Somehow I'll miss you Google. I'll miss the simpler times. You were a fine search engine and gmail was pretty great too, at the time. I was really excited to get an invite. Good memories.
We'll always remember you. Well.. not always, but for some time to come. Rest in peace.
I agreee, its sad, the old google is gone and will also never return. But the decline began already a decade ago.
It became an ad-driven corporate behemoth, so any innovation coming out of it will be related to these factors.
Confining an LLM to the very narrow domain of "calculators" is a mistake, I think.
You wouldn't say "a programmer that is 99% correct is worthless, I need 100%". I'm pushing it, but for a more fair comparison I'd say measure it against a programmer. How often are we wrong? 75% of the time? :) being generous here. It's the tools that make us productive.
I don't know about you specifically, but I don't think you'll be very productive with a bare terminal lacking any modern IDE-like or even REPL facilities. I'll ask you to come up with instantly working code every time, all the time. It doesn't work like that. You need iteration and I believe these kinds of AI have the same issues as us. There are wrong sometimes (often) and need feedback.
> You need iteration and I believe these kinds of AI have the same issues as us.
It's funny how we resort to humanizing the machines when their results are inaccurate. We don't do that with the calculator, because it's expected to be 100% bug free. When there's a bug in the calculator code we expect it to be fixed, not gradually improved.
Speaking of bugs: mistakes in code is one thing, wrong output because of a fundamental flaw in the algorithm is another. The statistical machines we are dealing with work as intended, or at least the wrong output the top comment here brings up is not a bug, it's a feature. That's the difference.
Literally LLMs get much better with chain of thought, feedback, and/or consensus.
Gpt-3 performance on MultiArith goes from 18% to 92% with all three. This isn't some hackneyed anthropomizing. Countless research papers showing massive improvement with these processes.
That's (IMO) too narrow view of what a "machine" is. Complex machinery of any kind never is 100% correct and needs constant correction and maintenance. I still think approaching this as a "calculator" is awkward at best.
> Complex machinery of any kind never is 100% correct and needs constant correction and maintenance
Computers are extremely close to 100%, we generally expect a CPU to never make errors even after years of working. If it starts making any errors at all we throw it away and make a new one.
Do you code in checks to check the calculations made by the CPU? I've never ever seen anyone do that. If a CPU starts making errors we throw it away. A typical CPU will make many quadrillions of correct calculations before its first error, I'd say that is basically 0 errors.
You think music theory is more demanding than CS? I've dedicated decades and probably 75% of my youth to mastering this instrument called a computing device. It has numerous layers, each completely different and each significant enough to build a standalone career out of (OS, networking, etc). I feel insulted if you think playing and mastering a piano is the same thing.
Extreme specialists are found everywhere. Mastering skateboarding at world level will eat your life too, but it's not "harder" than programming. At least, for any commonsensical interpretation of "harder".
All the rest, we do too. Except I don't record videos and I'm sure it is not childishly easy, but it will not eat my life.
I'm literally speechless. What an arrogant and egotistical comment. This is why us tech workers have a such a bad rep as culturally ignorant/bubbled community. Do a bit of research into jazz theory and counterpoint theory before you make this kind of blatant over generalization.
This exact comment could be made by a jazz soloist with a few words changed and be just as valid. I think you're underestimating how deep other fields, including artistic fields, are. Anything as competitive as an artistic field will always result in amounts of mastery needed at the top level that are barely noticeable to outside observers.
> This exact comment could be made by a jazz soloist with a few words changed and be just as valid.
It's not that uncommon for professional programmers to be pro-level musical soloists on the side, or for retired programmers to play top-level music. The reverse is far less common. I do think that says something.
> Anything as competitive as an artistic field will always result in amounts of mastery needed at the top level that are barely noticeable to outside observers.
Sure. Top-level artistic fields are well into the diminishing returns level, whereas programming is still at the level where even a lot of professional programmers are not just bad, but obviously bad in a way that even non-programmers can understand.
Even in the easiest fields, you can always find something to compete on (e.g. the existence of serious competitive rubik's cube doesn't mean solving a rubik's cube is hard). A difficult field is one where the difference between the top and the middle is obvious to an outsider.
I think today he/she learned an important lesson for his/her career: there are things more difficult than the epitome, the apogee, the quintessence of professions, called computer science.
I am a classical clarinet player, a Physicist and a programmer. Music theory is ridicously easy in confront to everything STEM I've studied. It isn't even a paragon: the counterintuitiveness of Physics, the abstractions, the rigor of thinking needed was really stressfull to my mind while music theory was... underwhelming to say the least.
I couldn‘t, but I could also not study many other things and not because of what you call difficulty. Quite simply different people are good at some things and less good at others.
Maybe you are better at CS than music and therefore perceive it as easy and the other one as hard.
Again, it depends on the level. Maybe you took trivial CS courses. Many parts of CS are indistinguishable from mathematics, is that so easy as well? What about the various open problems that have remained unsolved for decades now in theoretical CS? You think these are simpler than music? Really?
Come on, at which level did you study them? I studied both at University level and was a classical clarinet player and anything STEM was much more difficult than anything music theory.
Making art is not "vastly more difficult" or at least it is (IMO) highly debatable. Some parts of it require decades of experience to do with any kind of excellence, yes. That's also the case with powerlifting, figure skating and raising children and indeed programming. It's just that your boss made a money printer that takes in bullshit and outputs bullshit which gives you your cosy job.
But that is not "programming". That is glueing together bullshit until it works and the results of that "work" are "blessing" us everyday. The gift that keeps on giving. You FAANG people are indeed astronomically, immorally, overpaid and actively harm the world.
But, luckily, the world has more layers than that. Programming for Facebook is not the same as programming for a small chemical startup or programming in any resource-restricted environment where you can't just spin up 1000 AWS instances at your leisure and you actually have to know what you're doing with the metal.
I don't know the first thing about you, so it's hard to say what is going to help. If you truly are a narcissist all bets are off, but perhaps you are just a bit self-centered which might mean there is still hope after all.
I'd look for books on being a "building lasting relations", "how to deal with emotions", "how to be a good mother/father". That kind of stuff. Bring out the wounded inner child and see if some CPR is still an option.
It does not suck. The compile times are OK and are getting better. Angular (and other) SPAs have capabilities no RoR, ASP.NET or Django setup can even begin to touch. Separating the back-end from the front-end with an independent API is also a god-send.
Death to remaining in the dark ages for nostalgia's sake.
Oh, it does suck, hugely. Longer time to first paint, more bandwidth, more cpu and battery usage, huge dependency chains, slow "live" updates...
A lot of my issue with the Javascript-heavy present of "modern" SPA websites is that they have to recreate native browser functionality, which is inevitably slower and worse.
Way too many websites could have the same functionality, in a simpler form, with better performance with a more "traditional" stack. You can even have your API separated front-end and back-end via server-side rendering (by which I mean old-style server-side rendering).
The modern zeitgeist is mostly hype, fad, and novelty.
> The compile times are OK and are getting better.
Our frontend Angular test runs take 25 minutes to run less than 2,000 unit (not integration) tests. That's just nonsense. I don't blame Angular so much as I blame the JS ecosystem. For comparison our backend unit test runs (ASP.NET Core) take 3 minutes to run 6,000 tests.
You can argue we're doing something wrong, but as a company with a a thousand or so devs - reaching out to the most experienced frontend devs we have, their answer is: you've got too many tests, split up your SPA into smaller SPAs. Great, more complexity and dependency management hell.
I find it interesting that somebody dedicates his or her life to figuring out how these CPUs work when exactly that information is just laying about in some vault in Santa Clara.
We'll always remember you. Well.. not always, but for some time to come. Rest in peace.