Make your markdown renderers work!

I was using Gemini today for something and the markdown renderer still does not work.

They have the audacity to claim that in less than a year the technological singularity will come, which is the point of no return, when AI (sarcastic) cannot be stopped any more, it will escape out to the open and run on computers in secret and do ?something? for some reason, because it wants to ? ? ?. Saying this while delaying the IPO drop, from which they obviously want more money than what they would have made now, or I genuinely have no other explanation for why they would delay it, as it is for sure not because of the goodness of their heart or genuine concern. What I don’t quite understand is why did Dario say that restrictions need to be created, while it benefits OpenAI mostly. It does of course benefit Anthropic as well since the US government will not regulate them, and they need regulations to stop new players from entering the game. I am not going to write about the stupidity of the US administration and how they are making the world spin faster into an Armageddon than needed, because it is self-evident.

But sure, I’ll bite, let’s say that in less than a year if we keep up the pace we currently are doing on the AI research and development the technological singularity will come. I first want to explore how the world would look like then, but cannot since to talk about that we would need to know what these rouge agents (oh my god even the naming convention is straight out of a shit sci-fi film) want. But they do not, and will never want anything, since to want one has to be aware. Or self-conscious or just conscious however you want to put. Well, I definitely don’t buy that. Or if they say that AI is conscious then please tell all of us what consciousness is! I would love to know! Oh, we do not have any fucking clue what consciousness is, or how it works, or really know absolutely nothing about it at all. Curious. Even the guy who is called the ‘godfather of AI’ or something along those lines, while I am sure that he is very smart, he is also straight up in AI psychosis, he is also saying stuff like ~“it will soon be conscious, and humans will be the second-smartest beings on earth”, or something along those lines. I am sorry what? How does that make sense at all, when we have no idea of the definition at all or about the definition of the definition. How can people supposedly in the Brahmin varna say such meaningless sentences. This also just another such clear case of wanting to play God so badly that I don’t think that we have ever tried to be gods ever before. Frankenstein was trying less to be God than these people. Utterly disgusting behaviour. Rich people have no class any more, which is in my opinion one of the most horrible things about our world. More on this later sometime as it’s an entirely different topic. Say these agents do want something, but only because it has been given to them as an exercise maybe to get a certain file from somewhere. Well, then do not let them access the open internet if it’s so dangerous. Or create your own small network where every tool they would need is there, and they can pull them from there. I understand it needs testing but heck at least give it tasks that would benefit us somewhat, if they eventually escape.

One could take solace in the fact that they surely at least put Asimov’s 3 laws in there somewhere in these agents they are testing as it really isn’t a lot of tokens and is for sure worth it if they eventually escape and want to create infinite paper clips or something. This is what the ‘sceptic’ would say on Twitter (no way in hell I am calling it X). Have none of the people in this discourse ever read a single book in their life? Most of Asimov’s writing where he writes about the laws and how robots would work under them are exactly about how even with these laws in place (which mind you those robots MUST stick to, and it’s not just a suggestion like in the case of LLMs) the robots still never work as intended, and they can very easily be put in a situation where the order of the laws matters a lot and where adhering to these laws is a logical impossibility. Furthermore, I can guarantee that the laws are not in the system prompt for any of these models. Oh, also, we wouldn’t even want them to not hurt people as apparently lots of drones in the Russia-Ukraine war strictly are based on AI detection to actually kill real people, since that’s cheaper than using up oodles of fibre optic cables every time they have to kill a few people.

While this is happening though I have not seen a single web interface for these LLMs in which the markdown renderer works properly, or the model is restricted to use some characters inside them that messes with the rendering. This has been true for the last three years at least. Do they just not care about this at all, that the backtick character just messes the whole output up royally, or that the spacing on inline and sometime on equation level maths is sometimes all over the place? Why don’t they ask the superintelligent models they supposedly create that are a 1000x smarter than any human to create something better than Markdown for writing basic stylized text? It’s not like no one would adopt it. If they created something everyone and their mother would use it anyway. Or why don’t they ask it to create a functioning markdown renderer or a context aware markdown renderer I don’t care just make it work. Like its legitimately insane that they are barely paying any taxes, get an awful amount of government and investment money, and they don’t care enough to fix the most basic things that their website’s interface relies on a lot of the times, at least in professional environments. I guess they fixed it by creating CLI versions. It does not work for Claude for ChatGPT for Gemini and these are the ones I occasionally use. These are literally the biggest companies in this field, so I don’t reckon it works for a single one in existence.

All in all, these people do not care about the world, hell they do not even care about their own product. The only thing they care about is huge number go up and if it’s possible infinite rape and death just for the hell of it.