Hacker News: dimmuborgir

Interwhen: A Generalizable Framework for Verifiable Reasoning

dimmuborgir — Sat, 21 Mar 2026 07:30:28 +0000

Article URL: https://arxiv.org/abs/2602.11202

Comments URL: https://news.ycombinator.com/item?id=47464819

Points: 2

# Comments: 0

New comment by dimmuborgir in "AI gets more 'meh' as you get to know it better"

dimmuborgir — Wed, 08 Oct 2025 21:11:01 +0000

Nano Banana for me. After the initial wow phase it's meh now. Randomly refuses to adhere to the prompt. Randomly makes unexpected changes. Randomly triggers censorship filter. Randomly returns the image as is without making any changes.

New comment by dimmuborgir in "The Darwin Gödel Machine: AI that improves itself by rewriting its own code"

dimmuborgir — Fri, 30 May 2025 17:47:37 +0000

From the paper:

"A single run of the DGM on SWE-bench...takes about 2 weeks and incurs significant API costs." ($22,000)

New comment by dimmuborgir in "Firing programmers for AI is a mistake"

dimmuborgir — Wed, 12 Feb 2025 11:56:47 +0000

Yup, they are called GCCs (Global Capability Centers).

The Premise Trap

dimmuborgir — Thu, 09 Jan 2025 11:59:18 +0000

Article URL: https://world.hey.com/dhh/the-premise-trap-924b8cd9

Comments URL: https://news.ycombinator.com/item?id=42644385

Points: 3

# Comments: 0

New comment by dimmuborgir in "O3 "Arc AGI" Postmortem"

dimmuborgir — Sun, 22 Dec 2024 19:23:35 +0000

The result announcement blog post sounded too hypey and gave the impression that he changed his tone but for the last one year he has been consistently saying that LLM+"discreet program search" is the way to go. All the top scoring submissions before o3 had followed the same bruteforce strategy. Even o3 is more or less doing the same bruteforce under the hood, maybe not discreet.

New comment by dimmuborgir in "SumatraPDF Reader"

dimmuborgir — Tue, 24 Oct 2023 07:14:31 +0000

Dark mode finally!!! This was the only bummer for me all these years. Not anymore!

Getting from Generative AI to Trustworthy AI: What LLMs Might Learn from Cyc

dimmuborgir — Thu, 10 Aug 2023 10:56:59 +0000

Article URL: https://arxiv.org/abs/2308.04445

Comments URL: https://news.ycombinator.com/item?id=37074448

Points: 1

# Comments: 0

From word models to world models

dimmuborgir — Fri, 23 Jun 2023 10:57:26 +0000

Article URL: https://arxiv.org/abs/2306.12672

Comments URL: https://news.ycombinator.com/item?id=36445197

Points: 100

# Comments: 113

New comment by dimmuborgir in "How much of AI's recent success is due to the Forer Effect?"

dimmuborgir — Fri, 26 May 2023 18:44:07 +0000

We can extend this phenomenon to all of knowledge. You ask ChatGPT something and it gives you some very generic 10-point blogspam-esque answer. You immediately think ChatGPT is intelligent and that it has understood your particular question and that it has given you a tailor-made answer.

New comment by dimmuborgir in "AI's Jurassic Park Moment"

dimmuborgir — Wed, 21 Dec 2022 06:52:44 +0000

His book "The Algebraic Mind" goes into great detail about connectionism (neural networks), symbolic systems, limits of connectionism and proposals to integrate neural networks with symbolic systems (hybrid systems).

The "deep learning alone is enough" camp (especially LeCun) has abused him for years but now slowly coming to the realization that we need to feed neural networks with explicit inductive biases to attain AGI which is exactly what Marcus has been saying since the 90s. LeCun, for some reason, refuses to call these explicit biases as symbols and that's the only disagreement between Marcus and LeCun these days.

New comment by dimmuborgir in "A few thoughts on Ticketmaster and Taylor Swift"

dimmuborgir — Sat, 19 Nov 2022 16:44:46 +0000

I think her target audience is extremely broad. From teenagers to middle-aged people. Plus her latest album was marketed aggressively.

New comment by dimmuborgir in "AI language models are struggling to “get” math"

dimmuborgir — Wed, 12 Oct 2022 16:24:13 +0000

Those models are not trained on short loops. They are trained on whole songs just like image generation models are trained on whole images. And yet they struggle to repeat sections, modulate to a different key, create bridges, intros and outros. After a few seconds of hallucinating a melodic line they simply abandon the idea and migrate to another one. There is no global structure whatsoever.

New comment by dimmuborgir in "AI language models are struggling to “get” math"

dimmuborgir — Wed, 12 Oct 2022 15:29:31 +0000

AI is bad at music also. Even the state of the art transformer models can't produce more than a few seconds of coherent melodic phrases.

New comment by dimmuborgir in "AudioLM: A Language Modeling Approach to Audio Generation"

dimmuborgir — Thu, 15 Sep 2022 04:42:36 +0000

This one suffers from the same problem that previous audio generation methods had. It correctly mimics the piano timbre but there is no global structure in the generated melodic lines. There is some style imitation but no melodic coherence.

New comment by dimmuborgir in "DALL-E Chess in Jungle and Dunes"

dimmuborgir — Sun, 31 Jul 2022 14:32:40 +0000

Lots of weird artifacts which are very hard to fix.

The argument that users can now generate professional grade art by bypassing artists entirely feels so strange. I have access to Dall-E. To generate images without artifacts, you have to do one of these: a) Do a lot of cherry-picking which can be expensive. b) Prompt should be about an abstract concept which can "tolerate" any number of artifacts. c) Prompt should be about a common/generic concept that you have already seen a lot of times on the internet.

I think the biggest use case of Dall-E will be in removing creative block for artists.

New comment by dimmuborgir in "JSPatcher, a web-based visual editor to build JavaScript and WebAudio apps"

dimmuborgir — Thu, 08 Jul 2021 06:51:19 +0000

JSPatcher is similar to Max/MSP. Even the UI design is a blatant ripoff.

There are already full blown online DAWs like BandLab, Soundation, Amped Studio etc.

New comment by dimmuborgir in "A half-hour to learn Rust"

dimmuborgir — Sat, 02 Jan 2021 07:23:58 +0000

Previous discussion: https://news.ycombinator.com/item?id=22448933

New comment by dimmuborgir in "My Procedurally Generated Music Is Awful"

dimmuborgir — Mon, 07 Dec 2020 08:33:40 +0000

Interesting. Is there anywhere I can listen to the music generated by your algorithm?

New comment by dimmuborgir in "Ask HN: Top Coursera Courses?"

dimmuborgir — Sun, 29 Nov 2020 16:12:09 +0000

Audio Signal Processing for Music Applications

Professor Xavier Serra[1] is a highly respected veteran in the field.

[1] https://en.m.wikipedia.org/wiki/Xavier_Serra