I find this version unlikely, since companies very rarely genuinely believe what they preach in PR campaigns. It's always some sales and marketting dudes and gals trying to polish up something as something more than it is. Which is very annoying. We can now choose between Anthropic being the one exception to this, while having huuuuge incentive to hype up their product, or we just write it off as more marketting fluff.
I would be very surprised if this is an actual thought-out PR strategy. I am far more inclined to believe that their employees are just bought-in to the future where AI is genuinely transformative.
Whether they are right of wrong is another matter, but their claims also don’t seem too far out of the realm of possibility to me.
Coding agents have fundamentally changed my day-to-day job. In the last year, my work has shifted from me writing all of my code, to me writing very little code and spending most of my time on understanding problems better and setting direction, and reviewing, verifying, and polishing the output of coding agents. It has been quite a drastic change.
It is not that outlandish to suggest that coding agents could continue to improve at such a drastic rate over the next year. And the implications of that could be quite large! Even just the implications of more white-collar workers adopting tools like Cowork seems potentially very large, with tools that exist today. It seems sensible to at least consider this as a possibility.
Dario is no John hammond though. That'd be altman. He actually has the discipline and background as an ai scientist to tell what the potential failure modes are. You're right, he might still be just hyping things up, but generally i'd give more benefit of doubts to anthropic. Precisely because Dario was a scientist and I'd stand by it. People who get their phd in science already self-select, or proven at least to be made of different stuff.
Likewise, people don't as easily blame ilya for 'hyping things up' when he said these things.
Also talk about incentives, there are also incentives to lower their valuation. If you wanna be vigilant against social engineering i'd be wary of that too.
These are moot anyway though cause the article isnt even making any super strong claim. If you read it it's no big deal
It doesn't convert bogosort into heapsort either, despite the second being much faster than the first. I'm guessing that it's not that easy going from one to the other because the only thing they have in common is the output (and only after you have checked the last value), so if the transformation is not hard-coded into the compiler, the odds of it randomly discovering the optimization is close to zero
Yeah, I would expect such transformations to be implemented as optimizations. Just like maybe (the admitedly simpler):
(+ ((lambda () 1)) ((lambda () 1))) -> (+ 1 1)
A syntactical transformation, where it is possible as an equivalent transformation.
I may be overlooking special cases, but I thought the compiler is smart enough to infer that the array elements are integers and that `<` will result in a boolean, which is just `0` and `1` and will understand that having only the `if` without `else` branch is equivalent in this case. Guess I was wrong and the compiler is not sophisticated in this specific way.
The two code snippets do different things, apples and oranges... e.g. the array modification in the second example needs to move in front of the if. I bet then the compiler output is the same with -O1 or higher.
PS: e.g. note how bla() (first code snippet) and blob() (fixed second code snippet) have identical output, but the blub() function (original second code snippet) differs:
TL;DR: most 'branchless advice' that only tinkers with language features (like "x = a ? b : c") is useless because to the optimizer both '?' and 'if' are equivalent.
When there's a difference in the generated code then it's usually a bug and the before-after code are not actually equivalent (like in the code examples above).
Is it still chess, if humans cannot understand it? Because that's the point we are at in chess. Engines making moves, that humans cannot understand, but somehow they work out to be best or seemingly best. Look at the Leela Zero games, when it came out. These engines play kind of other-worldly chess.
Exactly. If we wanted "the best chess" we'd watch Stockfish against Leela Zero. Far better mechanically than human chess. But people are much more interested in Magnus Carlsen playing Gukesh. Both train with chess engines, but the thing that makes the chess game interesting are the human beings that understand their own moves and try to understand those of their opponent
Out of those renames, I agree with car->first and progn->do. setq is ugly, but I think using def is maybe questionable. lambda I would have just kept the same.
PEGs are just soooo much easier to read than regexes for anything more complex than a few words or single line matching. REs are a hammer that tempts people to see everything as a nail, but once one progresses beyond that phase one usually wants as few REs as possible.
That’s mostly an artifact of concise regex syntax I believe. When you write regular expressions in an ABNF-like form, they become much more intelligible.
Probably depends on whom you are asking. For me the essence is (1) having functions or procedures as the basic building blocks, not classes. (2) Having all the utility and higher order functions you need to deal with the functions and procedures first idea. (3) Having a very powerful syntax, that allows great semantic editing and is never ambiguous. Oh and can actually be extended in useful ways, without having to wait for a committee to decide upon "the one syntax to rule them all".
The critical point is collaboration on a document. I wish LO had this. You know, like other collaborative editors online, where you can see the other person's cursor and inputs live and all that, with a conflict free data structure driving it all. LO needs this, if it ever wants to replace MS Word in businesses. Even if it would be totally fine for most situations to edit sequentially, there are those cases, when multiple cooks are trying to change a company document, and now they can't, and need to be in a call, to organize their editing or share their writing ideas, because the tool doesn't allow them to edit at the same time.
It does not even have to be in a web UI or browser. Just somehow make it possible to easily connect and edit collaboratively. I know, I know, it is a huge ask, unfortunately.
>It does not even have to be in a web UI or browser
LibreOffice Writer online was not popular, so it is understandable you assume it doesn't exist. There is also the trivial headless Linux remote desktop cloud hosting with 100% identical functionality.
>"edit at the same time"
Perhaps you meant: "View -> Toolbars -> Track Changes..."
Most documents that support OLE have difficulty handling concurrent writes. Office 365 abandoned desktop publishing in 2019, and replaced it with a web document management system.
Best workaround is every user imports each instanced changes: "File -> Merge Document."
And manually handle any collisions.
I would post how to setup your own remote collaborative environment on an $8/month host, but people seem like they are not interested. =3
I have recently tried to use it, but couldn't find a way to open files from local computer in it. It showed me a fake filesystem and dragging and dropping files also didn't work.
I have not used another web browser LO thing, I think. If you are talking about another thing, do you have demo site/link?
> Perhaps you meant: "View -> Toolbars -> Track Changes..."
No, I mean a mode or so that is like online word processors like Google Docs, where you see the cursor of someone else, with their name, and can see them typing live, while you are also typing somewhere in the document. Unless "Track Changes" somehow enables connecting to someone else and seeing their actions concurrently, it is not what I am talking about. I really mean collaboratively editing a document, at the same time, seeing changes others make live, not merging afterwards (the git model).
reply