One thing that could explain this dichotomy is that writing such an article for oil refinery would be expensive task. Because data for such research may not be readily available online (not focussing on intentions). Just a guess. You will find more articles about things that are
easy to write about. If an article requires lot of field work, on foot research, talking to many people and synthesize and piece together disparate information in a coherent manner, don't expect it to be commissioned.
In similar vein, the following question is very apt. Please read the question because it captures all of our intuition when we try to understand something.
One answer is that it is only meaningful to answer this question if the answer has physical consequences. Popularly speaking, string theory is supposed to be the innermost Russian doll of modern physics, and there are no more dolls inside that we can explain it in terms. However, we may be able to find equivalent formulations.
I sometimes think "algorithms" understand me more than other people. Through my actions, they can diagnose me better than any doctor. e.g. meme therapy pages on facebook and tiktoks. I believe a constant stream of "best matched tiktoks/reels" to my situation" would be equivalent in value to going to a 3-star michelin restaurant and having their best dishes. It is available to everyone.
I think stoic ideas are from an era where their circumstances made them have those principles. We don't live in that era. It is possible to affect others and the associated cascading effect that can bring about a change in others action which were affecting you negatively. If Naval's idea of "individuals having leverage holds water" directly implies that you can change others, albiet slowly. If your reach becomes big enough that it becomes a threat that "other actors" need to curtail that reach through "algorithms" is another evidence that you were indeed having effects that they didn't like.
You think Marcus Aurelius was unable to affect other peoples' actions? That's not the idea. The point is that you can't directly make someone else think or feel a certain way, only act on them externally.
If causality holds, acting externally will have changes to internal assessments assuming good faith dialogue. Plus Marcus Aurelius was helpless in that it would take him lot of time and energy to give personal attention to each individual and clarify their doubts. He didn't have the technology to record his thoughts on a topic and refer people to it.
I could be wrong in this. But wasn't his writings only for himself and was published only later. How many people really referred to his writings? When was printing press invented? How popular was it compared to Bible? Was it possible for people to consume his writings in multimedia formats like video, audio? Were there meme-pages on tiktoks which contextualized his writings in different day-to-day scenarios so that the importance of his general ideas were imprinted on their minds? Did he have debate with others to defend his ideas watched by many, how would he respond to those counter-arguments? Would your mind change considering if his responses weren't that strong or on filmsy grounds?
> Were there meme-pages on tiktoks which contextualized his writings in different day-to-day scenarios so that the importance of his general ideas were imprinted on their minds?
I honestly cannot tell whether this is satire or if I just don’t want to be on this planet anymore.
Marcus Aurelis wrote in short snippets format. They were notes to himself which were latter organized into flowing organized essays by historians and writers.
I used to always wonder how do llms know whether a particular long article or audio transcript was written by say Alan Watts. Basically these kind of metadata annotation would be common while preparing training data for Llama models and so on. This could also be reason for the genesis for the argument that ChatGPT got slower in December. That "date" metadata would "inform" ChatGPT to be unhelpful.
Embedding is a transformation which allows us to find semantically relevant chunks from a catalogue given a query. Through some nearness criteria, you would retrieve "semantically relevant" chunks which along with query would be fed to LLMs and ask them to synthesize the best answer. Vespa docs are very great if you are thinking of building in this space. Retrieval part is independent of synthesis, hence it has its separate leaderboard on huggingface.
The other point is what if your Company's exit strategy is dependent on going big. Then it is pertinent that you plan for that 0.01% chance because there is no other exit.
With that money, Larry begged Jensen to take it and deliver him the GPUs. Onto the new grift. In yesteryears it was database and now it is AI. Benedict Evans always compared AI to databases. Oracle of AI is THE Oracle of Databases.
This idea could also be extended to domains like Art. Create new art styles for AI to learn from. But in future, that will also get automated. AI itself will create art styles and all humans would do is choose whether something is Hot or Not. Sort of like art breeder.
