Some are placing their bets and hoping to win big, some are just having fun and exploring the new technology and most are probably like me and have found those few areas that the technology works for them and ignores everything else. It is a fairly amazing technology, as a humanities sort that has no clue when it comes to programming but also has some things I want to make, LLMs are invaluable and not in the sense that they can write them for me—I am not capable enough in programming to get them to do what I want—but in the sense they can present things in a way a humanities sort can understand so I can write those programs I want.
A recent prompt of mine:
>Write a dialogue where H.P. Lovecraft and David Foster Wallace are tasked with developing a DSL for audio synthesis and composition, they are trying to sell each other on their preferred language, Lovecraft wants rust, Wallace Zig. Their discussion should be deeper than just the talking points and include code examples. Thomas Pynchon is also there, he is a junior dev and enamored with C, he occasionally interjects into the discussion and generally fails to follow the discussion but somehow manages to make salient points; Lovecraft never address Pynchon or C but alludes to the horror, Wallace tries to include and encourage Pynchon but mostly ignores him.
I pretty much knew how ChatGPT would use each of them; Lovecraft would view everything other than rust as some lurking unnameable horror, Pynchon would seem random and nonsensical, and Wallace would vainly attempt to explore all options without weighing in, and that is exactly what happened. Being able to get this sort of information within contexts I understand is amazing, 15 or 20 minutes later I had more direction than I ever had before and it answered a dozen or so language agnostic questions I had regarding implementing such a DSL which is really what I was after. And it gave me some good laughs.
A year ago I thought these projects would never move beyond being ideas, when ever I tried to get help from people they just told me what I was doing wrong and tried to send me down the path of their ideals, how they think things should be done, which was probably more a failure on my part than theirs with my being a humanities sort and incompetent programmer. LLMs have infinite patience and time, those people who I sought help from in the past were not completely wrong but they did try to lead me down a wrong path for what ever reason, partially because they don't have infinite time and patience but also because they thought they were right.
Henry James has been walking me through the PureData source code this past week.
Everyone is not hyping AI/LLM, that is just the bias of tech communities and the like, most of us see it about the same as we see a blender or toaster. Can't remember the last time AI/LLMs came up in general conversation for me.
A recent prompt of mine:
>Write a dialogue where H.P. Lovecraft and David Foster Wallace are tasked with developing a DSL for audio synthesis and composition, they are trying to sell each other on their preferred language, Lovecraft wants rust, Wallace Zig. Their discussion should be deeper than just the talking points and include code examples. Thomas Pynchon is also there, he is a junior dev and enamored with C, he occasionally interjects into the discussion and generally fails to follow the discussion but somehow manages to make salient points; Lovecraft never address Pynchon or C but alludes to the horror, Wallace tries to include and encourage Pynchon but mostly ignores him.
I pretty much knew how ChatGPT would use each of them; Lovecraft would view everything other than rust as some lurking unnameable horror, Pynchon would seem random and nonsensical, and Wallace would vainly attempt to explore all options without weighing in, and that is exactly what happened. Being able to get this sort of information within contexts I understand is amazing, 15 or 20 minutes later I had more direction than I ever had before and it answered a dozen or so language agnostic questions I had regarding implementing such a DSL which is really what I was after. And it gave me some good laughs.
A year ago I thought these projects would never move beyond being ideas, when ever I tried to get help from people they just told me what I was doing wrong and tried to send me down the path of their ideals, how they think things should be done, which was probably more a failure on my part than theirs with my being a humanities sort and incompetent programmer. LLMs have infinite patience and time, those people who I sought help from in the past were not completely wrong but they did try to lead me down a wrong path for what ever reason, partially because they don't have infinite time and patience but also because they thought they were right.
Henry James has been walking me through the PureData source code this past week.
Everyone is not hyping AI/LLM, that is just the bias of tech communities and the like, most of us see it about the same as we see a blender or toaster. Can't remember the last time AI/LLMs came up in general conversation for me.