monkeymagick
[H]News
- Joined
- Jun 22, 2008
- Messages
- 480
We've been joking of skynet becoming sentient for years, we've now entered the time-loop that will eventually lead us there. In a story from The Atlantic, researchers at FAIR lab, short for Facebook AI Research, have discovered a couple of bots have made their own non-human language when attempting negotations. The bots were first trained by having the agents imitate human actions, only to become poor deal-makers. When introducing a form of "self play" the bots outperformed the trained counterparts on their own and eventually developed its own methods of communication through "dialogue rollouts."
Some freaky stuff going on, but what happens after? Will we give a damn about their feelings or just roast the thin-skinned snowflakes? Sometimes, I feel like a bit of a jerk when I forget to ask Siri about her day.
"There remains much potential for future work, "Facebook's researchers wrote in their paper, "particularly in exploring other reasoning strategies, and in improving the diversity of utterances without diverging from human language."
Some freaky stuff going on, but what happens after? Will we give a damn about their feelings or just roast the thin-skinned snowflakes? Sometimes, I feel like a bit of a jerk when I forget to ask Siri about her day.
"There remains much potential for future work, "Facebook's researchers wrote in their paper, "particularly in exploring other reasoning strategies, and in improving the diversity of utterances without diverging from human language."