Hacker News: awestroke

New comment by awestroke in "Today, we hacked what we'd set out to hack: the iPhone"

awestroke — Fri, 23 Oct 2026 16:31:50 +0000

They found the PHP?

New comment by awestroke in "Show HN: Qwen3-4B with the AI writing tics trained out"

awestroke — Mon, 19 Oct 2026 13:19:27 +0000

Nobody knows what it means, but it's provocative. It gets the people going

New comment by awestroke in "Mistral Large 4"

awestroke — Tue, 06 Oct 2026 20:19:56 +0000

> Do you have any evidence to the contrary?

https://en.wikipedia.org/wiki/Snowden_disclosures

New comment by awestroke in "Sites in ChatGPT"

awestroke — Fri, 02 Oct 2026 19:17:13 +0000

something that solves a real problem for businesses, that you are uniquely positioned to execute on well. If you target the lowest common denominator, or consumers, or build something really quickly - good luck

New comment by awestroke in "iOS 27, iPadOS 27, and macOS 27"

awestroke — Mon, 14 Sep 2026 18:04:53 +0000

New Siri is not available in the EU. And no plans to change that.

I'll probably switch to Android once my phone and watch batteries degrade enough to annoy me

New comment by awestroke in "Benchmark: CadQuery vs. OpenSCAD for agentic CAD work"

awestroke — Sun, 13 Sep 2026 00:10:44 +0000

Replicad is another frontend to the same cad kernel. But can run in browser or via bun or whichever js runtime you want. I've found llms to be better at writing typescript than python code for things like this.

Using Replicad, I've successfully iterated on a 3d printed enclosure for my electronics project. It's fantastic to get everything parameter based and version controlled from revision 1.

New comment by awestroke in "Compute-efficient pretraining and scaling to trillion-parameter models"

awestroke — Thu, 10 Sep 2026 17:40:33 +0000

> What does that get you?

Cheaper model training runs? Ability to scale training to larger model sizes without extending training time?

New comment by awestroke in "Anthropic researcher believes more than 10% chance AI 'could kill all humans'"

awestroke — Wed, 09 Sep 2026 11:30:23 +0000

"The scenario has been explored in fiction" doesn’t mean "everything in fiction will happen." Your reply conflates familiarity with inevitability.

Science fiction is relevant here because it has explored the problem of humans losing control of what they create. Whether that could happen with AI needs to be assessed on its merits. Pointing to other fictional things that haven’t happened neither answers that question nor rebuts the original point.

New comment by awestroke in "The Navier–Stokes Millennium Prize Problem"

awestroke — Wed, 09 Sep 2026 07:13:43 +0000

AI companies use heuristics to filter sessions, then llms to further filter, then use various techniques too anonymize the session, then process it and add it to various datasets for further selection and refinement. they don't need huge storage for this.

New comment by awestroke in "Pigeon, a signed Pass for what a sub-agent may do"

awestroke — Sun, 06 Sep 2026 11:37:01 +0000

A trivial problem solved in a very complex way

New comment by awestroke in "Claude Fable 5.1 and Claude Mythos 5.1"

awestroke — Wed, 02 Sep 2026 12:47:30 +0000

It's only theft when people pay Anthropic for inference in order to improve their own datasets. It's not theft when Anthropic grabbed basically all ebooks and web content on the internet to build their own dataset, without paying anything to anyone

New comment by awestroke in "Mushroom behind 'tiny people' hallucinations identified"

awestroke — Fri, 14 Aug 2026 12:22:45 +0000

Imagine a person who does not want social media companies to shape their reality, yet does not want their perception of reality to be compromised by hallucinations. Or perhaps you would be fine with being snuck LSD on your way to work, or court, or picking up your kids from kindergarten? Most people do not seek out hallucinations

New comment by awestroke in "Accelerating GPT-5.6 Sol Ultrafast"

awestroke — Fri, 14 Aug 2026 09:37:52 +0000

the easier way to do this, with any harness, is to use an expensive main agent that is told to delegate all code reading, writing, exploration, research etc to weaker subagents to conserve tokens. It's an inversion of the pattern but the resulting split is the same

New comment by awestroke in "Mushroom behind 'tiny people' hallucinations identified"

awestroke — Thu, 13 Aug 2026 08:37:58 +0000

The large majority of people want absolutely nothing to do with hallucinations

New comment by awestroke in "OpenAI's new device will be hockey puck-sized and cost over $300"

awestroke — Tue, 11 Aug 2026 20:40:55 +0000

Oh, so starting Siri on my iphone is the exact same experience as having a voice conversation with a SOTA LLM?

New comment by awestroke in "The brain may be about to have its Ozempic moment"

awestroke — Tue, 11 Aug 2026 20:18:08 +0000

Title implies people will flock to a pill that makes them shed gray matter

New comment by awestroke in "OpenAI's new device will be hockey puck-sized and cost over $300"

awestroke — Mon, 10 Aug 2026 17:36:22 +0000

1. You can't just press a button on your phone to instantly start an ai voice conversation

2. You can't bring your smart speakers with you everywhere, also same as 1

3. See 1 and 2

New comment by awestroke in "Flint: A Visualization Language for the AI Era"

awestroke — Sat, 01 Aug 2026 07:20:01 +0000

Different charting backends support different kinds of charts. Being able to easily switch between "ECharts Sunburst" and "Vega-Lite faceted bar" - one of the examples on the project website - seems like a super useful ability.

New comment by awestroke in "Claude Opus 5"

awestroke — Fri, 24 Jul 2026 17:51:24 +0000

Doubleplus benchmaxxed

New comment by awestroke in "HMD Touch 4G"

awestroke — Sun, 19 Jul 2026 17:41:34 +0000

> Currently only available in India

Closes tab