Having worked at both FAANG companies and startups, I can offer a perspective on AI's coding impact in different environments.
At startups, engineers work with new tech stacks, start projects from scratch, and need to ship something quickly. LLMs can wrtie way more code. I've seen ML engineers build React frontends without any previous frontend experience, flutter developers write 100-line SQL queries for data analysis, with LLM 10x productivity for this type of work.
At FAANG companies, codebases contain years of business logic, edge cases, and 'not-bugs-but-features.' Engineers know their tech stacks well, and legacy constraints make LLMs less effective, and can generate wrong code that needs to be fixed
It might not quite be there yet, but one key advantage large codebases have that I think LLMs in time will be able to better exploit is the detection of existing patterns - presuming they're consistent - and application to new code doing similar things or to fix bugs in existing code that deviates from the pattern in some way that causes a bug.
It's a different thing to what you're talking about, but it's one way I'd expect to see LLMs contribute a lot to productivity on larger codebases specifically.
large application codebase - consistent - have you worked in the field? I feel like usually there are 3 or 4 patterns from different people/teams at different points in time that spearheaded a particular ideology about how things "should" be done.
latest llama 3.1 is in a different repo, https://github.com/meta-llama/llama-models/blob/main/models/... , but yes, the code is shared.
It astonishing that in software 2.0 era, powerful applications like llama has only hundreds of lines of code, and most work hidden in training data.
Source code alone is no longer that informative as Software 1.0
Can you elaborate a bit more why render is good? we are on heroku and I have evaluated alternatives every 6 months since heroku/github outage 2 years ago [1]. But I don't see how render is better. 2 years ago render postgres did not have PITR. now they have build it, but Render's postgres offering is even more expensive than heroku, and queries run a bit slower on similar spec machines based on my test. I also don't like render charges per seat in addition to infra cost.
Aside from the content itself, the "Listen to Article" button uses a robotic, outdated TTS voice.
Shouldn't a company like Google use their latest technologies in public-facing content, particularly when discussing AI progress?
I'm genuinely curious about the decision-making process behind this choice.
Musk's bet, that every executive watched like a hawk, was to fire a whole bunch of "spoiled expensive developers" who would flap their wings about the sky falling (their perspective).
If Musk simply wanted to cut costs and get Twitter to the core of its profitability because the company had pretty much peaked, then it was a valid business plan in the Gordon Gecko realm.
But Elon talked out of all parts of his mouth nonstop, said he wanted vision and features but fired and hired for maintenance. It's weird that all the other companies too that as direction for their layoffs.
Executives were appalled at their lack of power in COVID too. I think the layoffs are an attempt to gain authority, I don't even think it is about the dollars and cents. Elites don't actually care about how rich they are after a certain point, what they care about is the gap between them and the "plebes", and the developer plebes were far too uppity in their view.
