This is Brad DeLong's Grasping Reality—my attempt to make myself, and all of you out there in SubStackLand, smarter by writing where I have Value Above Replacement and shutting up where I do not… On Figuring Out What Your Comparative Advantage IsBeyond the 'bot: using “AI” to make writing painful self-assessment documents somewhat easier. Smart things to think about from David MacIver. Dealing with the fact that AI chatbots can churn out...
Beyond the bot: using “AI” to make writing painful self-assessment documents somewhat easier. Smart things to think about from David MacIver. Dealing with the fact that AI chatbots can churn out text that’s clear, competent, but—let’s be honest—indistinguishable from that of the TIS, the Typical Internet S***poster. For those who struggle with prose expression, this is a democratizing force. The rise of AI-generated writing will make it easier than ever to produce text that’s “good enough”. And “good enough” is often good enough. Raising the floor of serviceable prose is a blessing. But for anyone seeking insight, originality, or genuine clarity, the machine’s output cannot be more than just a very rough indeed initial draft. The real work begins with self-knowledge: understanding your own comparative advantage and knowing how to revise, rewrite, and push past the median of the digital ‘bot…
What are you particularly good at? David MacIver has a very good mental exercise régimen to help you find out:
Find that. And you have found what you are good at. But then what if you have to get that down on paper for some purpose? David has more good advice:
However, this won’t work unless you are already a good writer. Technology expands what’s possible, but progress depends on human effort. AI can mimic average internet writing, but it cannot create true insight or clarity. The digital world is full of text, but only good writers stand out. This work strategy is, I think—the initial complaint, the scenery-chewing, and then the ‘bot-created strings of words—is designed to produce bad writing that is nevertheless full of ideas. And, in my experience, that can provide you with a truly powerful springboard. After all, ChatGPT or Claude will, at bottom, give you nothing other than what a TIS—a Typical Internet S***poster—would write. This is how these models work: they learn from the mass of online text and spit out some interpolation driven by its compressed internal model of typical word patterns. There are ideas buried in the word-patterns it produces, yes. But they are buried. Now GPT LLM boosters, at this point, begin talking frantically about RAG, RLHF, prompt engineering, data curation, feedback loops, careful data chunking and indexing, embedding optimization, hybrid search and reranking, supervised fine-tuning training from high-quality and human-curated examples, human reward modeling, proximal policy optimization, continuous feedback loops, custom base prompts, context injection, dynamic prompt templates, semantic chunking, metadata tagging, regular content updates, fine-tuning models on domain-specific data, and so on. These are patches, not solutions. Relying on software to mimic a typical internet poster never works for good writers. If you write well, AI output can never be better than just a very rough draft. You really then do need to “let the hate flow through you. Delete whatever drivel it wrote and write something better, because you can definitely improve on that…” Of course, if your writing is currently worse than that of the TIS—or if you have to write in a language in which you are not fluent—AI can really help. ChatGPT and Claude can then produce text that can be clearer and less embarrassing. For weak writers, AI does powerfully lift the floor. And I see this as a very good thing, this democratization of the power to write serviceable prose. But the ceiling is relatively low. Especially in the eyes of people like me who so often read something and think: “Why didn’t they take just a little more time and express that better? It really is not very hard at all…” So: self-knowledge is a good place to start. And so here we have found another example of how new infotech can really be a cognitive force multiplier, but only if you can first start from a true picture of your own mind, and of its strengths and weaknesses. If reading this gets you Value Above Replacement, then become a free subscriber to this newsletter. And forward it! And if your VAR from this newsletter is in the three digits or more each year, please become a paid subscriber! I am trying to make you readers—and myself—smarter. Please tell me if I succeed, or how I fail…#on-figuring-out-what-your-comparative-advantage-is |
| ||||||||||
No comments:
Post a Comment