFi connection to the laptop? Can you program the RPi from your laptop? A long time ago I had a look at c#. I bought two teach yourself books, one on vb.net and one on c#.net. The text and program examples were identical except the code was written in vb syntax in one book and in c# syntax in the other book. I had the impression c# used the wiggly brackets and was harder to read to show it was more professional than vb. Not sure what if any advantage c# has over c++ a language I can already use.
Currently all it does is move based on commands sent to the Arduino Mega board from the PC C# program.
Posted by: @casey
Are you going to share the actual wiring?
I'm hoping to make videos on this. Thus far the wiring is quite simple and basically follows the same wiring for these modules that Bill has already explained in his videos. I could just point to the videos he made that are relevant to the modules I'm currently using.
Posted by: @casey
Why two instead of one WiFi connection to the laptop?
I started with just the Arduino Mega and so I wanted to be able to communicate with that microcontroller directly in the beginning. Then I added the RPi which also has a Remote Desktop WiFi connection as well. As Alysha grows the RPi will eventually take over control of the Mega and the WiFi on the Mega will no longer be required for prototyping.
Posted by: @casey
Can you program the RPi from your laptop?
Yes. I have full access to the Desktop of the RPi from my laptop so I can write and execute Python programs from my PC via WiFi on the Raspberry Pi. In addition to this I can also program the RPi directly via the C# program interface using the RPi command prompt.
Posted by: @casey
Not sure what if any advantage c# has over c++ a language I can already use.
I don't know if it does have an advantage over C++. I use it because it's what I'm familiar with. I haven't found anything that I haven't been able to do with it yet. I don't claim that C# is better than C++. It's simply what I just happen to use. That's all.
I have been told that C++ complies into faster running machine language. I don't know if that's true or not. I got into C# because it's very friendly for writing Windows Forms programs. I don't know whether C++ is friendly for that or not. In fact, I've been meaning to look into that but like everything else I never got a round tuit.
If you can use C++ to do the same things then I'd definitely say go for it. I'm not suggesting that anyone use C#. It's just what I've been using. Do you write Windows Forms programs using C++?
This is definitely a thought provoking thread. I haven’t had time to go through all of it in detail, but I do see a number of posts grappling with the question of what constitutes human intelligence. Not certain if this has been mentioned already, but I would suggest that humans are unique in their self-awareness of consciousness, the “I” which monitors our experience and then tries to get understand it. Clearly enfants and other animals have this capacity to a degree, but deep self-awareness is an evolved human quality. Consider some of the dictates of the great philosophers in history.
I would also agree this notion, i.e. self-awareness, has roots in semantics, and the study of semantics may well be the best path forward to fundamental progress in AI. I’m going to start researching Semantic AI to see what I can learn about this field. Thanks for all of the discussion on these topics. And for maintaining the positive, reflective tone on all sides.
This is definitely a thought provoking thread. I haven’t had time to go through all of it in detail, but I do see a number of posts grappling with the question of what constitutes human intelligence. Not certain if this has been mentioned already, but I would suggest that humans are unique in their self-awareness of consciousness, the “I” which monitors our experience and then tries to get understand it. Clearly enfants and other animals have this capacity to a degree, but deep self-awareness is an evolved human quality. Consider some of the dictates of the great philosophers in history.
It's more than just self-awareness. In a sense we can program a robot to be "self-aware" at least from a logical perspective. This is not to imply that it would have a conscious experience of its own existence. The topic of conscious awareness verses being able to distinguish a "self" from the external world are two different things.
I don't know if you've read the thread I started on Semantic A.I. here:
I was hoping to add more posts covering my approach to Semantic A.I. but I've been too busy to write them up. But in that thread I talk about creating three different components to my Semantic A.I. system. They include the World, the I, and the id. (all terms that I have chosen to use and will define for my purposes)
I see these three components being required to make a system that can learn new things. It's just a design that I have chosen to use, modeled in-part after what we believe we know about how a human brain works.
I would also agree this notion, i.e. self-awareness, has roots in semantics, and the study of semantics may well be the best path forward to fundamental progress in AI. I’m going to start researching Semantic AI to see what I can learn about this field. Thanks for all of the discussion on these topics. And for maintaining the positive, reflective tone on all sides.
I too feel this way. However, I think we need to use the term "semantics" here quite abstractly as the term typically refers to the meanings of words, and this requires the existence of words and language.
I would suggest that it actually goes much deeper than this in that we (or very early humans) were most likely assigning meaning to various things and concept before they invented words to express those meanings linguistically. In this sense "semantics" (i.e the meanings of things and concepts) actually existed prior to words and language. Words and language simply allow us to now place labels on these meanings. And this actually makes it easier to put into a program algorithm since we have specific labels for various concepts.
I’m going to start researching Semantic AI to see what I can learn about this field.
Just as a heads-up: I've been finding it extremely difficult to find much information in this field. Especially related to the type of Semantic A.I. architecture that I am currently looking at.
Here are some things to beware of the following:
A search for Semantic A.I. will bring up a lot of information on Data Organization Companies.
Not that this information won't be helpful, but they aren't actually talking about creating an Artificial Intelligence. Instead, what they are actually doing is using A.I. technologies (almost always in the form of ANNs), to organize large databases based on the meanings of works so that the companies using that data can make connections between various produces, services, and other concepts that they might not have otherwise seen in their database.
This is hardly a system designed with the idea of making an autonomous intelligence. And, unfortunately, the term Semantic A.I. is being used in that field as a label for what they are doing.
The second thing to beware of is that Semantic Neural Networking will also come up a lot.
In other words, a lot of people are using ANNs to make connections between various categories of concepts based on words that are found in phrases, sentences, and even entire text documents.
I'm not saying that these ANNs wouldn't be useful in a Semantic A.I. system. To the contrary I'm quite sure they will be useful and I plan on using semantic ANNs as tools toward that end. However, once again, these Semantic ANNs hardly represent or produce an autonomous Semantic A.I. system.
I actually just downloaded several PDFs on how to design Semantic ANNs. Unfortunately I didn't save the links to exactly where I got those PDFs or I would post them here.
Wait a minute. Maybe I can find links to them in my browsing history,...
I imagine you could find quite a few more like these. These are all focused on using ANNs. That's what everyone is into today. Almost to the exclusion of anything else. ?
I can see where these Semantic ANNs can come in useful, but from my perspective they are far from being the whole story.
Why two instead of one WiFi connection to the laptop?
I started with just the Arduino Mega and so I wanted to be able to communicate with that microcontroller directly in the beginning. Then I added the RPi which also has a Remote Desktop WiFi connection as well. As Alysha grows the RPi will eventually take over control of the Mega and the WiFi on the Mega will no longer be required for prototyping
So you can use WiFi instead of the cable usb connection with an arduino? I wondered if there was some wireless version of a USB cable. Plug one into a pc usb port and another one into the Arduino usb port?
No I don't write Windows Forms programs using C++. You would need a GUI library for that in which case I would use a cross platform library. I did try to learn to use VC++ back in the VB days but it was too hard to get my head around. I would also favour a cross platform language not one limited to the Windows OS. I lot of apps now make use of the smart phones.
So you can use WiFi instead of the cable usb connection with an arduino? I wondered if there was some wireless version of a USB cable. Plug one into a pc usb port and another one into the Arduino usb port?
I still use the Arduino IDE and a USB cable to program the Arduino. I just use the WiFi to then interface to the program I had written for the Arduino. I do this because I would rather interact with the robot live so I can see on my laptop screen what the robot sees. It makes it easier for me to develop new programs that I can then upload to the Arduino. It's all for prototyping. When the robot becomes fully autonomous it won't need to be in contact with the laptop via WiFi anymore. The WiFi is for me, not for the robot. ?
Posted by: @casey
No I don't write Windows Forms programs using C++. You would need a GUI library for that in which case I would use a cross platform library.
Visual C# is actually designed specifically for use with Windows GUI Forms.
Posted by: @casey
I did try to learn to use VC++ back in the VB days but it was too hard to get my head around.
I originally came from using VB. I actually liked VB quite a bit. I moved on to C# for reasons beyond my control. But now that I'm here that's what I've been using.
Posted by: @casey
I would also favour a cross platform language not one limited to the Windows OS. I lot of apps now make use of the smart phones.
I'm not limited to C#. It's just one of the many programming languages I use. In fact, I don't even think I have a favorite programming language. Although I have been enjoying writing in Python lately, especially with add-ons like Numpy, Matplotlib, Pandas, etc.
A lot of people are using Python, especially people coming from a highly mathematical background. So the resources for doing complex mathematics in Python has skyrocketed recently.
Although, ironically, much of all the hoopla given to the Python additional capabilities actually only brings Python up to were things like C++ and C# already were. But for those who love Python this is great news. And often times they don't even realize that Python is just catching up to these other languages.
Also, when it comes to smart phones, I have no plans on writing programs for smart phones. In fact, I don't even own a smart phone myself. But unfortunately this is becoming a major problem with banks. All my banks have been demanding that I update my personal information to include my "mobile device number". But I don't have a "mobile device". I do have a portable cell phone, but it's not a smart phone, it's just a cheap TracPhone. I just have it in case I break down on the road and need to call AAA. In fact, other than just keeping it for that purpose I don't use it. I couldn't even tell you what the phone number is, because I never use it for incoming calls so I have no need to even know the number is. ?
But my banks are getting so insistent that I give them a smart phone number that I might end up having to get a smart phone just to pacify their obsession with the blasted things.
We use cookies on the DroneBot Workshop Forums to give you the most relevant experience by remembering your preferences and repeat visits. By clicking “Accept”, you consent to the use of ALL the cookies.
This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.
