Btw I didn't mean to imply it's forbidden information. It's just not that good/impressive/sharp. It often just doesn't inspire going back for more.
Also like I said, I find it useful for software design. It ultimately makes a bunch of mistakes but I find the conversation useful and I walk away with usable ideas. I don't often start new software, so I don't go to it often for this.
I think it's useful at the high level stuff; giving you ideas and laying out options, like the chat log you posted. But as you dive into detail I find two things happen:
(a) It starts to make egregious errors.
(b) You bias it with your more detailed questions.
When I talked to it about electricity, it started contradicting itself about how energy is carried, whether by electrons' kinetic energy or by "the field". I asked it "so is electrical energy encoded in electrons' kinetic energy or not?" and then it started to say that oh yes, that's how it's encoded. It seems it used my question to reinforce that knowledge which previously it was confused about. Since I personally don't know how electrical energy is carried, I lost confidence that I'm learning correctly. This happens often. "I'm sorry about the confusion. You are right. ..."
As others suggested, this is likely a summer vacation dip, but speaking for myself as a non-student, my GPT (4 + 3.5) usage has plummeted.
If you use it in a "what should be done" or "what is the right thing" sense, you get useless generalities. "It is important to remember that ... <there are many sides to the situation>". Suppose for example that you talk to it about combating judicial bias. It will largely insist that judges are unbiased because they're supposed to be unbiased according to their job description... You won't walk away with any insights from engaging it in ethics/morals topics.
If you dive into depth with some subject, it will invariably make errors and contradict itself in the details. I dove into how electricity works and it was saying and enthusiastically confirming electrical energy is encoded in electrons' kinetic energy. Then it started claiming that electrical energy is definitely not in the kinetic energy of electrons, that electrons don't lose speed when they bump into atoms, and instead the energy is "in the field". Ok... YouTube it is.
It's at least less convenient than a search engine at routine fact-finding due to its knowledge cutoff, inability to display images and links, etc. Ditto with translation. It's good but so is Google Translate and Google Translate has in-line suggestions for misspellings, a menu of languages, etc.
Ok, it's pretty good at writing hilarious poems, imitate people ("in the style of", etc.). Friends and I sometimes text each other hilarious things GPT said. But even there the machine is sort of laid bare, and you can see its creative limits. Its poetry is ultimately pretty lame, basically, and gets predictable. It's like an insistent and repetitive 8 year old with a lot of knowledge to pull from. It's not something I do often.
It's good at writing boilerplate code but honestly I just don't reach for that use case often. I like to stay in my own flow rather than constantly delegate to a robot and check its work, which I find flow-breaking, especially if it doesn't understand my APIs and makes mistakes often.
I do, sometimes, engage in architectural discussion about how I should design something at a high level. It's full of mistakes but bouncing ideas around can be useful. But I only feel the need to do that once in a great while.
These issues are exacerbated by being forced to downgrade to 3.5 after using 4 for a bit. You might feel like you're getting somewhere with 4, and then 3.5 starts laying doozies on you. This is another UI/product issue, but affects usability.
> I think that we would get back to “organized society”, yes.
This is roughly what I meant to express. Didn’t mean to imply we’d return to where we are now, but ultimately the drive to survive and thrive would bring us back to some kind of organized society.
My core point being that “cavemanship” isn’t some kind of stable state like Kaczynski insists. It’s an ephemeral/unstable state on the way to some higher level organized society that balances tensions.
I kinda think that caveman life was inherently temporary, as we advanced toward organized society. You could destroy society and information, sending us back into a caveman lifestyle, but it would only be temporary as we would immediately start to climb back toward organized society. Organized society emerges out of the drive for survival. As you seek to secure food and security for your group, you build tools and you become weary of neighboring groups. You compete, you iterate on tools and strategy, you develop borders and politics, and eventually we’re back to some kind of global infrastructure, with everything standing in a complex tension where it’s no longer easy to expand quickly.
This "third position" is basically anarchy, or anarcho-primitivism.
It's not really a political position, it's more like a pathological insistence that civilization should be destroyed. Kaczynski went as far as saying we're biologically wired to live primitive lives and we're misusing ourselves in perversions that he calls 'surrogate activities', which include science and art. To him, hunting or trying to survive while being hunted in the woods is the perfect utilization of the human mind-body, and we should all abandon society and go do that instead.
This is true, especially Germany and Italy, and Germany in particular. Even now they're in the process of shutting down more nuclear power, to become even more dependent on Russian gas. It's idiotic and pathetic, really. I hope Scholz has more of a brain than Merkel.
Your reply seems rooted in the fairly widespread assumption that having your own online business that makes solid "mrr" is some kind of supreme achievement and you should relish in it solely.
Some people have higher-dimensional desires and interests in life.
Also like I said, I find it useful for software design. It ultimately makes a bunch of mistakes but I find the conversation useful and I walk away with usable ideas. I don't often start new software, so I don't go to it often for this.
I think it's useful at the high level stuff; giving you ideas and laying out options, like the chat log you posted. But as you dive into detail I find two things happen:
(a) It starts to make egregious errors.
(b) You bias it with your more detailed questions.
When I talked to it about electricity, it started contradicting itself about how energy is carried, whether by electrons' kinetic energy or by "the field". I asked it "so is electrical energy encoded in electrons' kinetic energy or not?" and then it started to say that oh yes, that's how it's encoded. It seems it used my question to reinforce that knowledge which previously it was confused about. Since I personally don't know how electrical energy is carried, I lost confidence that I'm learning correctly. This happens often. "I'm sorry about the confusion. You are right. ..."