This article on AI focused on the inherent untrustworthiness of LLMs, and attempts to break down where LLM untrustworthiness comes from. Stay tuned for a follow-up article about AI that focuses on data-scraping and the theory of labor. It’ll examine what makes many forms of generative AI ethically problematic, and the constraints employed by more ethical forms.
Excerpt:
I don’t find the mere existence of LLM dishonesty to be worth blogging about; it’s already well-established. Let’s instead explore one of the inescapable roots of this dishonesty: LLMs exacerbate biases already present in their training data and fail to distinguish between unrelated concepts, creating lucid lies.
A lucid lie is a lie that, unlike a hallucination, can be traced directly to content in training data uncritically absorbed by a large language model. MDN’s AI Help is the perfect example.
This article on AI focused on the inherent untrustworthiness of LLMs, and attempts to break down where LLM untrustworthiness comes from. Stay tuned for a follow-up article about AI that focuses on data-scraping and the theory of labor. It’ll examine what makes many forms of generative AI ethically problematic, and the constraints employed by more ethical forms.
Excerpt:
I don’t find the mere existence of LLM dishonesty to be worth blogging about; it’s already well-established. Let’s instead explore one of the inescapable roots of this dishonesty: LLMs exacerbate biases already present in their training data and fail to distinguish between unrelated concepts, creating lucid lies.
A lucid lie is a lie that, unlike a hallucination, can be traced directly to content in training data uncritically absorbed by a large language model. MDN’s AI Help is the perfect example.
#Mozilla has announced the new #MDN#Observatory would be released on January 31, 2024. But now all I get is a "HTTP Observatory is down" error. Has the web address for the tool also changed?
Appel aux (premières) contributions sur #MDN :) Si vous avez quelques minutes devant vous et que vous souhaitez traduire/mettre à jour des petites pages sur MDN, vous pouvez aider sur le Glossaire : https://github.com/mdn/translated-content/issues/16598
Some people criticize #tailwind/#tailwindcss, but I still think that it's great and even a good way to learn #css. It solves one fundamental problem of CSS (for me):
You can't go all-in with inline. What about transitions, states, etc?
The docs are also great and help you understand CSS. The docs may even be better than #mdn. Maybe not as comprehensive, but just as clear with good examples.
How do you keep up-to-date with new #html releases? I've done some research but I couldn't find a website where I would get a notification once a new release is out.
Should I just sign up to the #mdn#newsletter? Usually I notice because the awesome developers in my bubble write about it. Any help and boosts appreciated! 🙌
the #MDN#AI debacle continues. Apparently correctness is not what they are aiming for, and cannot be measured(!) It's really hard to know how to even engage with this level of thinking
It's very disappointing to see Mozilla decide that it's fine for MDN to use an LLM to provide inaccurate and misleading information to its users — because apparently the only thing that matters is that it looks helpful, not whether it's actually reliable — and refuse to listen to the many people in the community telling them it's harmful and they don't want this.
ChatGPT is a chatbot. A very impressive one, yes. But it cannot deliver reliable information no matter how much you want it to or how helpful it may appear to be. Because it's not a machine for doing that. It's a "'say something that sounds like an answer' machine". Sadly, it seems Mozilla doesn't care.
I "love" how Mozilla is busy renaming issues to hide the problem, ignoring criticism, and writing a post-mortem which ignores the problem.
I wrote a comment in the original issue which has since been closed, but this new one is just being ignored at this point, so I don't see the point in doing so again.
The enshitification of the internet continues, and it was silly of me to hope that MDN and Mozilla would be above it.
One thing about the #mdn controversy is that contrary to what too online people say, credibility is really hard to lose.
Ray Kurzweil hasn’t been right for at least 25 years, Marc Andreessen is still a genius investor and every “disruptive plan” of Elon Musk is still being taken at face value by media.
It’s literally impossible to lose credibility when you’re famous enough.