https://aibussin.com/tags/sicql/Recent content in SICQL on AIBussin — AI applications, systems and booksHugoen-USSat, 08 Aug 2026 14:44:00 +0100https://aibussin.com/books/models-from-first-principles/04-chapter/Sat, 08 Aug 2026 14:44:00 +0100https://aibussin.com/books/models-from-first-principles/04-chapter/<h1 id="sicql--building-a-model-from-q-v-and-policy-networks">SICQL — Building a Model From Q, V and Policy Networks</h1> <p>In the previous post we took the MR.Q idea and expanded it into something richer.</p> <p>Instead of asking one question of a shared representation, EBT asked several:</p> <div class="highlight"><pre tabindex="0" style="color:#f8f8f2;background-color:#272822;-moz-tab-size:4;-o-tab-size:4;tab-size:4;-webkit-text-size-adjust:none;"><code class="language-text" data-lang="text"><span style="display:flex;"><span>How good is this state-action pair? -&gt; Q </span></span><span style="display:flex;"><span>How good is the state more generally? -&gt; V </span></span><span style="display:flex;"><span>What action should be preferred? -&gt; Policy </span></span><span style="display:flex;"><span>How much better is Q than V? -&gt; Advantage </span></span></code></pre></div><p>That already gave us a more expressive system.</p>https://aibussin.com/post/cbr/Wed, 03 Sep 2025 23:52:29 +0100https://aibussin.com/post/cbr/<h2 id="-summary">✨ Summary</h2> <p>Imagine an AI that gets smarter every time it works not by retraining on massive datasets, but by <strong>learning from its own reasoning and reflection</strong>, just like humans.</p> <p>Most AI systems are frozen in time. Trained once, deployed forever, they never learn from mistakes or build on successes. Real intelligence human or artificial doesn’t work that way. It learns from experience.</p> <p>This is the vision behind <strong>Stephanie</strong>: a self-improving AI that gets better every time it acts, not by fine-tuning, but by <strong>remembering, reusing, and revising</strong> its reasoning.</p>