The purpose of a screening is not to determine competence, it is to determine whether recruit and company want to work together, which cannot be automated.

I understand and agree, but this is a technical screening. Typically you would have someone from HR have an initial conversation with the applicant to align. This comes after, if it's a go from both parties.

Furthermore, Niju does not automate the decision. AI is only used to create a transcript, a summary of the interview with, a list of important moments and a set of indicative scores on a number of criteria.

The decision is always with the human.

As a candidate I'm no more "done" evaluating the company after a first screening then the company is done evaluating me.

As someone with the privilege to be able to reject job opportunities, it's all but certain running into a tool like this would result in me immediately doing so.

If you can't take time to screen me, I can't take time to be bothered with your company.

Hiring is the most important part of your job.

I have a prompt to make it not rewrite, but just point out "hey you could rephrase this better." I still keep my tone, but the clanker can identify thoughts that are incomplete. Stuff that spell chekcer's can't do.

He's describing the Cuda execition model.

More interestingly, it's the same model as Intel's Implicit SPMD Program compiler (ispc, God that name is awful). But ispc works across SIMD lanes.

Yes this is just the GPU programming model without the hw perks (subgroups etc). Im impressed if he came up with it on his own but its pretty clear from the article that he didnt.

The GPU model works because the GPU is just wide SIMD with automagical scheduling.

To apply this to the CPU might be misguided unless you use SIMD aka like ISPC.

He makes such a good point. This is such a tragedy. Import std would of been an amazing thing that would of led some much less pain, but the committee got greedy.

For personal, I've got the nirvana life plan. It's great, for work, it's TODO.txt on a network drive.

Best idea of a solution I can think of is modifying bacteria to consume microplastics and releasing them into the wild. Make them a permanent part of the ecosystem.

Problem is they'd probably also start consuming macroplastics, and every plastic thing we use would degrade.

Then it evolves starts eating everything ahhh horror movie plot

Like most things in C++, I wish the default was `nothrow`, and you added throw for a function that throws. There's so many functions that don't throw, but aren't marked `nothrow`.

In my experience I've used exceptions for things that really should never fail, and optional for things that are more likely to.

If you squint hard enough, any potentially allocating function is fallible. This observation has motivated decades of pointless standards work defending against copy or initialization failure and is valuable to the people who participate in standardization for that reason alone.

For practitioners it serves mainly as a pointless gotcha. In safety critical domains the batteries that come with c++ are useless and so while they are right to observe this would be a major problem there they offer no real relief.

It would be nice if it just became a python interpreter. The concepts and build that CMake has is pretty good, but implementing it is a pain due to the quasi shell syntax.

These guys had a competition way back and settled on python.

https://www.scons.org/

This article should of had the conclusion of this is why you should use arena allocator.

This article could benefit so much from defining what an ERP is.

Enterprise Resource Planning. And any ERP admin would have seizures over the idea of just jacking in an AI to run wild in an ERP system.

Calling all this stuff "AI" is, in my opinion, the worst thing that has ever happened in the computer technology world. It's created this level of expectation in human minds that we're no where close to achieving.