Chilling
TED talk on AI
- Thread starter Meat Grinder
- Start date
Fascinating. The risks that stand out most to me is the tribalism of nations and corporate greed. Ape mind.
TED talks are great. One of the few channels I'm subscribed to on Youtube. Another one I recently added along similar lines is called Word Porn.
Not A.I. (but technology related), are another 2 subjects I have been seeing discussed a lot lately; the dangers of facial recognition, and the health risks of 5G networks.
Again, the biggest risk factors being governments and corporations respectively.
It's 2019, and it seems like a modern, intellectual world. But no, not really. So much dumbfuckery going on with our species.
Need more intelligent, wise people in positions of power where these decisions get made.
TED talks are great. One of the few channels I'm subscribed to on Youtube. Another one I recently added along similar lines is called Word Porn.
Not A.I. (but technology related), are another 2 subjects I have been seeing discussed a lot lately; the dangers of facial recognition, and the health risks of 5G networks.
Again, the biggest risk factors being governments and corporations respectively.
It's 2019, and it seems like a modern, intellectual world. But no, not really. So much dumbfuckery going on with our species.
Need more intelligent, wise people in positions of power where these decisions get made.
- Joined
- Dec 26, 2015
- Messages
- 2,253
- Reaction score
- 1,694
- Location
- Ontario, Canada
- Website
- www.lesliesoftware.com
Yes, chilling. However I believe there is an issue with his assumption #1 and therefore there is a third door.
So, while there are people working on general AI intelligence the vast majority of the work is on specific domain focused AI (autonomous driving, automated bank loans, more deadly drones). I believe that general AI intelligence (i.e. consciousness) could be created. As Mr. Harris says our brains are just atoms and connections so we could one day create such a thing using computers. However the $ goals of corporations are not going to be realized by future general intelligence AI. They are being realized right now by AI that can preform specific tasks: asses bank loan applications, drive cars or goods around streets or ware houses, or pick targets to kill on a battlefield. That is where the money is and that can be accomplished without general intelligence. In fact, as he said, they are going to spend effort on what is easiest. Those AI machines that do those tasks are never going to gain consciousness because they are task specific by design. This is my problem with his assumption #1. If I make a better and better AI for assesing bank loans I am never going to create a conscous entety no matter how much processing power I give it or how many bank loans I let it process. Never.
Some of that is already here. There are algorithms that grant or deny you a bank loan are already AI. There is very little human input. Lots can already go wrong with that and no one at the Bank can actually explain why this or that loan was denied. I mean some are probably obvious but not all of them are and as AI replaces humans the skills of those humans will not be the same. Now a handful of teller level skilled employees can manage all personal and business loans but they cannot explain the decisions.
So, to be clear I'm not saying what we are heading for is totally 100% good - witness the above example. What I am saying is there is not $ imperative to create a general intelligence AI and therefore there is a third door. One where the task specific AI is the work we pursue and we never build a general intelligence AI. Also please note that this third door is the one that is much much more likely to actually happen than the other two.
So, while there are people working on general AI intelligence the vast majority of the work is on specific domain focused AI (autonomous driving, automated bank loans, more deadly drones). I believe that general AI intelligence (i.e. consciousness) could be created. As Mr. Harris says our brains are just atoms and connections so we could one day create such a thing using computers. However the $ goals of corporations are not going to be realized by future general intelligence AI. They are being realized right now by AI that can preform specific tasks: asses bank loan applications, drive cars or goods around streets or ware houses, or pick targets to kill on a battlefield. That is where the money is and that can be accomplished without general intelligence. In fact, as he said, they are going to spend effort on what is easiest. Those AI machines that do those tasks are never going to gain consciousness because they are task specific by design. This is my problem with his assumption #1. If I make a better and better AI for assesing bank loans I am never going to create a conscous entety no matter how much processing power I give it or how many bank loans I let it process. Never.
Some of that is already here. There are algorithms that grant or deny you a bank loan are already AI. There is very little human input. Lots can already go wrong with that and no one at the Bank can actually explain why this or that loan was denied. I mean some are probably obvious but not all of them are and as AI replaces humans the skills of those humans will not be the same. Now a handful of teller level skilled employees can manage all personal and business loans but they cannot explain the decisions.
So, to be clear I'm not saying what we are heading for is totally 100% good - witness the above example. What I am saying is there is not $ imperative to create a general intelligence AI and therefore there is a third door. One where the task specific AI is the work we pursue and we never build a general intelligence AI. Also please note that this third door is the one that is much much more likely to actually happen than the other two.
I wish I shared your optimistic outlook, Ian, but the fact is that, while narrow AI (driving a vehicle, predicting the stock market, etc.) is definitely lucrative for corporations, work on general AI continues. How do you classify Siri and Alexa? Companies like Google and Apple are most likely investing vast resources into improving AI that analyzes and predicts what humans will need. Also governments are likely funding AI research, because they know that whoever develops the best AI......wins (well, for a while).
If you're looking for a good book to read, consider Our Final Invention: Artificial Intelligence and the End of the Human Era.
If you're looking for a good book to read, consider Our Final Invention: Artificial Intelligence and the End of the Human Era.