Out of Time

Some realistic questions about prospective machine intelligence regulation:

… we still don’t have a concrete answer about how to effectively regulate the use of algorithms. AI is just another very complex layer added to this already complex discussion, sometimes directly related to “big data” (in the case of deep learning, for example) and other times addressing far bigger questions (in the case of sentient machines, for example).

The UF (accelerationist) response is probably predictable: There isn’t time to reach answers. Acceleration means only (and exactly) that the problem is receding, or escaping. If it would only slow down, everything would be okay. It won’t.

Kant around the back

Schmidhuber exemplifies the path, while talking about robots:

One important thing about consciousness is that the agent, as it is interacting with the world, will notice that there is one thing that is always present as it is interacting with the world — which is the agent itself.

(Some room for quibbling, but it doesn’t get serious. This is where transcendental subjectivity comes from.)

Quotable (#208)

From an engrossing discussion of AI threats by Yampolskiy and ‘Spellchecker’ (?):

An AI researcher studying Malevolent AI is like a medical doctor studying how different diseases are transmitted, how new diseases arise and how they impact the patients organism.

If the diseases concerned could read medical papers, that analogy would be perfect.