Regulations exist for a reason. Polluting industries like the plastics business would obviously prefer not to be burdened by them. I am not particularly sympathetic to their concerns.
Clojure is Lisp, Guile is Lisp, Racket is Lisp, R and Julia have Lisp under the hood; but the spirit of Lisp also lives on in Javascript, Ruby, Rust, and every functional programming language ever, plus all the languages that incorporated functional features including Python, Perl, and Go, and languages that didn't until they did, like C++ and Java. And Emacs.
I question the coherence of this notion of intelligence as continuum of domain-general ability.
Specific measures of ability tend to cluster at the low end more than they do at the high end. This explains both the correlations among abilities, and the fact that high abilities tend to be domain-specific.
An IQ score over 70 has a clinical meaning: it defines the absence of mental retardation. Scores of 90 and 160 are equivalent for this purpose.
"Giftedness" in a domain-general sense is an abstract concept that may or may not map onto something real. The validity of IQ testing derives from measuring the other end of the scale.
No, stupidity is very general. The statistical association between different areas of ability, which is reified in the concept of IQ, is because people who are bad at one thing are bad at everything. The Stanford-Binet test was developed as a way to measure mental disability, not giftedness.
(Einstein would have had an unexceptional score on an IQ test, had he ever taken one. His schoolteachers thought him destined for failure.)
To put it another way, polymaths are unusual but idiots are everywhere. And people who are outstandingly good at say, computer engineering can be mediocre at philosophy or business administration.
Psychometrists distiguish between crystallized and fluid intelligence. Expertise is a combination of knowledge and ability. But ability itself is multifaceted, and raw talent goes untapped without the motivation to study and the opportunity to work.
For most areas of human endeavor, being smart enough is all that is required. Being a genius helped Albert Einstein reimagine physics, but did not make him a better patent clerk.
Speculation on Trump's use of GLP-1 receptor agonists is beneath the impeccably right-wing Economist. But clearly it would be too little too late to arrest his mental decline. He is too many cheeseburgers in.
The US is high-trust for insiders (rich white people). We allowed Donald Trump to loot the richest and most powerful society in history by imagining that he would follow the example of previous presidents instead of seeing him for the sociopathic con man that he has always been.
Conversely, the US is zero-trust for outsiders such as foreigners, racially disfavored groups, and the poor. Allegedly-dog-eating Haitians and the like. We have guns and are not shy about using them. Being killed by police is a leading cause of death for young men of color, as noted by Ice Cube, and confirmed by researchers at Rutgers (https://doi.org/10.1073/pnas.1821204116).
Agreed, although most of what humans do is pretty pointless in that context. Even if AI turns out to be a multiplier that makes everyone quicker and smarter, people will mostly just want to use it to make more money than the next guy. The history of capitalism suggests that the money-go-round will create useful technology along the way.
I agree with the author's overall conclusion about pricing LLMs, although I question some of the reasoning. He does not talk enough about one important input: the data used to train the models. To a first approximation, their utility is a function of the quantity of data available. Currently, the bulk of the training data is public (the internet) and that is a major reason why the performance of different models is not diverging more. One huge future area for AI is modeling the physical world and in that arena people are going to be creating their own empirical datasets. Waymo has achieved FSD and Tesla has not. It is not coincidental that Waymo has built a sophisticated platform to simulate the physical world.
> Tesla, who are Level 3 at best - curious, because with the ever-increasing rate of AI competence and their massive head start, you'd expect them to have cracked Level 4 by now.
Perhaps (1) Tesla's technology is incapable of level 4 (no LIDAR), or (2) Tesla is not as good at AI as its competitors, or (3) Tesla really truly is a droid company and the car thing was never a priority, or (4) the CEO is on drugs and should not be allowed to wave chainsaws around.