mgraczyk

this is needlessly pedantic

first, they are certainly not instructions so that is a much worse name

but more importantly, we use words in new contexts all the time. Do you object to calling the computer device "mouse" because it's not a mouse? how about "neural network"? "ignition" on an electric vehicle?

"cot" is no more misleading than thousands of words you use every day.

no, this is false. watermarking will not enforce anything like this

5g/day

I also started sleeping measurably better when I started taking creatine, but the broader evidence strongly suggests this was some kind of coincidence

"got caught"

to clarify, this behavior was announced with the model release

The numbers are public now, this is obviously false

For your math to make sense, Google would have to sell its stake this year

There may be more to it than buying compute but what you're saying does not make sense for Google. More likely Google wants a good relationship with SpaceX and possibly to buoy the stock, but it's a bad NPV trade

dangerous thing to believe IMO The models will get better, you will notice, everyone will notice. They will get better at coding and everything else. You should plan around that.

CSS is badly designed and uses a confusing, separate DSL with arbitrary rules designed before the Internet was widely used, before web apps existed, before smartphones etc

It's trash and throwing it out is good. Not learning it is good. Tailwind is a solution to a real problem.

More importantly, AI is good at it already and it's unlikely humans will need to understand HTML/CSS at all within a year or two. There's no reason to spend time learning how the gears work, just put the cover back on

Don't bet on the sigmoid flattening out any time soon. I don't know when it will flatten, but it definitely won't be within the next 2 years

This will probably happen but I wouldn't plan on it happening soon

What are you talking about, in what way is this supposed to be an argument about ads? It sounds like your dryer broke

It's not a good reason to be skeptical about cars as a technology (and by analogy brain computer interfaces)

I've never seen an ad delivered through any of these things. On smartphones I mean the phone/OS itself

It would be very easy to deliver ads via electricity. The utility could require you watch an ad before using more

I didn't list fridges because I've seen ads there, but these seem to have gone away in newer models (people don't like ads)

Except this hasn't happened with electricity, cars, washing machines, smartphones, smart watches, Bluetooth headphones, ...

Not all technology is bad

I don't understand in what sense they faked the process. What I've heard described is substantially similar to other SOC2 processes I've seen

And yes SOC2 is fake. Have you ever heard of a startup failing to get soc2 or doing more than a few hours of work to get into compliance?

What evidence did you collect that was not automated?

I haven't seen that and all the reports I got were under nda

Having gone through the SOC2 process multiple times and having worked with and read SOC2 reports from many public companies, it's difficult for me to understand the outrage.

The specific fraud allegations are bad (lying about US based auditors) but it's completely normal and common for soc2 reports to be templates with no company specific information. It would be unusual for reports to include anything about the specific information found during an observation window as some have suggested.

SOC2 is basically fake and it isn't possible in practice to fail to be compliant. You really can apply the same template to all companies and automate the audit process.

No the title is correct and you are misreading or didn't read. It was found with Claude code, that's the quote. This isn't a model eval, it's an Anthropic employee talking about Claude code. So comparing to other models isn't a thing to reasonably expect.

How many people have you interviewed and hired? I have interviewed around 400 and hired around 20, and I've seen data compiled on over 100,000 interviews. I have never worried about a false negative, except DEI stuff pre-2021

Maybe this is a Europe vs US thing?

> I am just not a fan of these types of interviews they tell absolutely nothing about the candidate.

Unfortunately this is wrong and I have seen tons of data at 5 companies showing this. These kinds of interviews really do correlate well with job performance

There is noise, but large companies in particular need a scalable process and this one works pretty well

Startups shouldn't do this though, but the reason is the opposite of what you're complaining about. It's too easy to accidentally waste your time on somebody who is good at leetcode

Still plenty of signal. You'd be surprised at how badly most people do at very simple questions.

Sure let's do it. I am pretty confident mine will be more maintainable, because I am an extremely good software engineer, AI is a powerful tool, and I use AI very effectively

I would literally claim that with AI I can work faster and produce higher quality output than any other software engineer who is not using AI. Soon that will be true for all software engineers using AI.

Somehow we went from writing software apps and reading API docs to research level astrophysics

Sure it's not there yet. Give it a few months

How about we do the following.

I have not done win32 programming in 12 years. Maybe you've done it more recently. I'll use an LLM and you look up things manually. We can see, who can build a win32 admin UI that shows a realtime view of every open file by process with sorting, filtering and search on both the files and process/command names.

I estimate this will take me 5 minutes Would you like to race?

Yes you have to be careful, but the LLM will read and process core and documentation literally millions of times faster than you, so it's worth it

Because it will take you years to read all the information you can get funneled through an LLM in a day

I don't pay for slack any more, I just picked the price of their enterprise plan. Large users probably get big discounts but it doesn't matter, the cutoff where this makes sense financially is probably around 4000 employees even at $10/seat