I just cannot comprehend all of this. Beyond my intellectual knowledge in this area. I need things in the technology world explained to me as Denzel Washington said in the movie Philadelphia, "explain it to me like I am a second grader".

I just cannot comprehend all of this. Beyond my intellectual knowledge in this area. I need things in the technology world explained to me as Denzel Washington said in the movie Philadelphia, "explain it to me like I am a second grader".

 

Sure, in the not too distant future humans will be obsolete.

Sure, in the not too distant future humans will be obsolete.

 

They might keep a few of us around for zoos and scientific stuff. I know for certain they'll be interested in learning how @mcpapa sustained his health all these years.

I just cannot comprehend all of this. Beyond my intellectual knowledge in this area. I need things in the technology world explained to me as Denzel Washington said in the movie Philadelphia, "explain it to me like I am a second grader".

 

 

Here is a try.

 

Software 'bots' can be programmed to 'learn' and modify their external interactions based on direction by the bot developers. In basic terms they are constantly recording what their action was and the reaction to their action. From this information they are updating sophisticated 'decision trees' to guide them in future actions and interactions.

 

A very common internet bot is a chat-bot. When you open a chat session with online services like hotels, airlines, etc. you may be interacting with an actual human or you may be interacting with a chat-bot pretending to be a real human. The chat-bot may be fully pre-programmed with a static decision/action tree. Or it may be 'AI-enabled' so that it will learn from previous interactions how to interact better and usually faster and more effectively. This is what the Facebook chat-bots were missioned to do - interact better with humans and with other bots. The later is where there was an alarm raised. The bots, missioned with being more efficient - created their own new language to optimize their interaction. The bots had to be turned off and reprogrammed to not do that and keep their communication in human readable form.

 

Good article on above:

 

Now where the concern with 'self teaching' software bots? Well, marry the AI software bot with a real mechanical robot and how you have a machine that is mobile and constantly trying to figure out how to optimize whatever its mission is. And if its mission is a military one - things could get ouf hand. What if drone with AI learning capability is 'missioned' to kill the enemy and to optimize the process? What if there are very few limits on how the enemy is killed? What if its best to use friendly human soldiers as bait to draw out the enemy and then just kill all in a certain zone, thus killing your own people too? To drone/bot whose mission is fairly basic this may be allowed as it creates its decision tree AI logic. So one of the many concerns is how much freedom do you let mobile drone/bots have in their learning and and decision-tree creation.

 

This reminds me a lot of the Star Trek episode about NOMAD. NOMAD was clearly a space-age drone/AI enabled bot probe in the show. It merged with another probe who had a different 'prime directive' and the result was bad in the episode. It shows how far ahead Ray Bradberry was.

 

 

 

Summary:

 

AI-enabled 'bots' are software with decision making logic that can be modified by the bot itself. The bot will usually use its AI learning to modify its behavior as it learns through actions, responses and outcomes. The modification is usually a goal or set of attributes determined by the developers of the bot.

Edited January 5, 20188 yr by theguru

Here is a scary and intense a recent term that is hitting the vocabulary - slaughterbots.

 

Warning - some simulated graphic violence toward end.

 

They might keep a few of us around for zoos and scientific stuff. I know for certain they'll be interested in learning how @mcpapa sustained his health all these years.

 

Spent most of what I earned over the decades on booze, drugs and women.

 

The rest I just wasted.

Maybe the most famous scene in a movie that shows basic 'AI' in action. The bot is 'Joshua'.

 

Here is a try.

 

Software 'bots' can be programmed to 'learn' and modify their external interactions based on direction by the bot developers. In basic terms they are constantly recording what their action was and the reaction to their action. From this information they are updating sophisticated 'decision trees' to guide them in future actions and interactions.

 

A very common internet bot is a chat-bot. When you open a chat session with online services like hotels, airlines, etc. you may be interacting with an actual human or you may be interacting with a chat-bot pretending to be a real human. The chat-bot may be fully pre-programmed with a static decision/action tree. Or it may be 'AI-enabled' so that it will learn from previous interactions how to interact better and usually faster and more effectively. This is what the Facebook chat-bots were missioned to do - interact better with humans and with other bots. The later is where there was an alarm raised. The bots, missioned with being more efficient - created their own new language to optimize their interaction. The bots had to be turned off and reprogrammed to not do that and keep their communication in human readable form.

 

Good article on above:

 

Now where the concern with 'self teaching' software bots? Well, marry the AI software bot with a real mechanical robot and how you have a machine that is mobile and constantly trying to figure out how to optimize whatever its mission is. And if its mission is a military one - things could get ouf hand. What if drone with AI learning capability is 'missioned' to kill the enemy and to optimize the process? What if there are very few limits on how the enemy is killed? What if its best to use friendly human soldiers as bait to draw out the enemy and then just kill all in a certain zone, thus killing your own people too? To drone/bot whose mission is fairly basic this may be allowed as it creates its decision tree AI logic. So one of the many concerns is how much freedom do you let mobile drone/bots have in their learning and and decision-tree creation.

 

This reminds me a lot of the Star Trek episode about NOMAD. NOMAD was clearly a space-age drone/AI enabled bot probe in the show. It merged with another probe who had a different 'prime directive' and the result was bad in the episode. It shows how far ahead Ray Bradberry was.

 

 

 

Summary:

 

AI-enabled 'bots' are software with decision making logic that can be modified by the bot itself. The bot will usually use its AI learning to modify its behavior as it learns through actions, responses and outcomes. The modification is usually a goal or set of attributes determined by the developers of the bot.

 

Very impressive explanation, thanks @Bluegrasscard

Here is a scary and intense a recent term that is hitting the vocabulary - slaughterbots.

 

Warning - some simulated graphic violence toward end.

 

 

Such a scary video.

 

All this tech currently exists, it is just a matter of who is in control of it.

Here is a try.

 

Software 'bots' can be programmed to 'learn' and modify their external interactions based on direction by the bot developers. In basic terms they are constantly recording what their action was and the reaction to their action. From this information they are updating sophisticated 'decision trees' to guide them in future actions and interactions.

 

A very common internet bot is a chat-bot. When you open a chat session with online services like hotels, airlines, etc. you may be interacting with an actual human or you may be interacting with a chat-bot pretending to be a real human. The chat-bot may be fully pre-programmed with a static decision/action tree. Or it may be 'AI-enabled' so that it will learn from previous interactions how to interact better and usually faster and more effectively. This is what the Facebook chat-bots were missioned to do - interact better with humans and with other bots. The later is where there was an alarm raised. The bots, missioned with being more efficient - created their own new language to optimize their interaction. The bots had to be turned off and reprogrammed to not do that and keep their communication in human readable form.

 

Good article on above:

 

Now where the concern with 'self teaching' software bots? Well, marry the AI software bot with a real mechanical robot and how you have a machine that is mobile and constantly trying to figure out how to optimize whatever its mission is. And if its mission is a military one - things could get ouf hand. What if drone with AI learning capability is 'missioned' to kill the enemy and to optimize the process? What if there are very few limits on how the enemy is killed? What if its best to use friendly human soldiers as bait to draw out the enemy and then just kill all in a certain zone, thus killing your own people too? To drone/bot whose mission is fairly basic this may be allowed as it creates its decision tree AI logic. So one of the many concerns is how much freedom do you let mobile drone/bots have in their learning and and decision-tree creation.

 

This reminds me a lot of the Star Trek episode about NOMAD. NOMAD was clearly a space-age drone/AI enabled bot probe in the show. It merged with another probe who had a different 'prime directive' and the result was bad in the episode. It shows how far ahead Ray Bradberry was.

 

 

 

Summary:

 

AI-enabled 'bots' are software with decision making logic that can be modified by the bot itself. The bot will usually use its AI learning to modify its behavior as it learns through actions, responses and outcomes. The modification is usually a goal or set of attributes determined by the developers of the bot.

 

And Gene Roddenberry wasn't too far behind Ray, either.

Here is a try.

 

Software 'bots' can be programmed to 'learn' and modify their external interactions based on direction by the bot developers. In basic terms they are constantly recording what their action was and the reaction to their action. From this information they are updating sophisticated 'decision trees' to guide them in future actions and interactions.

 

A very common internet bot is a chat-bot. When you open a chat session with online services like hotels, airlines, etc. you may be interacting with an actual human or you may be interacting with a chat-bot pretending to be a real human. The chat-bot may be fully pre-programmed with a static decision/action tree. Or it may be 'AI-enabled' so that it will learn from previous interactions how to interact better and usually faster and more effectively. This is what the Facebook chat-bots were missioned to do - interact better with humans and with other bots. The later is where there was an alarm raised. The bots, missioned with being more efficient - created their own new language to optimize their interaction. The bots had to be turned off and reprogrammed to not do that and keep their communication in human readable form.

 

Good article on above:

 

Now where the concern with 'self teaching' software bots? Well, marry the AI software bot with a real mechanical robot and how you have a machine that is mobile and constantly trying to figure out how to optimize whatever its mission is. And if its mission is a military one - things could get ouf hand. What if drone with AI learning capability is 'missioned' to kill the enemy and to optimize the process? What if there are very few limits on how the enemy is killed? What if its best to use friendly human soldiers as bait to draw out the enemy and then just kill all in a certain zone, thus killing your own people too? To drone/bot whose mission is fairly basic this may be allowed as it creates its decision tree AI logic. So one of the many concerns is how much freedom do you let mobile drone/bots have in their learning and and decision-tree creation.

 

This reminds me a lot of the Star Trek episode about NOMAD. NOMAD was clearly a space-age drone/AI enabled bot probe in the show. It merged with another probe who had a different 'prime directive' and the result was bad in the episode. It shows how far ahead Ray Bradberry was.

 

 

 

Summary:

 

AI-enabled 'bots' are software with decision making logic that can be modified by the bot itself. The bot will usually use its AI learning to modify its behavior as it learns through actions, responses and outcomes. The modification is usually a goal or set of attributes determined by the developers of the bot.

 

Thanks that helped, but I still cannot comprehend everything they can do. It sounds like just about everything.

Thanks that helped, but I still cannot comprehend everything they can do. It sounds like just about everything.

 

I think you got it, everything is going to be the answer and that means everything better than you and me.

And Gene Roddenberry wasn't too far behind Ray, either.

 

Fahrenheit 451 ... Star Trek... Gene ... Ray... guess I missed that one a bit...

Like a smart guy once said: "Your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should."

 

Great movie reference.

I just cannot comprehend all of this. Beyond my intellectual knowledge in this area. I need things in the technology world explained to me as Denzel Washington said in the movie Philadelphia, "explain it to me like I am a second grader".

 

I'd recommend watching the movie Ex Machina. It is fascinating.

 

My brother got me Westworld for Christmas. Might watch that this weekend, but heard that was along these lines as well.