Thank you sincerely for your thoughtful comment.
Here are some of my points:
> I lost the drive to endlessly customize my phone.
The way to avoid that is to 'start at the maximum and gradually decrease'. I don't want to endlessly fiddle with everything while not missing out on anything.
> I gladly put into "empathy and philosophy" ones.
Actually, my anticipated answer is that I would use Tool Intelligence to do this well. But as soon as I say that, people will come at me cursing. I'm trying to find something to say in response to that.
> you can't know what you need until you're facing a real problem in a real domain
My point is, many people can't figure out what tools are useful to them even when they have a real problem in front of their eyes.
> I really enjoyed how the post was laid out —- the structure caught and kept my attention and made the arguments easy to follow.
Thank you.
> Further, I think that it's easy to actually overwhelm yourself with too many tools, at which point you've exhausted your interest and willingness to learn them.
I think most of the counterarguments in the comments to this post will have a similar idea to this. My thought is that this phenomenon occurs because of a lack of tool intelligence in handling tools. What those comments show is that more content is needed on how to survive among numerous tools and how to easily filter tools. Some things are hard to use alone without something else, and this seems to be the case.
> I think the best tool use and evaluation comes from a sense of immediate necessity
You assume that people can know what tools they need. Considering my observations, that doesn't match reality. That doesn't mean a dancer has to go through all the Excel documents they don't even use, but it's easy to narrow down the candidates for what might be relevant to you.
This is what I think. I admit that the line between the main task and the tools is blurry, but usually the difficulty of the tools tends to be proportional to the difficulty of the main task. As you said, if it's a simple task done with limited cognitive ability, the tools for it are also likely to be simple. Another point is that while there is an upper limit to how difficult tools are to use, there is no lower limit, so there are always many tools that can be mastered simply.
And my view that people overestimate the difficulty of learning them can be shown by this example. If someone made 300 more keyboard shortcuts for a code editor, you'd probably think it's crazy and a mess. But for people learning a foreign language, memorizing 300 completely new words is just everyday life, and it can be done in a day or a few days. Also, this is a long story to tell, but in my experience, it's easier to memorize keyboard shortcuts than foreign language words.
Don't you think that's impractical? All attempts to stop people from using the Internet to do things on their own have failed. If the creation of such images is considered a crime in itself, then it will be obscured. Therefore, treating dissemination as a crime is not only logical, but also practical.
And when it comes to accidental leaks, even if the person who created the image didn't pay attention, the leak happens by somebody anyway. I think it would be more practical and lead a better result to punish those responsible for the leak itself
I admit that my criteria for what I consider creative is vague. To explain it roughly, it means work that is not just modifying something that already exists. More specifically, when I say that LLMs can't do creative work, I mean that I have never actually used an answer obtained from a prompt like "Create something new and original (text content, technology, novel scenario, etc.)" in real life.
Treating hallucination as an error rather than a fundamental limitation is simply a practical way of thinking. It means that, depending on how it's handled, hallucination can be mitigated and improved upon. Conversely, if it's regarded as a fundamental limitation, it would mean that no matter what you do, it can't be improved, so you'd just have to twiddle your thumbs. But that doesn't align with actual reality.
No advice has a 100% success rate. The existence of failure cases doesn't inherently make the advice bad. The advice to pursue your passion as a startup idea is still relatively good advice. It's the best way to avoid the most trite and cliché ideas. I don't know what someone passionate about spreadsheets would create, but it would be more unexpected than building a generic AI writing app. Whether it succeeds or fails.
You are confusing words. In the context of the OP, "having many options" doesn't mean indiscriminately hoarding everything. It's like making choices in chess that result in having more options for moving pieces. The "premature optimization is the root of all evil" saying is an even more lacking analogy, but it conveys a similar idea. In other words, the advice to increase your options applies even if you dedicate your entire life to just one thing.
Intuitively nonsensical but irrefutable thought:
Everything about how animals think is based on human assumptions. They might actually like being eaten by humans. How can we know, when the animals haven't said it themselves? In fact, there are people like that in the world. Such people have pain receptors and higher cognitive abilities, yet they still think that way. In other words, analyzing their physiology alone cannot determine whether animals dislike being eaten.
I agree. But this is just out of curiosity, what about jobs that don't have repetitive tasks and don't feel like coding, after you've spent enough time on coding. For example, business manager.
No, if I had to nitpick, I'd say 'talk is expensive'. The original text doesn't say "Somebody might even talk about how they want to do X" like your comment, and it doesn't need to. But maybe someone who finds 'some reason not to do' will say it, and they won't actually do it.
(B has never said he wants to learn to code)
A: Do you have any plans to learn to code? It seems to be useful.
B: No.... (thinking) Well, AIs.
(And B doesn't learn to code in the real world)
Actually, this comments doesn't matter at all, but I leave it in for fun
That's yes difference that it's short time relief. And you're already caring about the problem now. In fact, it's even possible that you can be caring about that because you've had a Big Mac to reduce your stress. In the context of this discussion, drugs mean something preventing you from doing this. The Big Mac is just junk food.
When we talk about after scarcity, we're talking about an endlessly sharable amount of wealth. (Not in a technical sense, of course)
Then the super-rich will start sharing their wealth, because if they don't, their competitors will do it first and take all the fame. It is not from the benevolence of the super-rich, the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own interest.
I'm guessing, but I think the parent comment was meant to imply that "solver of all problems that doesn't solve the problem of "people having nothing to do"" is a contradiction.
Their wealth is scarce in that they can't share it endlessly. No one is ultra-wealthy enough to say, "I'm going to give $1 million to random people so that people don't have to suffer anymore. That's a critical difference.
I'm curious to see how that actually compares in effectiveness to wearing headband-type speakers and listening to sleep brainwave stimulation music (which I tried).
I'm curious if the OP has tried any of the libraries that control the output of LLM (LMQL, Outliner, Guadiance, ...), and for those who have: do you find them as unnecessary as LangChain? In particular, the OP's post mentions the problem of not being able to generate JSON with more than 15 items, which seems like a problem that can be solved by controlling the output of LLM. Is that correct?