This is a pretty good introduction. However, in some ways it is swallowing the Kool-Aid of the AI companies trying to hype their stock and products and the love of sensationalism by the media. I've been in software development since 1976. Here is a translation of some of the key points.
Originally, what was called a computer program becam…
This is a pretty good introduction. However, in some ways it is swallowing the Kool-Aid of the AI companies trying to hype their stock and products and the love of sensationalism by the media. I've been in software development since 1976. Here is a translation of some of the key points.
Originally, what was called a computer program became an app for marketing purposes. Now, it is called narrow AI for the same reason. As explained in the article, they are just computer programs but every company wants to put AI into its product name to make it seem up to date.
What we think of as AI is simply pattern matching and process automation. That has been going on in computer science since before it was rebranded as AI. A good example of the ability of AI is the ability to recognize speech and convert it to text. There are rules and patterns. In its simplest form, it is a dating website that matches profiles.
There is a lot of media hype over AI programs, finding patterns and data that humans haven't. What they don't tell you is that many patterns mean nothing. It takes a human to determine if the patterns are important or not. Furthermore, it takes a human to tell the program if it is working well or not. And to establish parameters and rules. Therefore, the biases and beliefs of the humans that are directing the AI are Incorporated into the results. So AI is not independent. It is very biased according to its programmers and directors. For example, AI is used to remove hate speech and lies from social media. The natural result is also the removal of political statements contrary to the beliefs of the programmers.
And finally, there are many experts who say that true or strong AI will never be achieved. The possibility helps AI companies raise billions of dollars in venture capital. But just establishing a neural network doesn't mean AI can be achieved. That would take a whole article on its own. Research the skeptics of AI if you wish to learn more.
Using Elon musk's robot is misleading because all it is doing is repeating the actions taught to it by humans wearing sensory suits performing the same actions. There is no intelligence there. The same for automated driving. He is using the film footage from all the Tesla autos to teach it how to drive. It matches the patterns, it doesn't make intelligent decisions. No AI has any intelligence to date.
wait for the next essay on this topic. "No Speciation Without Representation"
;-)
I just needed to provide a general introduction so that we do not leave readers behind that might wish to follow along and participate in the discussion.
RE Tesla robot. Still in the development stage. Think it through. This is parallel development. The mechanics separate from the processor. You think that it will not incorporate AGI once it is available? Where do you think Elon is going here?
Elon tried the AGI approach for years and believes it isn't feasible, at least for practical tasks like driving and performing robotic tasks. At least, that's what he reported when he switched to direct mimicking and to using his millions of hours of film footage of driving. Naturally, if AGI ever comes into reality, it will be used for everything possible. And my remarks do not mean that AI isn't an important breakthrough. But it is more analogous to the breakthrough of modern search engines that founded Google. It's great for regurgitating a summary of information and for mimicking writing style. At least that's my educated opinion.
Yep. We all have biases. Imagine a guy programming a robot for mcdonalds. Subconsciously he prefers cheeseburgers and thus his programming making hamburgers is unconsciously less detailed so the robot makes great cheeseburgers but...
On a low level it is utilitarian… on any other level it can become quite dangerous. I think there is probably a direct link to all the desired/called for censorship.
Seems to me there are two categories of danger: 1) where AI functions as intended; 2) where AI makes mistakes and doesn't work as intended.
We think of computer programs as being objective and coldly analytical when run correctly, but consider the subject of genetic racial differences. We know there are mean genetic differences between races. An AI program to assign work opportunities developed by Democrats (equality) would be very different from an AI developed by populists (assignment by merit). At least until populists realized the consequences of their approach.
The use of AI allows leaders to offload difficult decisions and likely sets results in stone. Once the original developers have died off, nobody knows how to change whatever program they're relying on, however dysfunctional.
Then of course, if you develop an AI program able to detect any dissidence before it get a chance to spread, there is no chance at all of correcting mistakes or abuses.
And imagine if they allow the AI to write its own code based on everything it is trained on but without a conscience using only the limitations written by the original programmers. What could go wrong? 😒
This is a pretty good introduction. However, in some ways it is swallowing the Kool-Aid of the AI companies trying to hype their stock and products and the love of sensationalism by the media. I've been in software development since 1976. Here is a translation of some of the key points.
Originally, what was called a computer program became an app for marketing purposes. Now, it is called narrow AI for the same reason. As explained in the article, they are just computer programs but every company wants to put AI into its product name to make it seem up to date.
What we think of as AI is simply pattern matching and process automation. That has been going on in computer science since before it was rebranded as AI. A good example of the ability of AI is the ability to recognize speech and convert it to text. There are rules and patterns. In its simplest form, it is a dating website that matches profiles.
There is a lot of media hype over AI programs, finding patterns and data that humans haven't. What they don't tell you is that many patterns mean nothing. It takes a human to determine if the patterns are important or not. Furthermore, it takes a human to tell the program if it is working well or not. And to establish parameters and rules. Therefore, the biases and beliefs of the humans that are directing the AI are Incorporated into the results. So AI is not independent. It is very biased according to its programmers and directors. For example, AI is used to remove hate speech and lies from social media. The natural result is also the removal of political statements contrary to the beliefs of the programmers.
And finally, there are many experts who say that true or strong AI will never be achieved. The possibility helps AI companies raise billions of dollars in venture capital. But just establishing a neural network doesn't mean AI can be achieved. That would take a whole article on its own. Research the skeptics of AI if you wish to learn more.
Using Elon musk's robot is misleading because all it is doing is repeating the actions taught to it by humans wearing sensory suits performing the same actions. There is no intelligence there. The same for automated driving. He is using the film footage from all the Tesla autos to teach it how to drive. It matches the patterns, it doesn't make intelligent decisions. No AI has any intelligence to date.
wait for the next essay on this topic. "No Speciation Without Representation"
;-)
I just needed to provide a general introduction so that we do not leave readers behind that might wish to follow along and participate in the discussion.
You certainly accomplished that.
RE Tesla robot. Still in the development stage. Think it through. This is parallel development. The mechanics separate from the processor. You think that it will not incorporate AGI once it is available? Where do you think Elon is going here?
Elon tried the AGI approach for years and believes it isn't feasible, at least for practical tasks like driving and performing robotic tasks. At least, that's what he reported when he switched to direct mimicking and to using his millions of hours of film footage of driving. Naturally, if AGI ever comes into reality, it will be used for everything possible. And my remarks do not mean that AI isn't an important breakthrough. But it is more analogous to the breakthrough of modern search engines that founded Google. It's great for regurgitating a summary of information and for mimicking writing style. At least that's my educated opinion.
I’m thinking the Jetson’s
But where are our flying cars!
I’m sure Elon is working on those😉
The very thought I had when I heard of AI....who determines what is in the data base is it draws from? Bias is inevitable.
Garbage in, garbage out. OR select the database to get the AI bias that you want
Exactly. If Marxist or a hybrid control the data used…
Yep. We all have biases. Imagine a guy programming a robot for mcdonalds. Subconsciously he prefers cheeseburgers and thus his programming making hamburgers is unconsciously less detailed so the robot makes great cheeseburgers but...
On a low level it is utilitarian… on any other level it can become quite dangerous. I think there is probably a direct link to all the desired/called for censorship.
Seems to me there are two categories of danger: 1) where AI functions as intended; 2) where AI makes mistakes and doesn't work as intended.
We think of computer programs as being objective and coldly analytical when run correctly, but consider the subject of genetic racial differences. We know there are mean genetic differences between races. An AI program to assign work opportunities developed by Democrats (equality) would be very different from an AI developed by populists (assignment by merit). At least until populists realized the consequences of their approach.
The use of AI allows leaders to offload difficult decisions and likely sets results in stone. Once the original developers have died off, nobody knows how to change whatever program they're relying on, however dysfunctional.
Then of course, if you develop an AI program able to detect any dissidence before it get a chance to spread, there is no chance at all of correcting mistakes or abuses.
And imagine if they allow the AI to write its own code based on everything it is trained on but without a conscience using only the limitations written by the original programmers. What could go wrong? 😒
And this discussion is based on our country. Given the amount of viruses created by Bad actors, it's hard to imagine a good future.