<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
font-size:10.0pt;
font-family:"Calibri",sans-serif;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
span.EmailStyle19
{mso-style-type:personal-reply;
font-family:"Calibri",sans-serif;
color:windowtext;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;
mso-ligatures:none;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style></head><body lang=EN-US link=blue vlink=purple style='word-wrap:break-word'><div class=WordSection1><p class=MsoNormal><span style='font-size:11.0pt'>Before AI, there was code completion in IDEs that would suggest alternative functions that could be used. At the time this was introduced, I was suspicious of this because it seemed to show that developers could not remember their code. When I’m working on a code a lot, I do remember details and don’t need to be prompted. For programming languages I learned early in life, like C, the grammar is present as a motor skill. <o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'><o:p> </o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'>Later when I adopted functional programming, I also adopted the Wadler philosophy [1] and sought to make types reveal more meaning. Autocompletion makes more sense in this context, especially with languages like Lean, as the types can be rich in meaning. <o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'><o:p> </o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'>With generative AI, it is a step further. Drafting code is automatic but one must thoroughly read the code to see if there is a misunderstanding, or if underspecification is creating a mess. To some extent debugging code is automatic too. However, I sometimes find my intuition is quite different from the AI tactics. In part it is declarative programming, but also sampling the space of programs in the vicinity of an idea. Given the reality of context window limits, using AI for programming is iterative over multiple sessions. Thus, one communicates not just with natural language but with the evolving code itself -- code that can have informative types and even carry proofs. The weirdest thing about using AI is that it has no opinions. Claude will rewrite code without asking (seemingly having no self-control), but it will not confront you like a frustrated colleague might. It is happy to let you make a mess provided its sense of idiomatic code patterns are satisfied.<o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'><o:p> </o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'>[1] <a href="https://people.mpi-sws.org/~dreyer/tor/papers/wadler.pdf">https://people.mpi-sws.org/~dreyer/tor/papers/wadler.pdf</a><o:p></o:p></span></p><p class=MsoNormal><span style='font-size:11.0pt'><o:p> </o:p></span></p><div style='border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0in 0in 0in'><p class=MsoNormal style='margin-bottom:12.0pt'><b><span style='font-size:12.0pt;color:black'>From: </span></b><span style='font-size:12.0pt;color:black'>Friam <friam-bounces@redfish.com> on behalf of steve smith <sasmyth@swcp.com><br><b>Date: </b>Tuesday, February 11, 2025 at 8:22 AM<br><b>To: </b>friam@redfish.com <friam@redfish.com><br><b>Subject: </b>Re: [FRIAM] genai and critical thinking<o:p></o:p></span></p></div><div><p class=MsoNormal><span style='font-size:11.0pt'><br>On 2/11/25 8:20 AM, glen wrote:<br>> That's a fraught question. First, editors need not have been writers <br>> before they became editors. But barring that, my answer would be "No". <br>> But they prolly *do* lose facility for writing, the ease with which <br>> they write. It's simple reinforcement. Use it or lose it.<br>><br>> E.g. I can still code in Ada. But I'm way worse at it now than I was <br>> when I did it multiple days per week. A better question might be: Do <br>> editors lose their ability to read? And that question bears an even <br>> deeper problem ... something akin to Gell-Mann amnesia ... and I blame <br>> it for me losing my taste for reading for *fun*. Up until ~1998 or so, <br>> I did a lot of reading for fun. It was fun to read. Now reading is <br>> merely a means to some other end. Make something your job and it <br>> ceases to be a hobby. So even if editors retain their ability to read, <br>> the *quality* of their reading must change in deep ways.<br><br>I think this is the more salient aspect of the general question... and <br>it may even be "generational" in the sense not that readers/writers lose <br>their skills through atrophy but if they lose their "taste" or <br>"facility" for it and a new generation simply *never acquires it*. I <br>never acquired a significant ability or facility for writing <br>longhand/cursive, and I do think it limits me and how I <br>think/feel/perceive the world.<br><br> "kids these days" who have never read anything longer than a short <br>paragraph on the back of a cereal box (Boomers/X) or a Tweet <br>(millenial/Z) probably do perceive the world somewhat differently than <br>those of us who may still read novels, entire non-fiction books, and <br>long-form journalism. I myself have atrophied in this regard... I <br>tend to look to YouTube and Audiobooks (and Podcasts) to consume what I <br>once looked to full-length printed books.<br><br>But to your (glen's) point, there are qualitative thresholds which are <br>perhaps more salient than the quantitative ones...<br><br><br>><br>><br>> On 2/11/25 6:57 AM, Marcus Daniels wrote:<br>>> Do editors lose the ability to write?<br>>><br>>>> On Feb 11, 2025, at 6:43 AM, glen <gepropella@gmail.com> wrote:<br>>>><br>>>> The Impact of Generative AI on Critical Thinking: Self-Reported <br>>>> Reductions in Cognitive Effort and Confidence Effects From a Survey <br>>>> of Knowledge Workers<br>>>> <a href="https://www.microsoft.com/en-us/research/uploads/prod/2025/01/lee_2025_ai_critical_thinking_survey.pdf">https://www.microsoft.com/en-us/research/uploads/prod/2025/01/lee_2025_ai_critical_thinking_survey.pdf</a> <br>>>><br>>>><br>>>> It really doesn't seem that different to me from numerical analysis. <br>>>> It shifts the work from doing the computing to declaring what the <br>>>> computing should do.<br>><br>><o:p></o:p></span></p></div></div></body></html>