runnedrun

Yea ChatGPT wasn't better at non hallucinating, but RLHF made the model much more useful for the average consumer, so consumers used it, instead of just other ML people. And, like you said, consumers are a bit less critical than experts from the field.

Oh interesting, from what I understand then, this is great for small context size models, compared to RAG? Is there research into how to make this more effective for large context size models (since context size of major models seems to be 4xing every 6 months at this point)?

Yea I agree, it seems like the first user asked for the analysis using publicly available data, then the second user asked for the same publicly available data, and ChatGPT provided it in a different format.

I always assumed copilot and ai code assistants were replacing docs, for most use cases. I guess I'd rather just ask the question I have, when I have it, instead of having copilot preemptively create docs for questions it thinks I'll have later on.

Have you experimented with sending back the evaluated result in some way? How often does it give back broken code?

Couldn't tell in the pdf, are there entity size restrictions for the applicability of the law? Like, could this make launching a new software business in the EU even harder? Can you be marked as "important" but also be a 5 person company?

Ok thanks! I'll have to reuse that github query for other projects too, very useful...

Love the naming scheme, and I'm pumped that developer tooling is still able to raise this much. Without digging too deep here it's not super clear: Are these tools more for end user developers or are people using these for projects, and if so, what projects currently use these libraries?

Trying to figure out where's a good place to start trying them out...

Does anyone know what the supposed benefits are for this kind of bill? Are proponents overtly advocating for increased surveillance ability?

Love this. How long has this been going for? Also excuse my ignorance of mesh networks: how does the mesh connect back to the internet? Is that going through an ISP, or are you trying to circumvent ISPs altogether?

I've not tried Datasette yet, but, with office hours, seems like now is a good time!

Does it make dataset discovery easier as well? I spend a lot of time looking for sources that are close to what I need— perhaps even more than I spend on formatting or displaying. Does datasette (or some other tool) make that easier than Googling around?

Personally, It helps me write succinctly. I'll start with long paragraphs then remove content to get to 1-2 sentences.

"how to incorporate science into the concept of operations for the crewed mission to the lunar surface"

I wish I knew what this meant. I guess operations are a purely conceptual exercise for NASA?

I'd be really interested in how the authors (and researchers in general) measure semantic richness. The paper mentions that semantic richness seems to influence visual working memory (familiar but otherwise random geometric shapes are harder than celebrity faces). However, how do you know when an image is semantically rich for a given subject? Similarly, how did the researchers know when the experimental images were morphed enough to be semantically unrecognizable?

Big fan of reveal, thanks so much for your work. Useful for so many more things than just presentations. I normally use it for static marketing-ish sites, but been thinking recently about how to use it for more interactive applications...

From a personal perspective, I agree. You and I seek new information when we read online. Subtack siloes don't serve our needs very well.

However another pretty important "job to be done" when people read articles is moral affirmation. I need moral affirmation as much as the next person, but I solve this problem with different products (personal conversations, etc). If you use digital media as a reminder that you you're right, substack seems like a good deal.

It seems like that's the majority of digital media consumers, but I might be wrong about that.

Is there an evidence that limited viewpoint diversity (as a substack "community" tends to have) in any way financial harmful? It seems like only the opposite is true. The less viewpoint diversity a channel has the more the participants seem to enjoy and be willing to pay for it. But I might be missing a good example of the opposite.

Persuasion is often a side effect of engagement. Well written facts, even without persuasive language, can still give the impression that the listed facts are more important than other ones.

Two facts from March 2020:

We have a shortage of masks. Masks keep you from spreading disease.

If you read an engaging article about fact 2, you'll likely ignore fact 1.