Hacker Newsnew | past | comments | ask | show | jobs | submit | Clent's commentslogin

This is a crazy out of touch perspective.

Depending on salary, 2 magnitudes at $5k is $500k.

That amount of money for the vast majority of humans across the planet is unfathomable.

No one is worried about if the top 5% can afford DRAM. Literally zero people.


The vast majority of humans across the planet aren’t making their money with their computer, which was the qualifier in the first line of my comment.

Furthermore, even if they did, the vast majority of them still won’t be using their computer to generate revenue - they’ll be using an employer-provided one and the things I’m talking about have nothing to do with them.


It global warming exists, why is there snow?

Same thing here. You're dismissing a flood because it rains.


It's quite telling that this is nothing more than an abstract but it has gained votes so fast it hit the top link.

The abstract page has a link to the whole paper as a PDF:

https://pallais.scholars.harvard.edu/sites/g/files/omnuum592...


They still aren't learning. You're learning and then telling them to incorporate your learnings. They aren't able to remember this so you need to remind them each day.

That sounds a lot like '50 First Dates' but for programming.


> They aren't able to remember this

Yes, this is something people using LLMs for coding probably pick up on the first day. They're not "learning" as humans do obviously. Instead, the process is that you figure out what was missing from the first message you sent where they got something wrong, change it, and then restart from beginning. The "learning" is you keeping track of what you need to include in the context, how that process exactly works, is up to you. For some it's very automatic, and you don't add/remove things yourself, for others is keeping a text file around they copy-paste into a chat UI.

This is what people mean when they say "you can kind of do "learning" (not literally) for LLMs"


While I hate anthropomorphizing agents, there is an important practical difference between a human with no memory, and an agent with no memory but the ability to ingest hundreds of pages of documentation nearly instantly.


That is true, but does it actually matter if the outcome is the same? GP is saying they don't need to remind them.


The outcome is definitely not the same, and you need to remind them all the time. Even if you feed the context automatically they will happily "forget" it from time to time. And you need to update that automated context again, and again, and again, as the project evolves


They document how to do something they just figured out. They store/memorise it in a file.

It's functionally working the same as learning.

If you look at it like a black box, then you can't tell the difference from the input and output.


I believe LLMs ultimately cannot learn new ideas from their input in the same way as they can learn it from their training data, as the input data doesn't affect the weights of the neural network layers.

For example, let's say LLMs did not have examples of chess gameplay examples in their training data. Would one be able to have an LLM play chess by listing the rules and examples in the context? Perhaps, to some extent, but I believe it would be much worse than if it was part of the training (which of course isn't great either).


50 first new Date()


This implies the US is subsidizing the world's healthcare system and sacrificing it's own citizens for the benefit of every non-American.

You're ok with that?


I'm okay with implying that, yes. More than implying it.


Apple also supports select state drivers license.

I would definitely expect Google to follow quickly.



Generally the hangup/timeline is with the state government, not Apple/Google.


You went with a cynical view of the situation and that view got you upset. That's interesting.


Almond milk is 2% almond. Cow milk is 100% cow milk.

An gallon of raw almonds requires a massive amount of water.

Using numbers to create illusions of comparison is the marketing equivalent of parsley in your tooth.


Agreed. The feature set is in desperate need of the search option both on approved photos and when attempting to approve additional photos. Very often I have to go into the photos app, find the photo, make a mental record of approximately where it is in history and scroll scroll scroll. Obnoxious and cumbersome.


What I really want is to create a special photo album for (Facebook/Instagram/Slack/etc.) and have it automatically gain access to whatever photos I put in there.


Look at that blue sky through sun glasses. It may longer be blue to you but I assure you the sky is still blue to everyone else.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: