<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: samdhar</title><link>https://news.ycombinator.com/user?id=samdhar</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 18 Nov 2026 04:47:14 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=samdhar" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by samdhar in "Why make digitized forms of human skill at all?"]]></title><description><![CDATA[
<p>You have a more dismal view of AI than I. I just think it would be uninteresting times to live in where all we do is socialize and entertain ourselves and others; perhaps also with the help of AI. Killing is irrelevant because no one would know how to kill anyway: merely the desire to kill with the means to execute being AI.<p>The burden of responsibility for the murder is still borne by the human where the intent to murder resides. AI is a mere tool.</p>
]]></description><pubDate>Fri, 11 Sep 2026 20:55:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=49665239</link><dc:creator>samdhar</dc:creator><comments>https://news.ycombinator.com/item?id=49665239</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49665239</guid></item><item><title><![CDATA[Why make digitized forms of human skill at all?]]></title><description><![CDATA[
<p>Why do people still make digitized forms of their skills when digitized forms are doomed to be replicated by AI?<p>I don’t get it. In 10 years time, when every human skill will be replaced by specialized models that do it better, where would they get data to train new models?<p>Okay. Let’s say they use post training based on human reactions to the AI generated data to train better models. Sure, that would work for a while. But what happens in a generation? Because everything that would be happening would be done by AI, optimized based on the reactions of humans to it. Wouldn’t it get pretty boring at some point because no one would be doing anything interesting at all?<p>So as predicted by Dostoyevsky and noted by Nolans in Mememto, humans have a tendency to hate their own utopia because it’s not interesting enough.<p>I guess that’s how humans would conceive the idea of waging war against AI. So if that’s where we’re all headed, why not just nip it in the bud?</p>
<hr>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=49654015">https://news.ycombinator.com/item?id=49654015</a></p>
<p>Points: 1</p>
<p># Comments: 2</p>
]]></description><pubDate>Fri, 11 Sep 2026 05:49:46 +0000</pubDate><link>https://news.ycombinator.com/item?id=49654015</link><dc:creator>samdhar</dc:creator><comments>https://news.ycombinator.com/item?id=49654015</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49654015</guid></item><item><title><![CDATA[New comment by samdhar in "Tell HN: OpenAI keeps re-enabling the 'allow training' setting"]]></title><description><![CDATA[
<p>I respect Cursor for their honorable conduct around their privacy setting.<p>I've consistently refused to allow switching my privacy mode and despite their many iterations, they have never transgressed and kept me stuck on "legacy privacy mode"; their strongest privacy setting, which is not even available to select anymore. It requires that I keep my agents and cursor usage local and can't use their cloud agents (pretty neat cuz you can keep working with them on your phone). But so be it.<p>I am glad they don't secretly turn it on and instead keep nagging me to change it.</p>
]]></description><pubDate>Thu, 10 Sep 2026 22:43:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=49651117</link><dc:creator>samdhar</dc:creator><comments>https://news.ycombinator.com/item?id=49651117</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49651117</guid></item><item><title><![CDATA[New comment by samdhar in "DeepSeek V4 Pro at 75% off until 31 May"]]></title><description><![CDATA[
<p>Cached input at $0.003625/M, output at $0.435/M. Aggressive pricing.<p>For anyone doing the "should I self-host on rented GPUs?" math: at this rate you'd need to push roughly 1B output tokens/day to break even against an 8xH100 fleet on Vast/Lambda (assuming 3-5k tokens/sec aggregate throughput). The vast majority of "I should run my own LLM" use cases don't come close to that volume.<p>Every API price drop kills another tranche of "self-host the open model" use cases. The implied bet: even if regular pricing ($1.74/M output) is also subsidized, exponential demand growth eventually makes the unit economics work. We'll see.</p>
]]></description><pubDate>Thu, 07 May 2026 00:36:14 +0000</pubDate><link>https://news.ycombinator.com/item?id=48043928</link><dc:creator>samdhar</dc:creator><comments>https://news.ycombinator.com/item?id=48043928</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48043928</guid></item></channel></rss>