It's astonishing one company could generate this amount of electronic waste.

But, if you have a de facto monopoly on the desktop and the the productivity software (i.e. Office), it's basically like a government decree stating you have to trash those PCs.

It's like the "Cash For Clunkers" program, where the government destroyed 10s of thousands of perfectly usable vehicles (driving up the cost of used vehicles), except there's no cash being given to the affected people, and the computers aren't going to be recycled.

This phrase from the summary caught my eye: "Studies indicate generative AI could automate 30-40% of junior developer and tester tasks." This society better start developing young people and the lower skill tiers if they want that cream that it yields, the cream that drives technology and society. Cream does not automatically appear without regular milk.

If a society decides to prey on its young people instead of developing them, it is setting itself up for decline. The education bubble is a big deal. I remember watching the costs of books and classes skyrocket after the year 2000. The first housing bubble was starting around that time. These are two things that disproportionately harm young people.

The phrase "eating your seed corn" refers to farmers consuming corn seed instead of saving it to plant for next year. That appears to be what's happening over the past 20-some years. Consolidating wealth in the already successful, making it harder for lower skill and financial tiers to build themselves up.

I think it explains the whipsaw politics as well, as people cast votes for more and more nontraditional candidates.

Therefore they did not view Hiroshima and Nagasaki much different than other bombed cities. What really convinced them to surrender was that the USSR had invaded the largely undefended north, and with very rapid progress.

Quite a coincidence then, that the Japanese surrendered four days after the 6 August Nagasaki bombing, if the bombings had little or nothing to do with their decision.

Any public service that can be privatized and profited-from, they seek to privatize. Regulations that hinder profit, are being eliminated. Good for the business-owner class, not so much for the consumer.

"If money is speech, the wealthy have a lot more of it than you."

I think Citizens United was a disastrous decision. Instead of saying only physical persons could contribute to politicians, they went the opposite direction and said any logical construct, union or business, could contribute. Perhaps it was consistent with existing law but it really empowered the donor class to the detriment of the rest of the population. It's creating stresses in society. And both sides cowtow to money and moneyed interests.

In democratic societies, power has a tendency to coalesce to the center of power, to consolidate. A dictatorship is the ultimate form. Feudalism is an intermediate form. If it's not resisted, democracy becomes hollowed out, and we return to de facto oligarchy and feudalism.

Hitler was no spectacular physical specimen. More like a sickly little wimp.

The joke was, "The perfect Aryan man is blond like Hitler, slim like Goering, beautiful like Goebbels."

The big question is, "What exactly did the black box record?"
Was it the physical position of the switches?
Or the electronic status of the fuel run-cutoff state?

I haven't heard a definitive answer. Mostly, I'm seeing the question begged, assuming the physical positions of the switches was what was recorded.

There is a tremendous amount of existing code. Many times I've had to debug or enhance existing code. Reading code is hard. If no one is taught to program, and everyone is relying on code snippets from an LLM, that code is not going to be maintained and will slowly degrade.

Writing new code - you need to know data structures and algorithms for anything more than the highest level scripting. If you feed an LLM all the code on the planet, you still have to understand what it's returning in order to debug it.

If the LLM is going to generate new code, the programmer has to read it (hard - someone else's code and data structures and algorithms) and understand it before deploying.

I don't see how any of this is going to remove the need for programming and programmers. LLM's are amazing, they are natural language databases. They are productivity enhancers, but not replacements for people doing the querying.

If eradicating these flies is acceptable, why not mosquitos? There are certainly ecosystem impacts from removing the flies, the question it seems to me is, "Are they acceptable?" Apparently yes. How did they arrive at these conclusions - just test it and see? Or was the benefit worth whatever the costs to the ecosystem?

Spawning cults and driving people into psychosis is a strong pass of the Turing Test.

That's my big use case, and I think a lot of people's. Also, a good PDF editor. Notepad (I know Notepad++ works on Linux but notepad is as simple and quick as it gets).

If MSOffice could run seamlessly on Linux, that would seriously accelerate Linux desktop adoption.

Also, automatic OS updates would be big.

Semicolons create a harder stop than a comma, to encapsulate a thought; but not as hard a stop as a period, which is a more complete encapsulation of a thought.

Implication? People are expressing less compound thoughts in sentences, they stylistically seek faster flow and harder stops perhaps? Does social media consumption impact how people write and express thoughts? Article doesn't say, but interesting result regardless.

I'm a reasonably proficient touch-typist. I find it helps in software engineering tasks - writing comments in my code and writing informative (but not too long) variable names, to help readability and maintainability. It doesn't help with coming up with algorithms or decomposing larger problems.

But the best programmers I've known, algorithm and intelligence-wise, were hunt-and-peckers. Had they known how to type, their code would have likely been more maintainable and readable. One can be an excellent programmer without knowing how to type, as some of the comments attest to in this thread. However, it makes good code more maintainable due to the more verbose comments and superior variable names.

Existing code is hard to read. That's why even good programmers want to clear the board and start from scratch, rather than understand the existing code and optimizing it. Better comments and better variable names make it easier to read.

[Note: using &lt to represent less than operator because nothing lets me show the less than sign which is the start of an HTML tag]

// the hardware starts getting flaky beyond 10 connections
int maxConnections = 10;
for(int i = 0; i &lt maxConnections; i++){ ...

versus

for(int i = 0; i &lt 10; i++){...

The top tells you exactly what's going on and is more likely to be written by a touch-typist; the bottom more likely to be written by someone who cannot.

I think the ability to type is an advantage to the programmer and a benefit to the organization.
There are mediocre programmer who can touch type and excellent ones who cannot.

In order to understand LLM's, one needs, it seems to me, the appropriate amazement that the inventors of this technology were able to structure and store the information itself in such a way that stimulating it with prompts yields coherently structured information back.

It requires huge amounts of data and energy to ingest giant curated datasets and structure them in the way required to yield coherent information when queried/prompted/stimulated.

AFTER the data is structured and stored, perhaps it is a mere stochastic parrot (a term coined by Dr. Emily Bender in her well-known paper about bias in LLMs, whether they use too much energy and whether there is benefit to larger models - discussion of it here and here). At one point in the video discussion panel, she vehemently opposed that humans might be doing the same sort of thing that LLMs do because it's "dehumanizing". She goes on to say, "I will not engage in discussions with people who don't acknowledge my humanity," which seemed very... aggressively advocacy focused. I mean, we don't know how we digest food in that we are not consciously directing it, and we don't know how we retrieve information, we "just do it."

My point here is that this is a new tool, and no one knows everything it can do. Knowing it's a next token predictor is the most basic level of understanding of these software constructs, and can lead people to underestimate the tasks it can accomplish with tuning and optimization.