I dunno even isolated to language specific aspects someone with no experience could hang themeselves. I use a ton of Go & Javascript, if you take a fundamental part of both languages, async operations and compare them they have radically different approaches. Even if I was skilled at distributed systems with deep JS knowledge I would lack the skills to audit Go code effectively. goroutines and channels require a bit different mental model than promises and callbacks. You could easily let the AI architect a mess that works.
That is why I would be hesitant to review complex Rust or some other code I have no experience in, sure the language has ways to handle things like async, garbage collection, etc. but I would just be assuming the AI is doing it right or even worse trying to steer it to handle it in a Javascript or Go manner.
I dunno I find it almost more important than ever to have deep domain experience. You need to be able to scan output and spot any problems/improvements instantly. If I am just auditing AI output around an area I don’t have mastery of I am basically useless.
I understand the feasibility of this and sometimes in my lazier moments I skim the code changes and trust automated/manual testing to validate changes, but to just like... you don't even see what it did?
I think this all stems from Baby boomers controlling the narrative. Baby boomers had an insane relationship with the sun. Getting crispy brown tan, using tanning oils, using that metal collar to blast sun directly into their face, and frequenting tanning beds were viewed as totally normal and healthy things.
Big surprise they all got skin cancer. Then they swung the pendulum all the way back and now preach 24/7 sunscreen and never letting the sun touch you.
Is referring to these new models as "LLM"s still correct anymore? The frontier models are a more complex orchestration of LLMs and many other programming techniques, they aren't just a weight set anymore.
The whole purpose of this is to let non-technical users say things like "make a flappy bird games" and have it succeed after using 10 million tokens. A decent engineer could build the same thing in 20 mins with a small local LLM.
Same thing with a cyberattack, someone who generally knew what they were doing could quickly surface the info to perform each step of an attack. The govt just wants to ensure random 12 year olds can't type: "how I hack fortnite?" and get a step by step guide.
The computer should be a force for order, because being a living creature is chaos.
That said LLMs can be used in ways that promote order. People just got excited and wanted to believe they could be trusted in chaos mode.
For reference chaos mode would be prompting something like: "Look at my journal entries and tell me what I should do to fix my life". Versus using one to build a table of common themes and analyzing the resulting spreadsheet yourself.
Yeah being good is great but... some of the best engineers I have ever met are working at small companies making 150k, while mediocre ones fell into Meta/Google/Tesla jobs during the pandemic hiring frenzy and are now looking at retirement.
If the goal is to be good that is fine, but there is little correlation with salary.
Similar story where as soon as someone made a 100k+ offer to me I dropped out of school and never looked back (was making $12/hr at the time).
In retrospect it was not the best choice because college does give you a unique chance to get connections that let you into "the club" and will send you to the upper echelons of FAANG and early retirement.
I always thought I would just become a better and better engineer cutting my teeth at startups and then make the switch, but now it almost feels like you are a career startup engineer or a career big tech engineer, and making the switch is difficult. One is obviously insanely more lucrative, probably 3-5X with the way equity has exploded.
From 18-28 life was sharing apartments with various friends just wiling our time away working easy jobs, rock climbing, video games, girls, etc. Then the looming pressure of "you are wasting your life!" drove us all into serious careers and relationships one by one.
Is it societal pressure that pushes us out of that life? It would be considered an "empty" existence, but being a part time bartender who just meanders around and spends time with friends/family until their time is up would be a weirdly buddhist lifestyle.
Yeah they seem clickable because anything Anti-AI is a bit soothing right now, but he is constantly wrong and usually is pushing the angle of "these businesses aren't even profitable!"
Instantly close the tab as soon as the popup to subscribe to his newsletter pops up.
Same story, ended up just need to replace a fuse and clean out some filters.
I hate to be a "tech elitist" or whatever, but goddamn it seems like software is one of the most learning intensive careers you can pick. A little dabbling around the house has made me an amateur plumber/HVAC/handyman/gardener/etc. I should have gone into a trade, just for the comfort of having a skillset that doesn't need to be updated every 3 months. I guess my brain plasticity will thank me.
The dynamic of agent codes human reviews does seem like the only sane one for the foreseeable future. Even Anthropic themselves still fall back to this.
The problem is that sucks, even if all software engineers keep their jobs and salaries, the floor is still pulled out from under us. Imagine if a surgeons job was to supervise robot surgeons from a remote computer, or a woodworker just signs off on work before the machines do all the cutting and assembly. Sure they still have important jobs in their field but the soul & humanity of their skill is gone.
People outside of the tech industry are totally ignorant of how capable models are becoming.
Lawyers, Finance bros, Accountants, Doctors are all still barely being scratched at compared to the massive investments into AI tools for tech workers. I don't think they have had their "oh shit" moment where you realize you either need to adopt to this new world or be left behind, but it is coming.
If you believe extraordinary individuals are a product of nature and we don't know how to nurture them into existence then it makes sense. Whatever confluence of events triggers an Einstein has the greatest number of chances to occur.
Either way it seems dumb to structure our entire economy and future around elder care. I guess that is what you get when people older than 60 have 65% of the wealth and people under 40 have 7%.
Automating manual labor or repetitive brainless tasks is way different from automating high skill knowledge work.
I agree that automation is good when it frees us up for higher pursuits, but what is the end goal? All human labor is replaced? No human can produce anything of value relative to a machine? What does our life look like then?
I see a ton of people praising AI tools replacing software engineers. There is like a weird schadenfreude and an ignorance that this will only affect the tech industry.
They don't seem to realize that 100X the resources have been dumped into building coding/design tools as other industries. As soon as that hose gets turned towards their industry they are going to be in a similar or worse boat.
I don't think that is possible. Humans have always taken the path of least resistance, especially when it comes to work/school.
The idea that we just "trust everyone to carefully check and learn from AI output" as our barrier to human skillsets eroding is never going to work.
There is an Anthropic engineering post on HM front page that addresses this exact issue:
"... supervise the agent’s behavior via a human-in-the-loop. Claude Code previously protected against agents taking unintended actions by asking users for permission at each turn. Theoretically that works, but we’ve found the approach to be fallible. Our telemetry showed users approved roughly 93% of permission prompts. The more approvals a user sees, the less attention they pay to each, becoming over time much less diligent in their supervision. "
If every violation gets captured hopefully we can have law enforcement and judges that can use their discretion to make sure the "spirit of the law" is what actually results in punishment. Or we fire 80% of them if an AI can outperform them.
By spirit of the law I mean: rolling a stop sign at 1am on a flat country road is not the same as rolling a stop sign in a busy parking lot.
That is why I would be hesitant to review complex Rust or some other code I have no experience in, sure the language has ways to handle things like async, garbage collection, etc. but I would just be assuming the AI is doing it right or even worse trying to steer it to handle it in a Javascript or Go manner.