AI and machine learning
I commented on this is the last issue, but here's the official White House line on this meeting with tech executives. I'm not for over-burdensome regulation on emerging technology, but given, well, everything lately, a more circumspect approach to AI seems like a wise idea to start with.
This is a good approach to thinking about AI -- especially if you're wary of PR backlash or unforeseen consequences of overautomation:
"So we're thinking about not just brute force [in] applying technology because it's available, but understanding what the client’s need is and then applying technology the client needs."
This is interesting in the sense that you have to wonder what the economic value of a bachelor's in AI will be worth. Also, thinking about AI in the context of other early-college courses might be a good thing for expanding minds.
Nvidia's data center business is pushing a $3 billion run rate.
But again, that doesn't mean the U.S. or other countries are behind China. In part, it just means China is a different culture with, in certain areas, less technological and institutional debt that needs to be accounted for.
I feel like a broken record, but applying AI in areas like fishing could have huge positive effects. Overfishing is a major problem, and any way in which we can optimize the practice is only going to help everybody.
I link to this not just because I'm advising a synthetic data startup, but also because it's probably true. Data is a major obstacle for certain applications, and synthetic data done right can help overcome certain availability and privacy issues.
A really detailed breakdown of Google's newly announced TPU upgrade, for the true hardware geeks.
It still never hurts to read these types of posts and see where other folks are coming from. Defining terms like AI and ML and even their subgenres is a step toward having a lingua franca for AI, which would be a good thing.
Seems like a good approach for AI systems in areas where certainty is never guaranteed. Think about how many times you hedge your actions or decisions because you realize you don't really know what's happening.
These are some solid arguments about how tech could help and hinder, especially in contract situations. The obvious fixes seem to be human involvement and judgment in the actual decision-making (AI as only a tool) and explainability of models.