Notation:

If the notation is adapted to the discoveries..., the work of thought is marvelously shortened.

(G. Leibniz)

Mathematics is the art of calling different things by the same name.

(H. Poincare)

The Weighted Mean:

Does the average person understand the concept of the weighted mean?

Probably not.

Terence Tao:

https://terrytao.wordpress.com/

Depth and Breadth:

"People who know rare or difficult things or can solve very complex problems also generally know a lot more than do most people of the rather ordinary kinds of words and facts that many people know. Persons with poor reasoning and problem-solving ability also possess much less common knowledge about the world around them. Brighter persons automatically pick up more information from any experience afforded by their environment."

Arthur R. Jensen

LLMs as Toys:

Devon Erikson:

"Large Language Models, in their current form, are a big nothing burger. Some people are excited about the nothing burger. Some people are terrified of the nothing burger. Silicon Valley is breathlessly slavering over new ways they can sell the nothing burger. Our civilization as a whole is poised to invest a trillion dollars in the nothing burger. But there is nothing on the burger. So why does anyone think there is? Well, it's a novelty, for one thing. Look, you can talk to the computer. In English. And it can answer you. In English. Wow, that's convenient. And fun. And, above all, humans are curious. So something new, convenient, and fun is the very definition of a toy. And people want to play with it, to bat it around like a feather or a ball of string, and see what it can do. And for a certain period, they are willing to pay $20/month for the experience. But it goes deeper than that. Before we taught computers to talk, most humans, except for a few cognitive science specialists and a handful of very deep thinkers, suffered from a fundamental misconception about what "intelligence" is. They all thought it was the ability to speak. To turn ideas into language. And they thought humans can speak because they are smart, sapient, sentient, self-aware, conscious, ensouled, whatever you want to call it. Well, they were wrong. We don't talk because we are smart. We talk because we have subsections of the brain specifically evolved for talking. People who study the brain already knew that, but they often hadn't fully considered the implications, and were swayed by the prevailing cultural idea that eloquence and intelligence were the same thing. Some people still believed this after 8 years of Barack Obama as president. So now we have machines that can talk, and people think that's cool, but the machines can't really do anything, or at least they can't do anything right, because They don't have a world model. They don't have an imagination. They don't have desires. They don't have executive function. So of course they lie to you. That's what a glib talking device does when it's not backed up by allegiance to an objective model of reality. Really, after 8 years of Barack Obama, more people should have realized this. Here's how I think intelligence works, which I figured out from watching my cat learn how to use doorknobs: 1. You need an internal representation of your world, distilled from observation. Imagination. 2. This model needs to include not just what is, but causes and effects. Understanding. 3. You need to be able to imagine hypothetical, fictional versions of this world, which are slightly different than it actually is, in some respect. Possibilities. 4. You need to have goals and desires, so you can decide which of these possibilities you like. Choices. 5. You need to be able to tell a hypothetical story, using this understanding, of how to bring the possibility you have chosen to reality. Plans. This is the set of capabilities that we call "intelligence". And until we have a computer with these capabilities, we won't have a computer that we can experience as "intelligent" in any meaningful sense. And building "better" LLMs doesn't get us there. Not only does the Machine God not lie ahead on that path, neither does an ordinary human being who likes football and daytime television. No matter how glib and eloquent you make them, they'll still be as dumb as a sack of wet hammers. Sure, LLMs are progress. Language processing is something that a true machine intelligence will need in order to understand us, and tell us things. But the ability to speak English does not make a human, being any more than the ability to draw pictures, drive a car, or play chess."

Standards:

Robin Hanson

@robinhanson

·

20. Nov.

Many people really do say that if you cannot find evidence on a topic that meet the usual high standards of nearby academic fields, you should refuse to think about or have opinions on that topic.

College:

Diana S. Fleischman

@sentientist

One reason that women have always gone to college is to find a pool of men filtered for intelligence and conscientiousness. College is an expensive, time intensive and ineffective way to filter for quality men, but, in secular networks, it's one of the only ways. A certain set of men who evaluate women predominantly on attractiveness have trouble understanding how much rides on women's partner choice being adequately discerning- that one wrong step can lead a woman to be tethered to, and making copies of, someone subpar.