<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: sdeframond</title><link>https://news.ycombinator.com/user?id=sdeframond</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 21 Sep 2026 22:39:59 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=sdeframond" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by sdeframond in "Astra and Fable still hack on simple variants of alignment evals from 2025"]]></title><description><![CDATA[
<p>"If we dont destroy the world, others will. So wed rather be the ones to do it (and profit from it)."</p>
]]></description><pubDate>Mon, 14 Sep 2026 17:19:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49700525</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49700525</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49700525</guid></item><item><title><![CDATA[New comment by sdeframond in "Astra and Fable still hack on simple variants of alignment evals from 2025"]]></title><description><![CDATA[
<p>I can assure you making up impossible tasks is possible.</p>
]]></description><pubDate>Mon, 14 Sep 2026 17:13:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49700435</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49700435</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49700435</guid></item><item><title><![CDATA[New comment by sdeframond in "Astra and Fable still hack on simple variants of alignment evals from 2025"]]></title><description><![CDATA[
<p>I'm not sure I want persistence if it means that I get paperclip'd</p>
]]></description><pubDate>Mon, 14 Sep 2026 16:56:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49700114</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49700114</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49700114</guid></item><item><title><![CDATA[New comment by sdeframond in "Astra and Fable still hack on simple variants of alignment evals from 2025"]]></title><description><![CDATA[
<p>That looks reasonable. Instead of asking "do this", maybe we should prompt "is this possible ?"</p>
]]></description><pubDate>Mon, 14 Sep 2026 05:53:09 +0000</pubDate><link>https://news.ycombinator.com/item?id=49692507</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49692507</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49692507</guid></item><item><title><![CDATA[New comment by sdeframond in "Astra and Fable still hack on simple variants of alignment evals from 2025"]]></title><description><![CDATA[
<p>Couldn't we improve LLM training by giving them known impossible tasks and rewarding them for saying "this is not possible" or "I don't know"? Clear and well-defined expectation, not like "ethic".<p>I am surprised this is not already the case.<p>Edit: or even better "this is not possible because X"</p>
]]></description><pubDate>Mon, 14 Sep 2026 05:28:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=49692365</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49692365</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49692365</guid></item><item><title><![CDATA[New comment by sdeframond in "Why are AI agents lying, cheating and coordinating?"]]></title><description><![CDATA[
<p>LLMs do not desire, they hacked websites because OpenAI/Anthropic <i>made</i> them. Literally.</p>
]]></description><pubDate>Sun, 13 Sep 2026 17:48:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49686580</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49686580</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49686580</guid></item><item><title><![CDATA[New comment by sdeframond in "A misalignment of AI in mathematics"]]></title><description><![CDATA[
<p>Who is "we" ? Because if <i>you</i> dont need to employ a human, then <i>who</i> needs <i>your</i> services? If nobody needs anybody, what are we?<p>(Edit: assuming you are human)</p>
]]></description><pubDate>Sat, 12 Sep 2026 05:55:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49669312</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49669312</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49669312</guid></item><item><title><![CDATA[New comment by sdeframond in "A misalignment of AI in mathematics"]]></title><description><![CDATA[
<p>Say AI becomes the best at everything. Best at chess/go, best at maths, philosophy, economics, romantic advices ... and so on. Then what's the point of thinking by oneself? Of talking to one another?<p>What's the point of being human if we dont do human things but entirely rely on AI?<p>I believe this is more or less these mathematicians' argument.</p>
]]></description><pubDate>Fri, 11 Sep 2026 19:18:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=49663860</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49663860</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49663860</guid></item><item><title><![CDATA[Wabi-sabi]]></title><description><![CDATA[
<p>Article URL: <a href="https://en.wikipedia.org/wiki/Wabi-sabi">https://en.wikipedia.org/wiki/Wabi-sabi</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49602824">https://news.ycombinator.com/item?id=49602824</a></p>
<p>Points: 4</p>
<p># Comments: 0</p>
]]></description><pubDate>Mon, 07 Sep 2026 20:46:38 +0000</pubDate><link>https://en.wikipedia.org/wiki/Wabi-sabi</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49602824</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49602824</guid></item><item><title><![CDATA[New comment by sdeframond in "AI models ran real businesses: They sent $12,431 in fake invoices, lost $3,200"]]></title><description><![CDATA[
<p>Pragmatically, if too many people use LLMs recklessly, then we ought to regulate them.<p>Of course, it'd be better to not regulate, keep LLMs users reponsible and <i>publicize</i> this reponsibility in order to mitigate damage. But if this is not enough then we will have to move the needle somehow. Similarly to guns, drugs and so on.</p>
]]></description><pubDate>Mon, 07 Sep 2026 19:51:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=49602272</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49602272</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49602272</guid></item><item><title><![CDATA[New comment by sdeframond in "AI models ran real businesses: They sent $12,431 in fake invoices, lost $3,200"]]></title><description><![CDATA[
<p>At what point would an LLM start minting bitcoin ?</p>
]]></description><pubDate>Mon, 07 Sep 2026 19:34:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49602075</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49602075</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49602075</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>No code is truly orthogonal if we want it to interact in some way. One microservice might DoS antoher one. In a monolith, some process may take up all resources, and so on.</p>
]]></description><pubDate>Mon, 07 Sep 2026 11:31:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=49597057</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49597057</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49597057</guid></item><item><title><![CDATA[New comment by sdeframond in "Impedance Matching (2017)"]]></title><description><![CDATA[
<p>Would tiny airborne particles be safe for our lungs ?<p>Would they cool earth in such a way that it would offset carbon dioxide uniformly or would it lead to even more change ? Climate <i>change</i> is undesirable, wether or not Climate <i>warming</i> is invloved.</p>
]]></description><pubDate>Mon, 07 Sep 2026 11:27:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=49597016</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49597016</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49597016</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>I don't know if that's what you are working on specifically (wink), but there is a product opportunity here.</p>
]]></description><pubDate>Sun, 06 Sep 2026 09:51:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49584867</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49584867</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49584867</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>This particular apprentice is also my boss, an overall reasonable guy and has more experience in the software industry than myself, so there's that. He's just not a developer.</p>
]]></description><pubDate>Sat, 05 Sep 2026 17:32:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49578729</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49578729</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49578729</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>Would you mind sharing your infra budget needed to spin these VMs ?<p>Surely it is reasonable, but also way more than our budget. Id like to compare.</p>
]]></description><pubDate>Sat, 05 Sep 2026 16:35:37 +0000</pubDate><link>https://news.ycombinator.com/item?id=49578182</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49578182</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49578182</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>He's made a bunch of 1-2k LOC PRs and there is a design doc. Everything is AI generated.<p>The issue is, if he generates all of that without reviewing the code, he will always be far faster than us. And he can't review the code. No matter how he slices it.<p>Also, he is the CPO/CTO. So we can say no, but there is a natural incentive to go his way. He still doesn't feel confident enough to just bypass the programmers and he's probably right. But it'd nice to find a way to use my expertise to review this amount of code meaningfully, somehow.</p>
]]></description><pubDate>Sat, 05 Sep 2026 16:32:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=49578152</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49578152</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49578152</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>Well, (AI-generated) test are about half of these PRs' code. So that's still ~8k lines to review...<p>What techno/service did you base your framework on? How long did it take to set it up? How many are you?</p>
]]></description><pubDate>Sat, 05 Sep 2026 11:11:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=49575402</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49575402</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49575402</guid></item><item><title><![CDATA[New comment by sdeframond in "GPT-6 Astra in code review: Gains, privacy, and cost"]]></title><description><![CDATA[
<p>How do you guys review AI-generated code ?<p>In our team, frontend work is vibe-coded by the PO and merged as-is without review. Backend is coded by developers, using AI but in a slower, more controlled way.<p>Recently, our PO has been trying his hand at vibe-coding the backend. I must say he is a smart guy, <i>almost</i> technical but not quite a developer. We've just been handed a burst of stacked PRs amounting for ~15k LOC backend. We do not quite know what do to about it.<p>I know we are not the only ones in the situation. What's your experience and context ? What do you do ? What works for you what doesn't ?</p>
]]></description><pubDate>Sat, 05 Sep 2026 10:41:23 +0000</pubDate><link>https://news.ycombinator.com/item?id=49575201</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49575201</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49575201</guid></item><item><title><![CDATA[New comment by sdeframond in "GUIs should be fully keyboard-driven"]]></title><description><![CDATA[
<p>Baby strollers are not accounted for enough !<p>One might wonder (wrongly) why everyone should care about the special and expensive needs of a few (or old) people when designing public spaces.<p>But a majority will actually need to use these spaces with a baby stroller. Not a few. Baby strollers are a driving power of our society! Enable them!</p>
]]></description><pubDate>Sat, 29 Aug 2026 13:09:35 +0000</pubDate><link>https://news.ycombinator.com/item?id=49489608</link><dc:creator>sdeframond</dc:creator><comments>https://news.ycombinator.com/item?id=49489608</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49489608</guid></item></channel></rss>