Den_VR

Salary should go up, you say… how well is that holding true?

This makes me feel so much safer.

Didn’t care for Mecha Break?

Is that a change in the valuation of the pi or of the dollar?

No pelicans riding bicycles on this one, eh?

If only they would let people turn things on and off when they want to

There’s a meaningful difference between ML algorithms and the Grail that is (general) AI. Many of us spent years arguing that words matter and now with the VC backed hype train just about everything is AI, under penalty of stock market collapse

The tile references on-board ai. “On-board ai” here is automatic content recognition. Imagine my disappointment

That’s the Model Welfare speaking, not your bias.

It’s helpful for a model to be somewhat static when it’s involved in business processes.

I suspect that peanut boiling in a pot would also be problematic. The oil and water vapor and would provide a carrier.

I think the key part is the _boiling pot_.

Then there will be no more time spent trying to manage Orange Site public opinions regarding Anthropic.

Indeed, it can lead to the end of software as we know it. Instead everything may be written on the basis of a single user.

Tree law.

Situation Before: there are 14 competing standards.

Situation Soon: there are 15 trillion standards tailored to every possible situation.

2012, commercial availability (failed in 2018 and eaten by Google). 2017, used in scientific settings. 2021, sold commercially to consumers through kickstarter. 2026, multiple companies still sell light field cameras at modest quantities.

I wouldn’t say they “didn’t” take off. They’re gradually acquiring market share.

Why not both. The engineering side isn’t fraud, but the business side is doing business things “for the sake of the project.”

Orders-of-magnitude improvements in compute efficiency are needed to become a practical replacement for backprop… but those improvements are coming.

As a mixture, could activation-space search produce useful teaching targets for backprop?

Zeroth-order search would discover candidates, first-order learning would consolidate them. The potentially valuable step is converting a sparse judgment into a reusable training target.

This also changes the relevance of convexity.

Meanwhile, I’m increasingly convinced that what the regulators are to hold is fiduciary responsibility of Super Intelligence.

Doctors don’t need brilliant insight, they need to essentially leverage human medical knowledge against unwellness in human beings. This is confounded by two measures: too many patients per doctor, and too much human medical knowledge.

Unfinished, indeed.

Why J?

A speech by vice-chairman of the Federal Reserve

Seems like a story of knowing how to operate the machine of government and relationships.

What if you half spin the other direction?

The intent of OpenAI seems to have been to create a super hacking machine. It works, sometimes.

Reminds me of the Immersed team’s Visor vaporware.

Rationalists used to fear (entirely hypothetical) AI super intelligence for its ability to manipulate a human jailor. Now here we are.

Another broadcom related tragedy