JamesBarney

How long was the text? And mind sharing?

This should really be the #1 comment on this post.

This really depends what you mean. If you spend $5.00 on anthropic tokens does that cost Anthropic more than $5.00 to serve? No.

If you amortize all of their training and salaries over that $5.00 then yes.

If you only amortize the training costs of that specific model then again we're back to no.

Almost every. Pangram is pretty accurate on longer texts. I haven't seen any glaring examples of false positives.

Honestly academic writing is the only place where I think AI slop might be an improvement over the status quo writing style.

I don't know if Ed is far off the mark. But this article does nothing to help illuminate it.

He mixes estimated capex spend by like 3 different sources with actually commitments by the LLM providers.

He talks about how crazy it would be for ai providers to double revenue every year. But openai is doubling every 9 months and anthropic is doubling every 3.

It's obvious if AI consumption stops growing today those companies are in trouble, and if AI consumption keeps growing at current rates they'll be more than fine.

Most people expect growth rate to slow, just no one knows by how much. This will determine if there is an over build out or not.

The old school solution to this is have different schema in your database, and have views as the cross team public interface.

The issue wasn't sharing a database, it was not being clear about who owns what.

Having multiple teams with one code base that has one database is fine. Every every line of code, table and column needs to be owned by exactly ONE team.

Ownership is the most important part of making an organization effective.

The vast majority of people who write don't have a voice worth preserving. The rest can build out a voice document to make sure the AI doesn't strip it out.

If AI wrote and thought better by default then I wouldn't have to read the AI slop my co-workers send me.

My experience has been

(ordered from best to worst)

1. Author using AI well

2. Author not using AI

3. Author using AI poorly

With the gap between 1 and 2 being driven by the underlying quality of the writer and how well they use AI. A really good writer sees marginal improvements and a really poor one can see vast improvements.

If he was making an argument from data it would be cheap. But he's making an argument from lived experience against both data and someone who lives here.

Are you saying that the data is wrong and the only way to know what it's truly like to make it in America is to not live there? That's sounds insane.

You're completely out of touch. Even apartments are super expensive nowadays. I would gladly live in a house without A/C.

Why do you have so much certainty about what it's like in the US now vs 70 years ago when you're not American?

The American media writes articles about what gets clicks not what is true.

If you don't believe the enormous amount of freely available data on the internet. I am American, I had grandparents who were American. Poverty was a whole different beast in the 1930's compared to today.

But at least they could afford a house, right? I think a lot of people would accept living in a house without AC and more likely to catch fire. Is a house like that cheap today? No, right? It's crazy expensive as well.

I don't know many people who would rather live in a house without climate control than an apartment. A house from 1936 with no improvements is worth very little. When purchasing a house like that you're mostly buying the land.

Car technology in the past was worse, we know that. Cars were more affordable though.

Car ownership in 1936 was far below what it is today.

Like today then.

No, groceries were far more expensive. You can buy far more gallons of milks, eggs, lbs of ground beef, or potatoes at today's prices with todays median wage than you could in 1936 on the 1936 median wage. We have records of how much people made, and the cost of basic staples. This isn't something you need to guess about you can just google it.

Young people are rotting at home unable to go ahead with their lives because wages nowadays are not enough to pay for a house and a family. Why do people try to deny this obvious reality? Productivity didn't benefit everyone equally and people in the past had more opportunities to build a life inside a standard that was socially acceptable.

Because 100 years of data says that this is a difference in expectations vs people being poorer. Yeah housing is more expensive than it should be due to regulation but despite that people are still much better off.

You're ignoring the gorilla in the room. Why can't one live in a comparable manner today and bank the difference?

For two reasons.

1. They're illegal. You're not allowed to build a house to 1936 climate, safety, and fire codes with un-licensed labor. And boarding houses were effectively banned.

2. Market. Most people would rather live in a smaller apartment than 1936 style un-climate controlled death trap.

And the reasons are the same for cars. You legally can't sell a new 1936 car, and even if you could most people would rather drive an 10 year old civic.

They lived in a house or apt with a third the sqft/person that was far more likely to catch fire and didn't have AC.

If they had a car they most likely shared it. It was far less safe, didn't have AC, guzzled gas and polluted.

Never ate out and spent a third of earnings on cheap grocery store staples.

College and healthcare was much cheaper, and they got a lot less of it.

We're benefiting greatly from the increase in productivity. We just view our great-grandfather luxuries as our necessities.

They are large enough to cover their previous training costs but not their next gen training costs.

i.e They made more money on 3.5 than 3.5 cost to train, but didn't make enough money on 3.5 to train 4.0.

The most impactful censure is not the government coming in and trying to burn copies of studies. It's the the subtle social and professional pressures of an academia that has very strong priors. It's a bunch of studies that were never attempted, never funded, analysis that wasn't included, conclusions that were dropped, and studies sitting in file drawers.

See Roland G. Fryer Jr's, the youngest black professor to receive tenure, experience at Harvard.

Basically when his analysis found no evidence of racial bias in officer-involved shootings he went to his colleagues and he describe the advice they gave him as "Do not publish this if you care about your career or social life". I imagine it would have been worse if he wasn't black.

See "The Impact of Early Medical Treatment in Transgender Youth" where the lead investigator was not releasing the results for a long time because she didn't like the conclusions her study found.

And for every study where there is someone as brave or naive as Roland who publishes something like this, there are 10 where the professor or doctor decided not to study something, dropped an analysis, or just never published a problematic conclusion.

It's definitely not more productive per hour than 40. And if you're moonlighting another 30 hrs that'll definitely decrease your productivity at work.

Imnsho this paper is very low quality

Compared to like a phase 3 clinical trial, sure. Compared to your average paper, and especially your average business paper I don't think that's the case.

This work would suggest that the WFH movement would see a rise in sr. engineer salaries and a reduction in jr. engineers salaries, which we haven't seen.

You can't build a universal mind-reading device that doesn't require calibration.

I get the idea of publicly disclosing security issues to large well funded companies that need to be incentivized to fix them. But I think open source has a good argument that in terms of risk reward tradeoff, publicly disclosing these for small resource constrained open source project probably creates a lot more risk than reward.

Why? Why should we care where our pasta is made? Why not just buy the cheapest highest quality pasta, where ever it happens to be made?

Unfounded I don't know.

But if you're taking the time to build out side projects I'd open them so you can put them on your resume. Doesn't make sense to me to hamstring your resume so some kid you don't know gets a slighter better score in some cs class.

There are lots of restrictions and expirations around those IP rights.

Where are you getting these numbers? And how would you appraise the value of the windsurf IP which sold for 2.4b?

Inflation was stable and pretty low for the 10 years after the financial crisis.

Gold wasn't at $400 per oz before the financial crisis, and the spike in gold prices was fairly recent.

Basically the financial crisis response didn't cause inflation, the covid response did.

However, there's no speculation about what their earnings will be because they're currently selling their services below cost and there isn't really any story as to how they'll turn this profitable.

Do we know they're selling their services below cost? I'm pretty confident they're making money on inference and burning through large piles of cash on capex and research.