davebren

5 ms·

61.
by davebren 5mo ago
Literally suicidal empathy. Although the numbers would change drastically if it was real, it's easy to virtue signal in hypotheticals.
62.
by davebren 5mo ago
Or choosing blue allows them to feel self-righteous while not actually at any risk since this isn't real. Virtue signalling is the ethic of our day.
63.
by davebren 6mo ago
Somewhat relevant, I learned a mapping of individual characters to colors with a chrome extension I made, so the text size doesn't really matter since words are just sequences of colors. https://chromewebstore.google.com&#x2
64.
by davebren 6mo ago
I think the winner will be local model wrappers that are designed to do specific tasks well like search without being anthropomorphized sycophants.
65.
by davebren 6mo ago
It would be a good benchmark for humanoid robots
66.
by davebren 6mo ago
The ball trajectory gives the spin
67.
by davebren 6mo ago
It could already copy the art styles from its training data, what is the advancement here?
68.
by davebren 6mo ago
Learn how to draw simple circles and arrows, this is the epitome of learned helplessness.
69.
by davebren 6mo ago
It turns out that effort matters
70.
by davebren 6mo ago
The point of a diagram is that you have something in your head to turn into the diagram. There's no point if you can't do it yourself and the image generator is coming up with it for you.
71.
by davebren 6mo ago
> Preferring human work will be luxury status-signalling just like it is for clothing, food, etc. What? Those items are luxuries when made by humans because they are physical goods where every single item comes with a production and dist
72.
by davebren 6mo ago
The faces...that's nice that it turned a kid's book into an abomination
73.
by davebren 6mo ago
That's interesting, I hadn't heard of this. I guess IDEs kind of took some elements from that but with more flexibility.
74.
by davebren 6mo ago
At the same time treating everything as tokens and next word prediction will never produce any real understanding like what humans do when they learn how to program. The bitter lesson is an admission that we still have no clue what is at th
75.
by davebren 6mo ago
Getting back to a functional search engine is the most interesting part of this technology to me. Something that just gives links to the most relevant pages without a bunch a LLM editorializing on top of it. But do current LLMs solve that,
76.
by davebren 6mo ago
Are businesses all running on sunk cost fallacy now? These findings have been coming out for a while but it doesn't seem to change anything.
77.
by davebren 6mo ago
You copied two human coded native apps into a vibe coded react app? If the vibe coding is so good why wouldn't you keep the native apps and vibe code on top of them instead of spending a bunch of money to reach feature parity with a wo
78.
by davebren 6mo ago
If data is collected and stored on a server, that is what we mean when we say data collection.
79.
by davebren 6mo ago
Remember, every product they release expands the scope of their non-compete clause, and they like their lawsuits.
80.
by davebren 6mo ago
No it prevents businesses from selling directly from their site at a discount, and eliminates any incentive consumers have to purchase a product outside of amazon. It's one of the ways they became a monopoly, in addition to selling at
81.
by davebren 6mo ago
You can have legitimate use cases where it's a core functionality of the application to store it, so the user obviously knows it's being collected and agrees by using it.
82.
by davebren 6mo ago
I get it, but why would anyone trust what these companies say about their model performance anyway. Everyone can see for themselves how well they complete whatever tasks they're interested in.
83.
by davebren 6mo ago
This exploiting of benchmarks isn't that interesting to me since it would be obvious. The main way I assume they're gaming the benchmarks is by creating training data that closely matches the test data, even for ARC where the test
84.
by davebren 6mo ago
The point of the parable is not about the problem of induction but about how lying erodes trust.
85.
by davebren 6mo ago
As cooked as we were pre-LLMs knowing that security exploits are relatively easy to learn about online and use, yet things keep chugging along.
86.
by davebren 6mo ago
> If smaller models can find these things, that doesn’t mean mythos is worse than we thought. It means all models are more capable. It means "it's so dangerous we can't release it" was a blatant lie since anthropic wo
87.
by davebren 6mo ago
It should at least get the same coverage anthropic got then, if not more.
88.
by davebren 6mo ago
Seems perfectly comparable to anthropic's method, they just wrapped the same kind of prompt in a for loop.
89.
by davebren 6mo ago
You got the causality backwards, the cynicism is because it's most likely to be fake.
90.
by davebren 6mo ago
You've got to admit that crying wolf about how dangerous their new model is for the hundredth time right when the biggest story about the company was a leak that made them and their internal vibe-coding look totally incompetent is a bi

More ›