<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Design on Systems &amp; Engineering</title><link>https://valery.tech/ai/design/</link><description>Recent content in Design on Systems &amp; Engineering</description><generator>Hugo</generator><language>en-US</language><copyright>Copyright (c) 2014-2023</copyright><atom:link href="https://valery.tech/ai/design/index.xml" rel="self" type="application/rss+xml"/><item><title>Amortized Generation Problem</title><link>https://valery.tech/ai/design/amortized-generation-problem/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://valery.tech/ai/design/amortized-generation-problem/</guid><description>&lt;p&gt;There is &lt;strong&gt;amortized problem solving&lt;/strong&gt;: the model, viewed as an agent, has learned a mapping from prompts to plausible outputs during training and applies that mapping at inference time, without necessarily instantiating the problem as an explicit structure and solving that structure.&lt;/p&gt;</description></item><item><title>Prompt Driven</title><link>https://valery.tech/ai/design/prompt-driven/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://valery.tech/ai/design/prompt-driven/</guid><description>&lt;h1 id="prompt-specified-systems-and-prompt-only-enforcement"&gt;Prompt-Specified Systems and Prompt-Only Enforcement&lt;/h1&gt;
&lt;p&gt;A &lt;strong&gt;prompt-specified system&lt;/strong&gt; is one in which prompts specify a substantial part of the system&amp;rsquo;s functional behavior and an LLM interprets that specification at runtime. The prompts may describe analysis steps, transformations, decision criteria, domain rules, workflows, constraints, or output structure.&lt;/p&gt;</description></item></channel></rss>