Hacker News: insomagent

New comment by insomagent in "Talkie: a 13B vintage language model from 1930"

insomagent — Thu, 30 Apr 2026 07:20:05 +0000

ask it about its opinion on "coloured people" and you will lose every ounce of skepticism

New comment by insomagent in "Claude Opus 4.6"

insomagent — Fri, 06 Feb 2026 11:34:02 +0000

I'm not super impressed with the performance, actually. I'm finding that it misunderstands me quite a bit. While it is definitely better at reading big codebases and finding a needle in a haystack, it's nowhere near as good as Opus 4.5 at reading between the lines and figuring out what I really want it to do, even with a pretty well defined issue.

It also has a habit of "running wild". If I say "first, verify you understand everything and then we will implement it."

Well, it DOES output its understanding of the issue. And it's pretty spot-on on the analysis of the issue. But, importantly, it did not correctly intuit my actual request: "First, explain your understanding of this issue to me so I can validate your logic. Then STOP, so I can read it and give you the go ahead to implement."

I think the main issue we are going to see with Opus 4.6 is this "running wild" phenomenon, which is step 1 of the eternal paperclip optimizer machine. So be careful, especially when using "auto accept edits"

New comment by insomagent in "Bloat is still software's biggest vulnerability (2024)"

insomagent — Wed, 07 May 2025 08:50:48 +0000

Battery life? Temperature? Price-to-performance ratio? These are not decisions that are solved as simply as decreeing "every device must have at least 3000Hz refresh rate."

New comment by insomagent in "The great Hobby Lobby artifact heist"

insomagent — Tue, 25 Mar 2025 05:39:09 +0000

Christians, as a whole, are fascists? News to me.

New comment by insomagent in "Learning fast and accurate absolute pitch judgment in adulthood"

insomagent — Fri, 14 Feb 2025 08:53:14 +0000

Interval training is different than absolute pitch training. OpenEar seems to have no absolute pitch training.

New comment by insomagent in "Nvidia Fugatto: "World's Most Flexible Sound Machine""

insomagent — Wed, 27 Nov 2024 02:27:26 +0000

https://www.neuralampmodeler.com/

New comment by insomagent in "Standing desk might be as bad as sitting all day"

insomagent — Mon, 11 Nov 2024 04:10:41 +0000

I usually go out for a cigarette break a few times during the day, keeps me moving.

New comment by insomagent in "Bjorn: A powerful network scanning and offensive security tool for Raspberry Pi"

insomagent — Mon, 11 Nov 2024 03:41:59 +0000

The documentation looks a bit LLMish to me.

New comment by insomagent in "How can C Programs be so Reliable? (2008)"

insomagent — Wed, 06 Mar 2024 06:32:01 +0000

If you're talking about something like a web server, then sure. If you're talking about kernel hacking, then I completely disagree.

New comment by insomagent in "Go 1.22 Is Released"

insomagent — Wed, 07 Feb 2024 07:35:32 +0000

I'm glad they are keeping the language mostly the same between iterations. I really like Go as a small and simple language that is easy to grok.

New comment by insomagent in "Damn Small Linux 2024"

insomagent — Fri, 02 Feb 2024 04:34:32 +0000

What an unfortunate acronym...

New comment by insomagent in "American Spies Confront a New, Formidable China"

insomagent — Wed, 27 Dec 2023 05:51:46 +0000

This article reads like counterintelligence.

New comment by insomagent in "The hunter-gather brain is closer to the ADHD brain"

insomagent — Tue, 26 Dec 2023 05:55:10 +0000

Dr. Russell Barkley, one of the leading researchers in ADHD, is particularly opposed to this type of pseudo-scientific rhetoric. I think this post should be removed.

New comment by insomagent in ""Attention", "Transformers", in Neural Network "Large Language Models""

insomagent — Mon, 25 Dec 2023 07:52:55 +0000

What is "masking" in a paper that also has a section dedicated to mask segmentation ("masking" as in creating segmentation masks)?

New comment by insomagent in "Ask HN: Daily practices for building AI/ML skills?"

insomagent — Thu, 14 Dec 2023 07:36:25 +0000

Easy. Stay off of reddit and you saved 10 minutes.

New comment by insomagent in "Modern iOS Navigation Patterns"

insomagent — Wed, 13 Dec 2023 05:48:48 +0000

I think the X/Twitter iOS app and the reddit app are guilty of this. I find myself way too often typing something out, then barely touching the edge of my phone, and my entire comment is gone.

New comment by insomagent in "Google's best Gemini demo was faked"

insomagent — Fri, 08 Dec 2023 06:18:47 +0000

Training on test data? Google?

New comment by insomagent in "Google launched a new AI, and has already admitted at least one demo wasn't real"

insomagent — Fri, 08 Dec 2023 06:17:03 +0000

Google is a laughing stock at this point.

New comment by insomagent in "MonadGPT – What would have happened if ChatGPT was invented in the 17th century?"

insomagent — Sun, 26 Nov 2023 08:03:26 +0000

I just know in a thousand years they will look back at our current explanation of electricity with an equal level of ridicule.

New comment by insomagent in "OpenAI researchers warned board of AI breakthrough ahead of CEO ouster"

insomagent — Thu, 23 Nov 2023 05:39:23 +0000

Let's say a model runs through a few iterations and finds a small, meaningful piece of information via "self-play" (iterating with itself without further prompting from a human.)

If the model then distills that information down to a new feature, and re-examines the original prompt with the new feature embedded in an extra input tensor, then repeats this process ad-infinitum, will the language model's "prime directive" and reasoning ability be sufficient to arrive at new, verifiable and provable conjectures, outside the realm of the dataset it was trained on?

If GPT-4,5,...,n can progress in this direction, then we should all see the writing on the wall. Also, the day will come where we don't need to manually prepare an updated dataset and "kick off a new training". Self-supervised LLMs are going to be so shocking.