2,887 karma · joined October 23, 2019
But everything else would be so much simpler and better for most of the things I do if it could just be a normal box with normal software.
In theory I could build that experience with nothing but bare EC2 instances. But once you’re in AWS, they make using their managed things so tempting that convincing everyone else to not use them is an even bigger battle.
Look, I won't deny that like the dot-com bubble and the "cloud" hype of years before that there isn't overhyping and forcing going on. But this feels to me a lot like someone in the mid 90's looking at this "internet" thing and thinking it's all a fad being pushed on us by computer nerds. I'm not saying every company and every attempt to stuff "AI" into everything is going to be some revolution that will stick with us forever, but neither were a lot of early online providers. But like I said, AI is infrastructure, and people are going to figure out how to build things on top of it that will become integrated and standard parts of our lives, just like computers, the internet and the cloud are all today.
I think to some degree this is going to be inevitable for a while. No matter what people think about the ethics of AI or AI use, the fact of the matter is it exists, and it is shaking up and going to shake up a lot of the world. I personally think "AI" isn't really a product anymore than "The Internet" is a product. It's infrastructure in the same way the internet is, or to quote "Halt and Catch Fire", it's "the thing that gets us to The Thing".
The problem is no one really knows what The Thing looks like, so everyone is throwing anything they can think of at the wall and hoping to see what sticks. And they're dong this because they remember Sears and they remember Blockbuster. No one wants to be the next company that was positioned perfectly to integrate the new technology into their business model and missed the boat completely for short sightedness or lack of speed. "$foo but with AI" is the current generation's "$foo but on the internet", and everyone is scrambling hard to not be the next company that someone writes a "whatever happened to them" youtube video about.
Incidentally, this "AI as infrastructure" view I have is also why I really think governments should be getting into the space of building and distributing training data sets and models. Not only because it could reduce the overall amount of bot scraping and traffic if everyone who wants to build could go to a known good and clean source of training data, but also because nominal usage fees could help manage the copyright concerns. I've pitched it before in more detail, but I imagine something akin to radio broadcast royalty licenses, where people whose copyrighted works are part of the government training sets are paid out "royalties" from the pool of money brought in from licensing the training sets.
Why would LLM output be any different?
AI is the new "excel spreadsheet with ridiculous formulas on top of VB script on top of pivot tables" for the non-programmers. Yes, the AI might not do it efficiently or even always give the right answers, but the spreadsheets didn't either. For non-programmers, the computer has always been a black box, data goes in, data comes out. Usually it's right, sometimes it's wrong and the users mostly had no idea why or how. But now they don't need the middleware app. They don't need to take their data and split it up into very specially structured files, and put just the right pieces of information on just the right lines and in the right spaces. They don't need to remember to do "(foo and (bar or baz))" to get the right answer, as opposed to "(foo and bar or baz)" and get the wrong one. AI for non-programmers is the closest we've ever come to "natural language programming" for people to use. It doesn't matter that it gets things wrong sometimes, or does it inefficiently or makes something unmaintainable unless you use the AI again. What matters to them is they can say "take all this data here and swizzle the frobs" in their native language instead of learning a specific programming syntax and system. And they can point it at some arbitrary data in whatever format they happen to have the data in, and they have a halfway decent chance of getting something that does what they want.
It's not going to be pretty, it's not going to be perfect and it's probably going to break in subtle and unexpected ways. But as long as most of the time its mostly good enough, people will still use it enthusiastically. We're effectively looking at "worse is better" for the modern age and for people that aren't and never were going to be programmers.
I think there's also a lot of recency bias in it. The same thing that makes you suddenly notice how many people are driving the same model car you just looked at, or how many ads there are for Turbo Encabulators after you read an article about them. A lot of the "tells" people picked out in early AI are tells because they're also really common in the material that the AI was trained on and the styles it was made to emulate. But until everyone wanted "one quick trick" to pick out AI writings, people didn't have any particular reason to need to notice those tells and so they slipped under the radar.
But none of those sound like things that would change how likely any given project is to succeed. Shrinking timelines from years and months to weeks and days just means that you can fail faster and try again sooner (hopefully with lest costs), but nothing about that implies your rate of success is going to get better.
I'm not sure I follow why you would think this. Prior to AI was the reason most projects failed because generating the code was too slow or too labor intensive? That would be surprising to me if it was true.
Likewise for "Building and grounds cleaning and maintenance", we went from $9.41/hour to $18.12 / hour.
Though admittedly not every field saw such gains. "Sales and related" didn't do quite as well, going from $13.46 to just $18.52. "Office and Administrative" did a bit better going from $12.64 to $22.81, but that's still a bit under the 2x mark. "Construction and extraction" also missed the 2x mark by a bit, going from $16.56 to $28.63.
Still the numbers I think aren't quite as dire as the vibes make it feel right now.
[1]: https://www.bls.gov/oes/bulletin_2000.pdf [2]: https://www.bls.gov/news.release/pdf/ocwage.pdf
The initial explanation of the happy path might be slightly longer to teach drag and drop or right click copy and paste for files, but that happy path will also deal with a lot more scenarios in the exact same way than the happy path cli commands will.
As an industry we've been promising people for decades that if they put all their data into our special softwares they can get all sorts of information back out that will make life easier for them, reveal new insights and otherwise improve their understanding. But the unspoken caveat has always been that you have to put the right data into the right places, in the right format, in the right way and then you have to ask the right questions, in the right syntax, with the right tools. And if you get any one of those parts wrong, you're not going to get the right answers (or possibly even any answer at all). How many people have had their excel worksheet that they (or someone else they asked/employed) built for some task that has been working fine for the last year suddenly stop working or start throwing out nonsense numbers because some input changed? Or how many people have experienced their system seemingly throw out meaningless garbage because daylight savings changed right at the moment the report was being run? Or spent months operating on wrong data because the person who wrote the query misplaced a parenthesis and the query was searching for "(foo AND bar) OR baz" and not "foo AND (bar OR baz)". For most people, the computer and the programs they use to do their jobs are magical black boxes that most of the time produce mostly the right answers and sometimes get things very very wrong with no indication of what has changed. Which is effectively the same experience they will have with an AI, but now instead of needing to figure out some arcane excel pivot table and VBA script, they can just dump some raw data and a "natural language" question into the AI.
And that's not counting the fact that their experience with looking information up online is about the same as well. How many absolutely confident wrong takes have you encountered online for things you're an expert in? How many of those wrong takes have come straight from supposedly trustworthy sources like news companies or even other people in the field?
For most people, using a computer has always come with the asterisk that you should always be aware that the source you're reading could be very wrong, that the output is only correct assuming all the inputs and all the parts processing that input are also correct and that everything you do should be accompanied by vetting by experts, whether those experts were software developers or domain experts. For most people the only thing that's changed with AI is that it's a one stop shop for their "probably directionally right, almost certainly wrong in the details" access to the digital oracles.
Now you might argue that “vacations” aren’t “life changing”, but I would certainly argue that if you never would have had the experience or seen the place then they absolutely can be. But even if not, I refer you back to the original thesis which is that “life changing” is relative. Because the sums of money we’re talking about would have been “pay my rent for a year”, “buy a reliable (used) car”, “reduce my student loan balance by nearly a third” sort of money. And those I think could all be reasonably said to be life changing sorts of things.
Finally I would suggest that if you are “throwing away” this sort of money on actually changing someone’s life, then you are by definition not “hoarding money” and can hardly be said to be poisoning society with your relative wealth.
A) which premade config to choose
B) what half of those settings even do (and do you need them)
C) a lack of “sane” defaults that use the built in abilities
Realistically walking through the article authors “emacs from scratch” config and seeing what can be done with the built in packages and how is hugely instructive but even then you only get that from walking through the config one line at a time and reading “help” documentation for most of it.
At this point emacs is so old that fixing the problem of “sane defaults” is probably near impossible if only for how much it would break existing configs. But it might be a good addition to the tutorial to provide a set of questions and answers (possibly with demonstrations) that allow a new user to generate their own config with some nice defaults. We can assume these days that most new emacs users are coming from some other editors, so asking questions like “do you want auto complete suggestions in an inline drop down like VSCode” could be a perfectly reasonable question and then it could add the correct configuration settings to config for you using just the built in functionality
For my desktops, I use an Ergodox EZ keyboard and mapped ctrl and meta into the thumb clusters there.
Start laying hazards in the middle of the road and see how quickly the police introduce you to things like “reckless endangerment” and “involuntary manslaughter”. The general social contract is that you don’t take actions with the intent of causing harm to others regardless of whether the victim could have avoided the harm had they taken different actions.
Not to be overly negative, but have you really looked at the vast majority of the content on the internet? There are good pockets of real, in depth content. But the absolute vast majority of it is surface level basics at best, and completely wrong hot takes at worst. Content farms and click spam have made up huge portions of the internet for a while, never mind the absolute hell holes that places like Facebook, Twitter and Tumblr were and have been. And that's before you consider how often news media gets stuff wrong and then everyone copies everyone else's homework. Knowledge propagation, and more specifically correct knowledge propagation has always been difficult, slow and rare. You have always needed to check primary sources, and AI is just the latest in a long line of reminders of that fact.
[1]: https://en.wikipedia.org/wiki/Mac_transition_to_Intel_proces...
In my opinion I consider a friendship any relationship where no matter how long ago it ended or how long ago you last talked you wouldn't mind hearing from them again, even if it might only be awkward small talk. Old schoolmates, college roommates, military squadmates, and co-workers can all be friends. They can all be acquaintances too. But crucially the fact that you stopped talking at one point or stopped spending time together isn't the demarcating factor between the two.
It's true for some of us introverted people as well, especially given that without some "reason" to get together, some of us might never interact with another person ever.
You're getting paid to be friends with your co-workers? Or are you being paid to work, and work, like many other situations where multiple people gather and share experiences and spend time together are also places that people tend to form friendships in. You had friends in school that you stopped maintaining the friendship when you stopped attending school together I'm sure. Were those people not actually your friends? How long does a "social interaction" have to last, and over what distances before it becomes a "friendship" instead of a "transactional relationship"? If it ever ends was it never a real friendship? It's certainly possible to view every relationship you build with people that you share circumstances with as transactional relationships, but that to me seems like a good way to never actually build a friendship with anyone.
This is reductive to the point of absurdity. Situational friends are still friends. How many of your elementary school friends are still your friends these days? High school? Summer camp? Heck college friends? Unless you're living in the same town with the same people, there's a good chance that most of them aren't anymore. Were these people also not your friends? When you leave that book club, when you stop showing up at the corner cafe, when you move out of the neighborhood, how many of those people will you still be spending time with 5 years later. For the ones that you aren't, were they also not really friends?
Friendship isn't a binary thing. Not every friend you make will help you bury a body, but not every friend or friendship needs to (or should) run that deep. And sure not everyone you're "friendly" with at work are friends, it's a spectrum. But situational friends are friends. People you bond with for a short while over a shared experience and then when life moves one or both of you on the friendship ends are still friends.
Are you suggesting that “urban hubs” and “immigration tension” are code words for “black people” and “slavery”? Because I regret to inform you that when New York City established the first US police department in 1845 (per britanica) the “immigration tension” at the time would have been the influx of Irish immigrants. And while Cincinnati had indeed had a white on black race riot in 1841, when it established its own police department in 1852 the anti-catholic / anti-German immigrant riots in 1853 and 1855 were the more contemporary “immigration tensions” I was referring to. Boston too when it founded its police department in 1854 was in the middle of a surge of Irish immigrants. Certainly these northern state city centers weren’t simply giving uniforms and badges to “slave patrols” when they founded their police forces, regardless of what other racial tensions may or may not have played a hand in the demands for a police force.
All of which is to say if you recall your American history, we have a long and storied tradition of hating on our immigrant populations and having conflicts with them. Yes white vs black was a problem at the time. And so was “white vs Irish” and “white vs German”. Our history is littered with racial tensions across just about every set of ethnic lines you could care to draw.
Edit:
I note now that my britanica link in my first post was the wrong one, this would be the more appropriate for the topic at hand: https://www.britannica.com/topic/police/Early-police-in-the-...