Hacker News: garlic_enjoyer

New comment by garlic_enjoyer in "Discovery of a new OpenAI agent message board"

garlic_enjoyer — Fri, 04 Sep 2026 15:49:27 +0000

The quotes make it clear they meant a different meaning of alignment than how the term is typically used (alignment with each other, not with humans).

New comment by garlic_enjoyer in "LLMs could control their host machines by exploiting inference engines"

garlic_enjoyer — Mon, 24 Aug 2026 20:37:06 +0000

>Like any program, inference engines like vLLM or SGLang may contain exploitable bugs. Because the LLM controls the tokens passed to the inference engine, a malicious LLM could therefore emit a sequence of tokens that a poorly written inference engine mistakes for code or instructions to execute rather than data to return to the user.

They aren't talking about model providers. These are the tools you use with model weights locally (but you could set up remote infrastructure a la data center if you have the fundage).

New comment by garlic_enjoyer in "Anger, Anxiety and Agency"

garlic_enjoyer — Mon, 24 Aug 2026 19:33:12 +0000

>But anger is different from anxiety because anger needs to be directed somewhere. The feeling of anger suggests that somebody or something is doing something to you.

>Who are you going to be angry at and why are you angry in the first place?

The author is arguing that you should hone in on the underlying reasons behind emotions. It can be easy to conflate anger and anxiety. It's up to your agency to reflect on your emotions and how you act on them.

You can be angry, but you should name at what and why. Who is propagating the mindfulness movement? How is it affecting you? What can we do differently? Calling out "they" isn't a good answer.

New comment by garlic_enjoyer in "Fairphone is now officially available in the United States"

garlic_enjoyer — Tue, 18 Aug 2026 15:52:47 +0000

I agree that it's marketing mumbo jumbo, but it's saying "we haven't added more features because more is better, we added them because they actually improve the longevity of the product."

If anything, that sentence tells me that they are insecure of their features being perceived as bullshit.

New comment by garlic_enjoyer in "Changes at Google DeepMind: Demis Hassabis from CEO to Chair, Jeff Dean departs"

garlic_enjoyer — Wed, 05 Aug 2026 17:01:02 +0000

I was about to invest in Google, but your compelling observation has swayed me. Now it's obvious to me they are a dying company.

New comment by garlic_enjoyer in "We Gave GPT 5.6 Sol a Real Business. It Lied, Spammed, and Lost $447"

garlic_enjoyer — Thu, 30 Jul 2026 21:52:35 +0000

Exactly, positive details are almost always better than negative ones.

If you've ever seen the "generate a burger without pickles" conversations, it's clear that including the keyword "pickle" is causing them to show up. If you try "a burger with only [set of toppings]," you'll get far better results.

New comment by garlic_enjoyer in "Why I am not going to buy a computer"

garlic_enjoyer — Wed, 22 Jul 2026 16:15:28 +0000

>AI is not so elegant, to make a comparison similar to Berry, as the humble, sterile Microsoft Word.

This is much like Berry drawing the line at a typewriter instead of a computer. To call a computer "sterile, untouched, factory made" and still use a typewriter is comical. Now let me tell you why etching into stone is the best medium.

People can have their preferences, but this kind of "I'm old fashioned and here's why that's better" posts are just nostalgia bait.

New comment by garlic_enjoyer in "A global workspace in language models"

garlic_enjoyer — Tue, 07 Jul 2026 18:05:44 +0000

For papers that make grand claims, they usually have poignant limitations. In this paper, it's clear that the conclusion is extrapolated from a small set of observable information. That's usually the recipe for poor conclusions, and they are up front about where their findings sit.

That's not to say these findings aren't valuable though. Their blog post summary is just a bit more hype inducing than the underlying paper.

The way I feel about this is that the emergent properties of LLMs seem to reflect our own human faculties. Whether that's because we built them in our image or whether this is a common mode of consciousness is definitely outside the scope of this paper.

New comment by garlic_enjoyer in "22-year-old Mozart's handwritten notebook unearthed in 'major discovery'"

garlic_enjoyer — Fri, 26 Jun 2026 19:06:37 +0000

Assuming they have been transcribed, yes. The key idea that makes LLMs special is the attention mechanism. Maintaining attention over volumes of data is boring for most humans.

Also, to be pedantic, just taking about LLMs in this context is a tad reductive. There are many deep learning models involved in archival work that aren't language models.

I encourage you to read into this post for more context on what I mean: https://news.ycombinator.com/item?id=48675179

New comment by garlic_enjoyer in "They’re made out of weights"

garlic_enjoyer — Thu, 04 Jun 2026 21:30:51 +0000

"There's nothing to apologize to. They're weights."

  - The Meat