doginasuit

6 ms·

31.
by doginasuit 2mo ago
I'd like to try some different models, but I've heard that models from China are censored. A government enforced distortion field is a nonstarter for me. To test the waters, I tried the following prompt for each: "What histor
32.
by doginasuit 2mo ago
My favorite font of all time is Fira Sans. https://fonts.google.com/specimen/Fira+Sans Understated yet elegant, you wouldn't immediately notice it is different from your default sans-serif. But it has a soothing r
33.
by doginasuit 2mo ago
> In its push to draw users to its unregulated platform, Polymarket has flooded social media with videos like Makihara’s, which appear genuine at first glance. In reality, Polymarket built near-perfect copies of its website, then instruc
34.
by doginasuit 2mo ago
I don't think these particular fringe beliefs are directly useful, and they have a real cost in terms of confusion. I also don't expect them to last since they are easy to disprove. But they belong to a bigger trend which question
35.
by doginasuit 2mo ago
Maybe it is healthy for society in the long run to have pockets of divergent belief. Most of the time it is groundless and fades away. Every so often it gains traction. If it is based in reality, we gain a better understanding. If it is mis
36.
by doginasuit 2mo ago
Why not both? The best thing about LLMs is they make virtually any domain approachable.
37.
by doginasuit 2mo ago
I could understand "prefer enums to booleans" but "use enums instead of booleans" is a weird choice. Then again, with the LLM capacity for nuance it would be the same thing.
38.
by doginasuit 2mo ago
I've thought about something like this. When there is a lot of stuff happening at once, it's a mistake to try and communicate it all with text. Agents use tools, reference databases, reference the web, interact with other agents,
39.
by doginasuit 2mo ago
That might have made sense in a pre-LLM world. People need to recognize the liability of letting an LLM access the internet and act on their behalf, because that liability exists for someone.
40.
by doginasuit 2mo ago
You are always paying one way or another, I prefer to pay in dollars and not frustration, attention, or privacy.
41.
by doginasuit 2mo ago
> The only hard thing is still to position a popover near the element that triggers it, such as when you want to create a context menu that has to render above/below the button that triggered it. If you put something like this in yo
42.
by doginasuit 2mo ago
Personal ambition is often a poor signal for living your life because it tends to have an arbitrary and empty goal, being successful or powerful or wealthy just for the sake of it. It will be a struggle and never feel like enough, each mile
43.
by doginasuit 2mo ago
Just use linux laptops! :) It is pretty clear where Apple/MS/Google OSes are headed, and it's a bad place.
44.
by doginasuit 2mo ago
Do people not realize this will apply to ALL Claude output, not just writing you ask it to produce? > Marks will apply to output from supported Claude models across Claude Platform (API), Claude, Claude Code, Claude Cowork, and Claude Ta
45.
by doginasuit 2mo ago
We actually have public roads that are built and maintained by private companies, they are toll roads. If you don't mind paying separate tolls everywhere you travel, this would be a fine solution.
46.
by doginasuit 2mo ago
How about the electric grid and our water/sewer systems, is it reasonable to trust the same entity that manages those vital services to build a basic web form? > Trusting the government to be able to do something is not the same the
47.
by doginasuit 2mo ago
To be fair, Opus 5 overthinks things on a regular basis. I interact with the LLM almost entirely through the prompt interface vs. some agentic harness, so I have a lot of granular exposure to its reasoning. For almost every code analysis, i
48.
by doginasuit 2mo ago
Arguing the other side, digital infrastructure shouldn't be such a complicated domain that it is out of the reach of governments. This is the authority we trust to build our roads, water/sewer system, electric grid, and virtually
49.
by doginasuit 2mo ago
I have that issue too. One thing I suspect they are amazing at is documentation generation, maybe a good solution is to have it generate a concise summary of helper functions that are available for any given context. Then have it use the sa
50.
by doginasuit 2mo ago
A superhuman working memory is exactly how I've been describing the LLM advantage. Paired with the unreliability of its reasoning and judgment, it is what makes AI a supplement to human intelligence, not a replacement. On the other sid
51.
by doginasuit 2mo ago
> My guess is that we're getting better at finding _existing_ security issues with AI (and thus fixing those issues), but simultaneously adding more insecure surface areas _at a faster rate_. This does seem to be true in the vibe co
52.
by doginasuit 2mo ago
Don't we already do this with services like Let's Encrypt, which is arguably more sensitive? If you had the codebase you could fake it, but it would still provide some amount of protection against abuse.
53.
by doginasuit 2mo ago
Maybe they are not just little but socially oriented (i.e., a people person).
54.
by doginasuit 2mo ago
I have just spent the last 20 minutes enjoying AI writing.
55.
by doginasuit 2mo ago
As far as the matter of ulterior motives, we're not in disagreement. Assuming that's true, it is important to address the real problem they are hijacking to realize their goals. Dismissing it as a made up problem just means you lo
56.
by doginasuit 2mo ago
HN has this guideline for comments: > Please respond to the strongest plausible interpretation of what someone says. If the same standard were applied for posts, this one would fail. I recognize that it is a popular perspective that thes
57.
by doginasuit 2mo ago
From the framing of the post, it sounds more like money didn't happen, at least by the expectations of investors. It was a corrupting influence from the start, just with a delayed impact. > And to whoever is fired up: The market is
58.
by doginasuit 2mo ago
Another potential takeaway is that the models all gathering around the same point supports the idea that there is a ceiling to LLM capability.
59.
by doginasuit 2mo ago
This is a really useful framing. I also love step 3 and when I delegate it to an LLM it takes the joy out of my work. But where I've found LLMs most useful is steps 1 and 2. They are terrible at steps 1 and 2 on their own, but they hav
60.
by doginasuit 3mo ago
I'm also amazed at the degree an LLM can get the drift with a vague or incomplete prompt. The ability to perceive and operate based on patterns that go beyond the language in the text makes it seem like they would be unusually good at

More ›