@ Five
2025-04-02 09:48:16
## Physical prototyping
3D printers have evolved as useful tools for rapid prototyping of physical products for inspiration and demonstration purposes.
Nowadays people also use them for hobbyist endeavors. They create nice little items for pure enjoyment or very simple substitution parts that can be used regardless of industrial precision and material limitations.
## Prototypers of Ideas
Similarly with LLMs, you can bounce ideas off them to better understand your own thinking and also let them generate variations of ideas. They provide inspiration, summaries, and proof-of-concept boilerplates. They are rapid prototypers for and of your ideas, creating a feedback loop.
However, to use LLMs successfully, people must understand that LLMs cannot and will never discover key insights. They are not creative in a human sense, just like 3D printers cannot produce a real car.
From engineering to art and business ideas, people use tools for almost all phases. The key innovation of LLMs is that they offer an effective tool to iterate faster in the ideation phase.
And yes, that's true of coding too. The code that LLMs generate is not production-ready, but it is a good starting point. Beware of using the prototype as the final product, as it will inevitably backfire. Use LLMs where their high-frequency but low-quality output is justified. Situations where low-quality boilerplate makes sense include experiments and other instances like code snippets, text summaries, sketches, demo videos, mockups, etc.
So, take your key idea and start the conversation. Apply critical thinking to identify where the answers lack important aspects and make mistakes, but most importantly, learn how to formulate your questions better.
## Economic Impact
3D printers lower the barrier to entry in the manufacturing of physical products. LLMs, on the other hand, lower the barrier to generating useful ideas. That is to say, they lower the barrier to produce anything and everything because ideas are the basis of any product. This has a deflationary impact and, in my estimation, will further lower the need for mediocre, unimaginative people across the board. It's the bureaucrats' nightmare.
Another impactful aspect of LLMs is that they are very effective learning tools, just like 3D printers. Learning in a feedback loop used to be available only by hiring a human. Now that is one prompt and a few sats away.
## Progress
LLMs need "quality data" to improve, i.e., more human insight and true creativity. Therefore, they feed on open-source. Closed-source proprietary information and "intellectual property" slow their progress, but not for long. Humans, overall, yearn for innovation more than they fear censorship and punishment. It's another Pandora's box, and the effects will never be reversed. People are essentially and foremost explainers of the world, creative innovators.
If you ask me, the human touch will be necessary at least until we have true AGI, and the path LLMs are on right now can never achieve that, not in a million years. I agree with David Deutsch's take on this: To produce real AGI, we must first crack serious concepts like how Meaning, Consciousness, Creativity, and Free Will work. We don't know how far we are from that, but it's not to be found along the path of what we call AI today.
If you are striving to become a better version of yourself and learn every day, rest assured: LLMs are your leverage. Otherwise, you will need some fiat authority, like the state today, to force others to value your work while it has an overall negative impact (90% in that category in my estimation). People like that need a shift of perspective or will suffer the consequences.
Keep printing ideas and execute on them!
Learn and repeat. The path to success.
#llm #ai #agi #3dp #programming