If the quality of LLM's keep increasing, will it emulate the abstraction shift that compliers gave us?
i.e. can useful deterministic complier-like behavior ever be found with a non-deterministic LLM approach?
In my view the answer is yes (for most people). I don't think the technology has to formally perfect to create a significant shift in how we write (most) software.
There will still be some who review AI code. Probably in the domains where people review complier code. But not everything actually needs that level of formal verification.
I concede that [3] is not the best case of this reading more into it. My point still stands though.
If visitors are one immigration mistake away from weeks in a detention camp, or any other unjust punishment it's reasonable that visitors would not want to visit.
I miss PyCon US a lot and I'm sad I can't go. As a Canadian, recent USA ICE government actions have made it really hard for me to attend.
No conference is worth getting thrown in an ICE detention camp. This actually has happened to people from my country. [1] [2]
A big part of this conference is the non-USA residents who show up.
> We attribute this largely to the sad but understandable decline in willingness of international attendees, as well as some vulnerable domestic attendees
It seems like part of the hotel problem is the lack of international attendees that are stopping travel to the USA travel due to recent government actions.
In general USA-Canada Travel has been down all year. [3]
Hoping for a future PyCon that is as big, but I don't have to take risks around my freedom to attend!
I had a professor in university that would do this pre-LLM.
He would take the first result on google and modify his problem to be slightly different.
Students that copied from google were easy to find and missed a key part of the problem.
Seems like a natural progression.
https://github.com/srid/nixci
Is this the project or is this a completely different Nix based CI/CD tool?
I can't find a Github or anything on the website.
The experiment is conducted with 11 subjects diagnosed with ADHD by pediatricians and psychiatrists. Binaural Beats (10 Hz via wired earphones, sine waves) are used for audio, and pulses of light (10 Hz via VR device) are used for visual entrainment. This audio-visual entrainment is done for 20 days with 15 minutes of entrainment per day. EEG was recorded pre and post entrainment sessions using an Emotiv Epoc X device. The analysis revealed an improvement in 8 subjects out of 11 subjects in terms of attention and spatial learning. [0]
This is cool, though I'm reluctant to give praise when they have been so weird with Linux support on their games.
It was annoying after buying Rust to learn that you can't play on official servers on Linux. The game runs fine on Linux, the devs just don't allow it.
> Do you not enjoy the pride that comes with attaching your name to something you made on your own? It's great!
This is like saying a photographer shouldn't find the sunset they photographed pretty or be proud of the work, because they didn't personally labor to paint the image of it.
A lot more goes into a blog post than the actual act of typing the context out.
Lazy work is always lazy work, but its possible to make work you are proud of with AI, in the same way you can create work you are proud of with a camera
I agree with this sentiment. Perhaps this is why there is such a senior / junior divide with LLM use. Seniors already build their theories. Juniors don't have that skill.
For stuff that's not in Nixpkgs like this, IMO flakes is the best option. Both of these require flakes.
On the topic of deploying flakes to consumer users. I'm currently exploring using https://github.com/nix-community/nixos-generators to create pre-made install iso's for a full "batteries included" experience. You can just "overwrite" the regular nixos image. Its Nix all the way down.
i.e. can useful deterministic complier-like behavior ever be found with a non-deterministic LLM approach?
In my view the answer is yes (for most people). I don't think the technology has to formally perfect to create a significant shift in how we write (most) software.
There will still be some who review AI code. Probably in the domains where people review complier code. But not everything actually needs that level of formal verification.