All this, and yet, people are so angered by the term "stochastic parrot".
I use LLMs every day, I use Claude, Gemini, they're great. But they are very elaborate autocomplete engines. I'm not really shaking off that impression of them despite daily use .
I am hearing this term for the first time but I love it. It is novel and creates a picture. Exactly what Scott Adams says about labels used for persuasion. I usually say "highly trained autocomplete" in discussions at work, but I am going to say "stochastic parrot" from now on.
It's weird. It's literally what they are. It's a gigantic mathematical function that takes input and assigns probabilities to tokens.
Maybe they can also be smart. I'm skeptical that the current LLM approach can lead to human-level intelligence, but I'm not ruling it out. If it did, then you'd have human-level intelligence in a very elaborate autocomplete. The two things aren't mutually exclusive.
People are hung up on what they “really” are. I think it matters more how the interact with the world. It doesn’t matter if they are really intelligent or not, if they act as if they are.
I listened to the podcast linked in the article, and my understanding of the timeline is:
- The owner originally had two dogs. Both disappeared from her backyard one day. One dog returned home. The other vanished without a trace.
- Eleven years later, a random girl found the missing dog outside. She befriended the dog and brought him home. She talked with her parents and contacted ACCT Philly, who in turn found the original owner through a microchip.
Does this make sense? To me, this story managed to be a rare mix of heartwarming, insightful and frustrating.
Eleven years seems like a very long time to be a Philly street dog - kinda makes you wonder if it wasn't adopted by somebody in the interrim before ending up with the girl somehow.
i cant wait until we can have these cars in the US. looks like I'm going to be pretty geriatric by that time but it's absolutely stupid how many ICE cars are still all over the place at this point
book publishers who want to charge lots of money for exposure to fine new authors is absolutely a great place for a line to be drawn. not much different from concert ticketholders furious that singers were lip syncing some years back.
the whole point of art, music, and literature is that humans love to do these things. The only reason AI would be injected into these topics is for profit-driven purposes, which is a dumb purpose. The job of the robots was always supposed to be, boring, repetitive tasks. Coding, washing dishes, folding laundry. So far we have one of those three checked off.
not speaking for upwards failures in general, but for the extraordinary cases of convicted frauds being pardoned, the incentives are:
1. his $1.8M donation to Trump shows other felons and fraudsters that paying Trump will pay back in dividends (Trump profits)
2. By pardoning thousands of frauds, con artists and outright violent nazis (Jan 6), Trump builds himself an army of loyalists who owe him their lives
3. By putting pardoned frauds, con artists and violent nazis in charge of government functions, Trump replaces the entire US government with one that will do his personal bidding
coding with an LLM works if the model you are following is: you have the role of architect and/or senior developer, and you have the smartest junior programmer in the world working for you. You watch everything it does, check its conclusions, challenge it, call it out on things it didnt get quite right
it's really extremely similar to working with a junior programmer
so in this post, where does this go wrong?
> I am not your average developer. I’ve never worked on large teams and I’ve barely started a project from scratch. The internet is filled with code and ideas, most of it freely available for you to fork and change.
Because this describes a cut-and-paster, not a software architect. Hence the LLM is a gambling machine for someone like this since they lack the wisdom to really know how to do things.
There's of course a huge issue which is that how are we going to get more senior/architect programmers in the pipeline if everyone junior is also doing everything with LLMs now. I can't answer that and this might be the asteroid that wipes out the dinosaurs....but in the meantime, if you DO know how to write from scratch and have some experience managing teams of programmers, the LLMs are super useful.
> it's really extremely similar to working with a junior programmer
Right, which is why LLMs aren't useful if you actually know what you're doing. It's a drain on your time to have to carefully check everything a junior writes, but you do it because he will learn and eventually return on that investment. With an LLM, there is no such long term payoff.
It's nuclear powered autocomplete . You can have it spit out pages of boilerplate chainsaw tests for kunernetes , fix sql queries the way you tell it to, it's enormously helpful for automating all kinds of things you previously have typed by hand 4000 times. I've put many changes from LLMs into production and there is no issue. An actual junior programmer makes way more mistakes in my experience.
oddly enough the TikTok referred to here was to be shut down in the US. But then the executive branch ignored the law while it could organize handing the company over to Larry Ellison instead. But these allegations date to when the company was fully under the control of ByteDance, and not US-regulated entities at all.
> oddly enough the TikTok referred to here was to be shut down in the US. But then the executive branch ignored the law while it could organize handing the company over to Larry Ellison instead
Which should make people think twice when they call for government regulation on speech as a solution to content they don't want other people to see.
The more you give the government power to control speech, the more they'll use those laws to further their own interests.
this is an incompetent, corrupt change that will be reversed when Trump leaves office in 2029. Companies should likely not change their quarterly reporting since it will only be temporary.
I use LLMs every day, I use Claude, Gemini, they're great. But they are very elaborate autocomplete engines. I'm not really shaking off that impression of them despite daily use .
reply