The Role of AI in Accelerating Skill Development

Sure, some historical books in the US South are wildly inaccurate!

This is why we deal with multiple sources, as well as defining, validating, and exploring our sources. ChatGPT does not do this. It does not pull from definable sources, it is an LLM that outputs text based on statistical word sequences from an almost bottomless source.

You can learn from it, but the sources may be sketchy, and it will mix source biases. I tried to ask it for sources on the fall of ancient Rome:

>> As an AI language model, I don't have the ability to consult sources or remember my previous responses. However, the information I provided about the fall of Rome is widely accepted in historical scholarship and can be found in many reputable sources such as encyclopedias, textbooks, and academic articles.

Personally, I think we need to tune the LLMs according to use. Creative? Then it one way. More factual, the other way. This does mean we would need to know sources for LLMs.

Example:

>> If I'm feeling a little horse, where am I?

>> If you're feeling a little horse, you might be experiencing a sore throat or hoarseness in your voice. This is a common symptom of many respiratory illnesses, such as the common cold, flu, or bronchitis, as well as other conditions that affect the throat and vocal cords, such as acid reflux, allergies, or overuse of the voice. It's important to stay hydrated and rest your voice if you're experiencing hoarseness, and to seek medical attention if your symptoms persist or worsen over time.

Long answer, following my latest involvement with AI stuff:

I haven't really played that much with text models, ChatGPT and such, but I can kind of relate to this when Stable Diffusion is involved. I'm a fairly competent artist, and I enjoy drawing both elaborate art pieces and doodling random stuff when bored or when someone says something funny in my socials, and despite the hate of many of my peers, I picked it up and got my own local instance to have fun with. This thing has fully revolutionized my world as an artist. It's only been like half a year and I don't think I have drawn so much since high school, specially during tough work seasons where I'd normally be busy and too tired to draw in my free time. You can say that every year my quality goes up by a roughly equivalent amount, and this year I'm already outpacing myself.

I don't know, I can't quite explain it in simple terms, but I'd wager it's both the stimulus of seeing art of topics and characters that would normally take months or years between a single new image, and the AI failing at certain elements often (hands, fine details and decorations, eyes, etc...) making you more aware of those elements and thus becoming more mindful of them when drawing, even at a somewhat subconscious level. Not to mention when the AI does something hilarious and you kinda make it a thing or draw it "for real" from scratch.

And, thinking about it, there's also some ancillary art exercises the AI has gotten me into. Editing images (for training or "this generated image will be 100% perfect if I redraw this part myself" cases, which makes you learn a lot because of the huge amount of different styles this can generate. Gotta learn to recreate that shading style on the spot!), stuff like ControlNet accepting quick sketches as input, and wanting to duplicate some of the more fascinating things the AI generates. Training models can become a full "art project" given enough love and care and manual art supply. And this also incentivizes to be very clear and tidy for the training to go well. And all of this becomes practice and experience.

Being able to generate really decent images in a few seconds is never going to take away my fun or motivations when drawing, and it's opening new paths to have fun with art, too. In that regard, and finally tying with the thread's content, if the AI has managed to not only help me improve but make me even more active in art, I'm sure it can have similar effects in other fields as well. A writer can easily improve, learning from the terrible goofs, hilarious twists and occasional "getting it just right" moments the text models provide. I can see the same thing happening in areas like music or whatever comes next. An observant and motivated use can turn a ML system into a goldmine of custom examples, and at a ridiculous fast pace, without tiring and on demand!.

And it's fun to tinker with the code, too.

The only way what you're saying is true if there is a limit to the amount of new opportunities everyone being so productive will create. There's nothing to suggest there is.

I often feel depressed about how hard it is to make products better, if this tool enables that to happen, guess what I'll be doing? Building more products.

Also, many developed countries such as Japan, South Korea etc were almost facing economic collapse due to not having enough workers, so they basically "need" these tools to survive in the future.

Maybe we're going to see a world with less people, and an abundance of awesome shit?

For getting shipping minimum viable product and say hello to shipping maximum awesome product?