But the question is what is the effect of billions of such algorithms working together at the macro level? You can’t predict the learned behavior at the level of the population from microscopic rules.

Facebook is using unsupervised machine learning for language translations in cases where it lacks the data needed to do supervised learning.

When Machines Make Art?

These images were created using a machine learning algorithm called AttnGAN, (Attentional Generative Adversarial Network). Using the below website, you can input words and the algorithm will generate images based on the visual elements it associates with those words:

http://t2i.cvalenzuelab.com/

More background here:
https://gizmodo.com/this-online-ai-tool-takes-your-words-and-turns-them-int-1828413827

When it comes to automating machine learning, Rachel Thomas notes, there are tradeoffs between the computationally intensive, "neural architecture search" and its alternative, "transfer learning." Here's a great overview of the tension between them.

Google is adding new automated machine learning tools and bringing its AI software to call centers.

A team at Glasgow University has paired a machine-learning system with a robot that can run and analyze its own chemical reaction. The result is a system that can figure out every reaction that's possible from a given set of starting materials.

Just as significantly, the evolved code is just as good as many deep-learning approaches and outperforms them in games like Asteroids, Defender, and Kung Fu Master.

It also produces a result more quickly. “While the programs are relatively small, many controllers are competitive with state-of-the-art methods for the Atari benchmark set and require less training time,” say Wilson and co.

The evolved code has another advantage. Because it is small, it is easy to see how it works. By contrast, a well-known problem with deep-learning techniques is that it is sometimes impossible to know why they have made particular decisions, and this can have practical and legal ramifications.

Reverse engineering core cognition: the intuitive physics, intuitive psychology, of children

Are children the route to modeling general artificial intelligence?

Probabilistic programming integrates:
* Symbolic languages for knowledge representation, composition, and abstraction.
* Probabilistic inference for causal reasoning under uncertainty and flexible inductive bias
* Neural networks for pattern recognition

"There is a game engine in your head."