My primitive "understanding" of AI is that it inherently depends upon being "trained" by massive amounts of appropriated data, the ultimate sources of which are largely uncompensated. If Ms Johannson chooses to challenge OpenAI and succeeds it will be a pyrrhic victory for the many millions of little people whose data has also been scraped but who lack the resources to do anything about it.
My primitive "understanding" of AI is that it inherently depends upon being "trained" by massive amounts of appropriated data, the ultimate sources of which are largely uncompensated. If Ms Johannson chooses to challenge OpenAI and succeeds it will be a pyrrhic victory for the many millions of little people whose data has also been scraped but who lack the resources to do anything about it.
Yes that is AI, there are decision trees that govern what path or paths an AI takes then the data lookup and finalizing the results , all of this is done in a optimized way , i.e. this is where they cheat in order to stay fast . All of the above points the process in a direction of the creators choosing . AI is not thinking for it's self there just isn't the processor speed that could really do that . This is important it's not thinking it's just following code and data and decision trees all programmed , into code or databases .
The power of artificial learning is in creating code that creates its own decision trees. Of course, it does so based on its programming.
People who are worried about what AI can do should be more worried about what people who write AI allow it to do outside of the code. One example being "self-driving" cars. In most cases, much safer than human drivers, but the mistakes they can make are not mistakes we expect from other drivers, so it's hard to share the road with them.
Another issue, as governments in the western world are doing their best to eliminate reliable energy grids, is that data centers require more and more power. Relying on applications that use modern data centers may be dangerous because they won't work when energy is rationed.
I'm not so worried about the legality of the training of AI. Everything we know that we haven't learned first-hand comes from appropriated history. It's not so much whether data is legally scraped. The important issue is whether the data is being scraped accurately. I do not trust the tech bros in California any more than I trust the bigots at MSNBC or The New York Times.
People in IT are never worried about the law until they get served , I was in the business for 40+ years and saw it a lot. AI will be served once the owners of data start seeing there data given away . I worked next to a 40 story building full of lawyers looking for the right case to sue , and talked to a few of them about how it all worked , that building was only 1 of 3 . Chicago , Germany and AU a 24 hour operation . They had there hands in a lot of pockets .
I used to build out data centers as a side job, there is always a data center that will give you 24x7 power all for a price , and these types of data centers are across the county (and a lot of the world) with 3 forms of uninterrupted power this is pretty standard these days so power will always be there for the right price . If you have the money you can run it without much worry . If your a startup using some cloud then I might worry . The big guys build clouds , the little use them.
My primitive "understanding" of AI is that it inherently depends upon being "trained" by massive amounts of appropriated data, the ultimate sources of which are largely uncompensated. If Ms Johannson chooses to challenge OpenAI and succeeds it will be a pyrrhic victory for the many millions of little people whose data has also been scraped but who lack the resources to do anything about it.
Yes that is AI, there are decision trees that govern what path or paths an AI takes then the data lookup and finalizing the results , all of this is done in a optimized way , i.e. this is where they cheat in order to stay fast . All of the above points the process in a direction of the creators choosing . AI is not thinking for it's self there just isn't the processor speed that could really do that . This is important it's not thinking it's just following code and data and decision trees all programmed , into code or databases .
The power of artificial learning is in creating code that creates its own decision trees. Of course, it does so based on its programming.
People who are worried about what AI can do should be more worried about what people who write AI allow it to do outside of the code. One example being "self-driving" cars. In most cases, much safer than human drivers, but the mistakes they can make are not mistakes we expect from other drivers, so it's hard to share the road with them.
Another issue, as governments in the western world are doing their best to eliminate reliable energy grids, is that data centers require more and more power. Relying on applications that use modern data centers may be dangerous because they won't work when energy is rationed.
I'm not so worried about the legality of the training of AI. Everything we know that we haven't learned first-hand comes from appropriated history. It's not so much whether data is legally scraped. The important issue is whether the data is being scraped accurately. I do not trust the tech bros in California any more than I trust the bigots at MSNBC or The New York Times.
People in IT are never worried about the law until they get served , I was in the business for 40+ years and saw it a lot. AI will be served once the owners of data start seeing there data given away . I worked next to a 40 story building full of lawyers looking for the right case to sue , and talked to a few of them about how it all worked , that building was only 1 of 3 . Chicago , Germany and AU a 24 hour operation . They had there hands in a lot of pockets .
I used to build out data centers as a side job, there is always a data center that will give you 24x7 power all for a price , and these types of data centers are across the county (and a lot of the world) with 3 forms of uninterrupted power this is pretty standard these days so power will always be there for the right price . If you have the money you can run it without much worry . If your a startup using some cloud then I might worry . The big guys build clouds , the little use them.