https://futurism.com/google-artificial- ... -built-ai/
There is no way this can go wrong right . .
Even if the timetravelers from the future warnings about this get more and more desperate . .
Also:
I could have sworn we had a singular topic for stuff like this somewhere, but i can't seem to find it on the first 5 pages of topics in here . .
Todays THE FUTURE IS NOW: AI makes better AI than humans!
Moderator: Moderators
- Stahlseele
- King
- Posts: 5976
- Joined: Wed Apr 14, 2010 4:51 pm
- Location: Hamburg, Germany
Todays THE FUTURE IS NOW: AI makes better AI than humans!
Welcome, to IronHell.
Shrapnel wrote:TFwiki wrote:Soon is the name of the region in the time-domain (familiar to all marketing departments, and to the moderators and staff of Fun Publications) which sees release of all BotCon news, club exclusives, and other fan desirables. Soon is when then will become now.
Peculiar properties of spacetime ensure that the perception of the magnitude of Soon is fluid and dependent, not on an individual's time-reference, but on spatial and cultural location. A marketer generally perceives Soon as a finite, known, yet unspeakable time-interval; to a fan, the interval appears greater, and may in fact approach the infinite, becoming Never. Once the interval has passed, however, a certain time-lensing effect seems to occur, and the time-interval becomes vanishingly small. We therefore see the strange result that the same fragment of spacetime may be observed, in quick succession, as Soon, Never, and All Too Quickly.
- Judging__Eagle
- Prince
- Posts: 4671
- Joined: Fri Mar 07, 2008 7:54 pm
- Location: Lake Ontario is in my backyard; Canada
As far as I can tell, the paper that article is referencing is here: https://arxiv.org/abs/1707.07012
The researchers were able to automate the architecture engineering stage of a neural network design. That's totally neat, don't get me wrong. But we're still solidly in the Weak AI paradigm. The paper also found that their automated architecture engineering procedure was better than random architecture searching, but not by a huge margin.
So not quite Skynet.
The researchers were able to automate the architecture engineering stage of a neural network design. That's totally neat, don't get me wrong. But we're still solidly in the Weak AI paradigm. The paper also found that their automated architecture engineering procedure was better than random architecture searching, but not by a huge margin.
So not quite Skynet.
Out beyond the hull, mucoid strings of non-baryonic matter streamed past like Christ's blood in the firmament.
- RobbyPants
- King
- Posts: 5201
- Joined: Wed Aug 06, 2008 6:11 pm