mortsnort

6 ms·

61.
by mortsnort 1y ago
And yet thousands of people have liked (and presumably read) his long article and hundreds have replied. Substack provides levels of information sharing that a book can't provide. With books, some small amount of people read difficult
62.
by mortsnort 1y ago
You can write Google cloud functions to disable your credit card when certain thresholds are met pretty easily, but it's unethical that this isn't just a toggle somewhere in settings.
63.
by mortsnort 1y ago
It's not a binary choice though. We have the technology for an OS to enable sideloading for people who want it and disable it for people who don't...
64.
by mortsnort 1y ago
It's a race to see which runs out of steam first: AI investment or Ed Zitron's schtick.
65.
by mortsnort 1y ago
Fair but the Gemini "flash" branding implies it's their model for speed/scale in my mind.
66.
by mortsnort 1y ago
At $0.02 per image, it's prohibitively expensive for many use-cases. For comparison, the cheapest Flux model (Schnell) is $0.003 per image.
67.
by mortsnort 1y ago
Similarly spec'd laptops can be found for ~1/2 the price. I don't think a deep economic analysis is necessary...
68.
by mortsnort 1y ago
The computer you listed is specifically designed for running local AI inference, because it has an APU with lots of integrated RAM. If that isn't your use case then AMD 9000 series should be better.
69.
by mortsnort 1y ago
It seems like a lot of "agentic features" these days don't really require LLMs. The dinner reservation agent is basically just Kayak for dinner reservations. They're even working with the reservation vendors to ensure co
70.
by mortsnort 1y ago
All of the foundational models are converging on the same performance. For most use cases you can swap Anthropic, OpenAI, and Google's top models interchangeably. Barring some major methodology breakthrough, the next decade for AI will
71.
by mortsnort 1y ago
Does the author really feel alone? There's posts like this on the front page of HN everyday. If you don't want AI to be useful for coding then you can find a way to convince yourself it isn't. Surely, at the bare minimum, yo
72.
by mortsnort 1y ago
While I agree SEO and click-based ad revenue have degraded the media greatly, we should still be very concerned about the potential collapse of journalism. Proper journalism requires large budgets, trained journalists and things like report
73.
by mortsnort 1y ago
It's a blog post about whether GPT 5 lived up to the hype and how it is being received, which is a totally legitimate thing to blog about. This is Gary Marcus's blog, not BBC coverage, of course it's slanted to the opinion he
74.
by mortsnort 1y ago
There are lots more applications than coding and Open Router hosting for open weight models that I'd guess just got completely changed by this being an Apache license. Think about products like DataBricks that allow enterprise to use
75.
by mortsnort 1y ago
Releasing this under the Apache license is a shot at competitors that want to license their models on Open Router and enterprise. It eliminates any reason to use an inferior Meta or Chinese model that costs money to license, thus there are
76.
by mortsnort 1y ago
I believe it's roughly the same size as the model files. If you look in the transformers folder you can see there are around 9 5gb files, so I would expect you need ~45gb vram on your GPU. Usually quantized versions of models are eve
77.
by mortsnort 1y ago
You can also vibe tackle tech debt tho. AI is pretty good at deciphering code for you when you need to jump into it.
78.
by mortsnort 1y ago
I get the MS Copilot hate, but Perplexity, Dia, Open AI and others are entering the AI browser space. I don't think Google and MS can just sit it out and hope AI browsers don't catch on. At this point most of us have a ChatGPT tab
79.
by mortsnort 1y ago
Convincing hipsters to switch their urban hives to more obscure bees actually seems achievable.
80.
by mortsnort 1y ago
It says they can be "fine tuned," but it looks like the agents are all using the same model with different system prompts? This would be more intriguing if they trained a debugger model from the ground up that could be used for th
81.
by mortsnort 1y ago
Big update, they removed _preview from the model name.
82.
by mortsnort 1y ago
Chrome has it, but it's hidden in menus. Android has it as an OS feature with a really unintuitive interface. It also fails the first time you try to read any page, but works the second time... It's clearly not something they w
83.
by mortsnort 1y ago
Where is popular science videos coming from? Blogs, Wikipedia, research papers, Substacks, newspapers, magazines, ChatGPT, etc.
84.
by mortsnort 1y ago
Tik Tok is obvious brain rot, but what if one's time at the dopamine carnival is spent consuming "brain-growth" content? Phones essentially put all human knowledge at our fingertips, where is the line of diminishing or negati
85.
by mortsnort 1y ago
S&P 500 companies don't get any inflow of cash when their stock is purchased
86.
by mortsnort 1y ago
Something like this may kill games before they're even made, which is even worse. Barriers to entry don't seem like a good idea to me.
87.
by mortsnort 1y ago
Kendall the blog link at the end for semantic parsing gives a 404 error.
88.
by mortsnort 1y ago
I know tech people like to villainize bean counters for ruining tech companies, but this man has zero business sense and needs a good bean counter. It's crazy that he thinks people will read this and feel like he was in the right with
89.
by mortsnort 1y ago
Apparently all of the buildings made with the excavator are also not valid evidence of its utility.
90.
by mortsnort 1y ago
It's just a reskin of Flux Kontext. You can pay $.04 per image on Flux website and have full control over the prompt.

More ›