BDSM, LGBTQ+, and sugar dating apps have been found exposing users’ private images, with some of them even leaking photos shared in private messages.
BDSM, LGBTQ+, and sugar dating apps have been found exposing users’ private images, with some of them even leaking photos shared in private messages.
Vibe coding is the current trend of having an LLM build your codebase for you then shipping it without attempting to understand the codebase.
Most developers are using LLMS to some extent to speed up their coding, as cursor and Claude are really good at removing toil. But vibe coders have the LLM build the entire thing and don’t even know how it works.
In other words, vibe coders are today’s technologically accelerated script kiddie.
That’s arguably worse as the produced scripts may largely work and come with even less demand for understanding than a script kid’s cobbling together of code may have demanded.
100% accurate.
Thanks. What is an LLM?
Basically, think ChatGPT
Large language models (LLM) are the product of neural networks, a relatively recent innovation in the field of computer intelligence.
Since these systems are surprisingly adept at producing natural sounding language, and is good at create answers that sound correct (and sometimes actually happen to be) marketers have seized on this as an innovation, called it AI (a term with a complicated history), and have started slapping it onto every product.
Neural networks have been around for quite some time. The simplest forms of it have actually existed since around 1795.
Ahhhhhh… that’s a really simple explanation thanks
A machine learning model that can generate text.
It works by converting pieces of text to “tokens” which are mapped to numbers in a way that reflects their association with other pieces of text. The model is fed input tokens and predicts tokens based on that, which are then converted to text.
Large Language Model
To the extent of my understanding, it is a form of slightly more sophisticated bot, as in an automated response algorithm, that is developed over a set of data, in order to have it “understand” the mechanics that make such set cohesive to us humans.
With such background, it is supposed to produce new similar outputs if given new raw data sets to run through the mechanics it acquired during development.
Clever, thanks 😊
What is toil in this context?
Boring/repetitive work. For example, I regularly use an AI coding assistant to block our basic loop templates with variables filled in, or have it quickly finish the multiple case statements or assigning values to an object with a bunch of properties.
In little things like that, it’s great. But once you get past a medium sized function, it goes off the rails. I’ve had it make up parameters in stock library functions based on what I asked it for.