> “So why isn’t it engineering? Well, is asking a question “engineering”? Excuse me for being pedantic but from what I’ve learned it’s part of the field of “philosophy”. There is no engineering involved. And even if there is “skill” required to formulate a well-structured question for the LLM, that is not an engineering pursuit.”
Not a convincing argument when we’re talking about software.
Is designing a data structure engineering or philosophy? Go back 200 years and ask scholars whether determining the abstract properties of a non-physical object is engineering. They’d say “no” and point you towards Kant. Yet today that kind of thought process underlies engineered things that we call names like “infrastructure” and “products” even though they don’t fit the definition those words had as recently as 60 years ago.
Data structures require mapping a conceptual model ("philosophy") to a representation by a physical machine. As a data structure engineer you will need to make engineering decisions such as alignment of data, data-precision (*), and behavior under different runtime regimes (single threaded, concurrent). All those are technical decisions that a mere 'philosophy' is not equipped to address.
* a hot topic in ML world actually. Precision required for the params.
Philosophy is not a bucket where "soft" or "unskilled" knowledge work goes either. The author is doing themselves a serious disservice in not really explaining what about prompt engineering makes it not "engineering" and what about philosophy matches it better-- and I say this as someone who is on the caboose of the AI hype train.
Comments
> “So why isn’t it engineering? Well, is asking a question “engineering”? Excuse me for being pedantic but from what I’ve learned it’s part of the field of “philosophy”. There is no engineering involved. And even if there is “skill” required to formulate a well-structured question for the LLM, that is not an engineering pursuit.”
Not a convincing argument when we’re talking about software.
Is designing a data structure engineering or philosophy? Go back 200 years and ask scholars whether determining the abstract properties of a non-physical object is engineering. They’d say “no” and point you towards Kant. Yet today that kind of thought process underlies engineered things that we call names like “infrastructure” and “products” even though they don’t fit the definition those words had as recently as 60 years ago.
Data structures require mapping a conceptual model ("philosophy") to a representation by a physical machine. As a data structure engineer you will need to make engineering decisions such as alignment of data, data-precision (*), and behavior under different runtime regimes (single threaded, concurrent). All those are technical decisions that a mere 'philosophy' is not equipped to address.
* a hot topic in ML world actually. Precision required for the params.
Philosophy is not a bucket where "soft" or "unskilled" knowledge work goes either. The author is doing themselves a serious disservice in not really explaining what about prompt engineering makes it not "engineering" and what about philosophy matches it better-- and I say this as someone who is on the caboose of the AI hype train.