Except McLuhan himself was quoted as saying "We shape our tools and thereafter our tools shape us". I think what the blog post was trying to get is the dangerous framing of calling it a tool given...
Except McLuhan himself was quoted as saying "We shape our tools and thereafter our tools shape us". I think what the blog post was trying to get is the dangerous framing of calling it a tool given how we've previously thought about tools, but McLuhan would likely argue that all tools have this property and its not safe to ever think of tools as some self-containing thing apart from us.
To be clear, I'm not suggesting LLMs are "merely a tool", but rather we should approach all tools with a great deal of caution. Anytime we use a tool to accomplish something we're giving something up. And specifically with LLMs what we're giving up is harder to see than with any other tool we've encountered before. The shape of that "giving up" is still a complete unknown, I'm personally still grappling with it and feeling really ambivalent in the process.
You could say "a computer is just a tool". But then observe the lives of many people waking up to look at a screen, looking at a screen whenever they have a free moment, looking at a screen all...
You could say "a computer is just a tool". But then observe the lives of many people waking up to look at a screen, looking at a screen whenever they have a free moment, looking at a screen all day for their job, looking at a screen after work to relax, looking at a screen in the moments before they go to sleep.
Nice read. My thoughts on tools is that the learning curve of using any tool is changing the way you perceive the world itself, just like driving a car is an entirely man-made abstraction of...
Nice read. My thoughts on tools is that the learning curve of using any tool is changing the way you perceive the world itself, just like driving a car is an entirely man-made abstraction of multiple perceptions (watch the road) and actions (steer, pedal, break). A big part of design is into designing that abstraction to be total and exhaustive, like the handle on a teapot is basically one. In a way, we trade off an old set of mental models for a new one in order to integrate our tools.
AI (I assume in this case, LLM), subtly shifts our perceptions, because it works in the domain of words, in other words, symbols detached from the symbolized. In writing, one has to choose the words carefully to mean what they mean, no more and no less; this is not the case with LLMs, where the corpus will happily fill in whatever unrelated words you don't specify yourself and hence contaminate your mental context.
Using AI has no deliberately designed mental model, because language is an always-shifting thing, and we cannot freeze the meaning of words to be an exact semantic set at any time. Hence the language of many different contexts (same word can mean different things) are blended together. Words don't mean what they mean anymore, and we grow to lose the plot.
Except McLuhan himself was quoted as saying "We shape our tools and thereafter our tools shape us". I think what the blog post was trying to get is the dangerous framing of calling it a tool given how we've previously thought about tools, but McLuhan would likely argue that all tools have this property and its not safe to ever think of tools as some self-containing thing apart from us.
To be clear, I'm not suggesting LLMs are "merely a tool", but rather we should approach all tools with a great deal of caution. Anytime we use a tool to accomplish something we're giving something up. And specifically with LLMs what we're giving up is harder to see than with any other tool we've encountered before. The shape of that "giving up" is still a complete unknown, I'm personally still grappling with it and feeling really ambivalent in the process.
That’s rather ominous, but vague. What would be an example of the sort of thing they’re warning about?
You could say "a computer is just a tool". But then observe the lives of many people waking up to look at a screen, looking at a screen whenever they have a free moment, looking at a screen all day for their job, looking at a screen after work to relax, looking at a screen in the moments before they go to sleep.
Nice read. My thoughts on tools is that the learning curve of using any tool is changing the way you perceive the world itself, just like driving a car is an entirely man-made abstraction of multiple perceptions (watch the road) and actions (steer, pedal, break). A big part of design is into designing that abstraction to be total and exhaustive, like the handle on a teapot is basically one. In a way, we trade off an old set of mental models for a new one in order to integrate our tools.
AI (I assume in this case, LLM), subtly shifts our perceptions, because it works in the domain of words, in other words, symbols detached from the symbolized. In writing, one has to choose the words carefully to mean what they mean, no more and no less; this is not the case with LLMs, where the corpus will happily fill in whatever unrelated words you don't specify yourself and hence contaminate your mental context.
Using AI has no deliberately designed mental model, because language is an always-shifting thing, and we cannot freeze the meaning of words to be an exact semantic set at any time. Hence the language of many different contexts (same word can mean different things) are blended together. Words don't mean what they mean anymore, and we grow to lose the plot.