Even being unable to see the eclipse live, I felt compelled to read Nightfall again, yesterday.
One of the things I appreciate about Asimov as a scifi author is that he never goes too deeply into explaining the science or physics of his novels (no technobabbles or midichlorians), he is more interested in the Big Ideas and characters.
I'm not sure that a world like Kalgash can exist, and many details about the story could be nitpicked to death, but it doesn't make it any less fascinating and complelling.
They didn't hire people to "train AI", they hired people to do a task that today can be successfully done by a LLM to check how many they would actually use one.
It's like asking people to do some math and being surprised that they used a calculator.
In Midjourney you get fantastic results just by using their discord and a text prompt.
To get some similar results in Stable Diffusion you need to set it up, download the models, understand how the various moving parts work together, fiddle with the parameters, donwload specific models out of the hundreds (thousands?) available, iterate, iterate, iterate...
Setting up the environment and tooling around in the code is not a burden, it's a nice change of pace from the boring code I have to deal with normally. Likewise, playing around to build intuition about how prompts and parameters correspond to neighborhoods in latent space is quite fun.
Beyond that, being able to go to sleep with my computer doing a massive batch job state space exploration and wake up with a bunch of cool stuff to look at gives me Christmas vibes daily.
Sure, but if Midjourney outputs a low quality results for your prompt, they are going to be much more difficult to improve. It's a black box at this point.
While with SD there can be multiple solutions for a single problem, but yeah, you have to develop your own workflow (which will inevitably break with new updates)
That's not stopping you from selling the cucumbers, just normalizes "quality classes" so customers know what they get. EU does stuff like that for many regional product, you can make cheese, you just can't name it say "Parmigiano Reggiano".
"or maybe that regulation is there for a reason (most times to protect consumers)"
Erm, can you tell me in what way a cucumber is dangerous to consumers, if its shape is bend a little bit more or less, so they need to be protected of that threat?
These regulations are descriptive. They exist so that there is a law somewhere that describes what a cucumber is.
Sidenote, this whole exchange is .. urgh. People criticizing things they haven't spent a minute to try to understand :/
Imagine if I came up to your monitor and started looking at individual lines of your code, and then I see an "x = 0", and I start criticizing it in a vacuum, completely cluelessly, as someone who has never coded before?
"People criticizing things they haven't spent a minute to try to understand"
Or maybe I have indeed worked on fields and packaging when I was younger and travelling and witnessed the throwing away of perfectly fine food, that just did not met some arbitary size regulations?
"Sidenote, this whole exchange is .. urgh"
But I agree to that. I am not really here to discuss the sense of defining cucumber sizes. If you are into that, have fun with it.
> can you tell me in what way a cucumber is dangerous to consumers, if its shape is bend a little bit more or less
I can't tell why a bent cucumber is dangerous, but I can tell tell why specifications for cucumbers are sometimes needed. When someone in the food industry needs cucumbers, he can't use any kind of cucumber, because it will go through machines with a calibre expectation. The industry that builds the machines also need specifications for their inputs.
Many regulations on fruits and vegetable are meant to help the food processing, since nowadays most of the raw food is processed by the food industry, not by consumers in their kitchens.
Oh, and you can still sell vegetable that are out of spec. AFAIK that's still legal in Europe, but since they can't have the right label, wholesale buyers may be hard to find at the same price.
" When someone in the food industry needs cucumbers, he can't use any kind of cucumber, because it will go through machines with a calibre expectation"
We were talking about consumer products.
"Oh, and you can still sell vegetable that are out of spec. "
Now you can. The regulations regarding sizes of cucumbers have been abolished in 2009, even though the whole theather about it was rather populism.
As far as I can see, if you produce a fruit that is healthy, with good coloration and the only thing “wrong” about it is the shape then you can still sell it.
You just can’t call your torus shaped fruit a cucumber because that’s not what anyone expects a cucumber to look like.
I'm looking at the USDA regulations and they seem reasonable to me. Looks like they're protecting consumers from being sold garbage labeled "cucumbers" instead of cucumbers labeled "cucumbers".
"The maximum diameter of each cucumber shall be not more than 2-3/8 inches and the length of each cucumber shall be not less than 6 inches"
Why is that reasonable?
This regulation means, that perfectly fine cucumbers that happen to be a bit smaller will get thrown away. Or well, into the lower grade category, but even in the lowest category
"maximum diameter of each cucumber shall be not more than 2-3/8 inches and the length of each cucumber shall be not less than 5 inches"
These are arbitary size regulations.
You do not need to regulate that to ban garbage. You can have a cucumber with the perfect shape, that is still garbage, because sunburned or too ripe or the plant was sick or whatever. Yes it should be illegal to sell garbage as fresh food and it probably is, but I do not see the connection to shape at all. This is merely aesthetics and personally I prefer a weird shaped ecological cucumber over a perfectly sized and shiny tasteless thing full of pesticides any day.
In all seriousness, if you’re paying by weight (or volume) then it should be OK not to have standard sized cucumbers… but pricing per unit should be reasonably consistent.
The thing is, chatGPT seems to be already capable to perform Google's function reasonably well.
If you ask for the top websites about [topic], it will output a shortlist of web links each with a description of the website.
You can narrow down, ask for a specific number of links, ask to exclude videos or specific websites from the results...etc.
If there was a chatGPT service with UI/speed/availability/up-to-date database of Google (a 2-months-old technology vs. a 25-years-old one), I would probably do my searches there almost exclusively.
In a world where we are bombarded by nearly limitless information, the ability to synthesise and focus on what one is already looking for is far more valuable than the breadth of results Google will output.
Salaries for devs in the US are insane compared to the rest of the world, I don't think they can be taken as a reference, especially as most people do not have an easy way to just choose to move there.
One of the things I appreciate about Asimov as a scifi author is that he never goes too deeply into explaining the science or physics of his novels (no technobabbles or midichlorians), he is more interested in the Big Ideas and characters.
I'm not sure that a world like Kalgash can exist, and many details about the story could be nitpicked to death, but it doesn't make it any less fascinating and complelling.