No, everyone is not using AI for everything(gabrielweinberg.com) |
No, everyone is not using AI for everything(gabrielweinberg.com) |
- I'm getting my roof replaced due to hail damage. Insurance originally covered only $5k due to depreciation. I fed the insurance policy to AI. I learned about the appraisal clause and invoked it. At the end, I got another $6,500 back.
- I was having issues with plumbing. Four different plumbers came, they all said the cast iron pipes under the house need to change. Quotes ranged from $35k to $55k. I had AI walk me through the process. It taught me about the yard line vs. under-slab distinction, and suggested getting just the yard line replaced first because it's much cheaper and can fix the issue. I did that and spent $6k. The issue was fixed. I "saved" $30k for now by deferring that massive month-long project. (For brevity, I'm omitting a ton of boring technical stuff I learned about plumbing that helped me make the optimal decision - none of the contractors bothered explaining any of it.)
- My 2010 Hyundai Santa Fe is starting to show its age. I've taken it to multiple different repair shops, then fed their diagnoses and recommendations to AI and figured out which ones are trying to fleece me and which ones are being more careful and conservative with their repair recommendations. Probably saved several thousand dollars there. Learned a lot about cars too!
- My partner and I are converting the backyard to a wildlife sanctuary. The AI helped us plan what to plant where (depending on lots of factors like sunlight location, irrigation access, etc.) and it has been going really well. Also planned out a dragonfly pond to deal with mosquitoes. AI created a project plan, including schematics, material purchase list and step-by-step instructions.
- I've been wanting to do various other home improvement projects, but only ones that make financial sense. I took photos of my house, both inside and outside, and fed them to AI, and said "give me a list of projects I can do that will have high ROI for when I decide to sell this house". It spent 15 mins doing deep research, then came back with a long, prioritized list. If I do all the projects, I'd be spending about $40k and it would improve the house valuation by about $90k.
I can go on. There's probably dozens of stuff that I've used it for over the past year that led to massive time and money savings, and I've learned a ton as well about topics I normally would not have been exposed to or bothered to research myself. And I'm not even including all the work-related usage, both for my employer and my side business. That would be its own very long list.
Nor should they! It's such a shit thing to be emotionally invested in. Imagine people would have been upset about databases. It's really fantastic software and we should be happy to have it, and now go and make the most of it, for all of us.
I also just bought a completely mechanical film camera to learn a new old skill with no tech to fall back on.
[1] https://sparktoro.com/blog/new-research-20-of-americans-use-...
and for the ones that are using it (especially the paid subs). the lure is undeniable.
Actually anything that is about 90% great and 10% disastrously wrong is utter crap given the way people want and do use AI models.
They are great tools in the right hands and awful in the wrong.
the tech is pretty good at helping identify simple bugs when they happen and to write short sections of code given very explicit instructions but yeah I have yet to see good examples of short one sentence ideas turned into a working product that looks better than anything that could be a UDemy tutorial app.
My wife uses it for a (non-computer related) business though and it's great for all sorts of normally tedious marketing/social media type jobs though. Stuff that doesn't really require accuracy just needs text on pictures that looks good quickly.
I think everyone just has FOMO and doesn't want to lose to competitors. Eventually it'll die down.
That aside, this piece is interesting and ties together some useful numbers and studies.
I hadn't seen the recent Microsoft paper showing:
> 30 percent of the US working-age population is using AI [...] with at least 90 minutes of usage time in a given month.
I'm honestly impressed at how high that number is! That's a lot of adoption for a technology (LLM chatbots) that didn't exist four years ago.
"Everyone Is Using A.I. for Everything. Is That Bad?" - subheading: "Either way, let’s not be in denial about it."
It's clearly intended as rhetorical hyperbole - like "everyone's on their phone at the movie theater" or "everyone's fed up with AI hype".
If you read the actual transcript it makes it very clear that it's not claiming "Everyone is using AI" almost immediately:
> ChatGPT is the sixth-biggest website on Earth. Something like 43 percent of Americans in the work force use generative A.I.
It's tough to answer because you want to hedge for both an AI enthused employer and an AI hesitant employer with limited information about who they are and how they personally use these products. I've been responding with a sort of long winded answer about how 'there is clearly a learning curve for how this technology fits into any process and how I always always always double double double check yadayadayada'
I'm probably using the chat/ask functionality on a daily basis for quick debugging / new technology learning questions but I have yet to really use the fully agent or computer-use products because I've had more bad results than good the few times I've tried them (re-factoring a big repo of decades old fortran+C code for modern compiler/OS some things started to work but ultimately I abandoned that effort).
Have you considered just answering truthfully?
Would you even want to work somewhere where you need to play a role and where they flip out when you say the wrong word you should've correctly guessed through mind reading? That sounds not like a job but a toxic relationship.
This just sounds like a standard tech interview. Mind reading to find and perform the secret “signal”. Nobody flips out if you don’t find it, they just move on to one of the other 1,000 candidates for the role.
I remember the graduate recruitment days - If you told the truth you were the only candidate they saw all day that wasn't the captain of the football team, top of the class and voted most likely to succeed - aka the worst candidate they saw all day.
We all filter and “nudge” the truth during interviews. We all cater our responses to the person in front of us. Let’s not pretend otherwise. Your interviewers sure aren’t.
It's 2026, you gotta sell your soul just to get a phone screening
To be honest, I don't think I would want to work with or hire you, based on your response here.
I'm an old hat on both sides of this type of discussion from a post-grad view.
Recommendation: use it to own the conversation and to signal mutual fit. Yes, your idea of AI lover versus hesitant matters. I recommend reframing the question to pivot to your fit to the org (and org fit to you) question. Show/concisely explain how you consider whether LLMs are fit to a task and how to tell it improves outcomes.
An outcome focus and willingness to show thought process around a common use case will be a substantially strong response.
- Any long-winded answer to a question is immediate out and has been for years.
- Not having used agents and not being able to comment on what to do and what not to do with them is immediate out since early this year.
From all the tech that we have, agents are really not that hard to learn on the job. They're also not a magical silver bullet.
want a Flutter developer who is unusually strong at directing AI-driven software delivery. This is not a traditional "write the code yourself" role.
Why?
If the winding path is actually interesting and gives you insights into how the person works, why would that be a bad thing?
That’s a bit ironic, given the typical output of LLMs.
Having been in academia in the past and now in software I can say with a lot of certainty that this will take a lot more upfront work than otherwise.
Academic code does not have a lot of structure. And usually lacks a lot in terms of tests. While AI is best when it can mimic patterns as well as there are tests to target.
So you will probably need to budget a few weeks to establish good patters, docs as well as testing patterns before you can seriously make it really do what you want it to do.
Even with 3 weeks I'm just not the Fortran/C programmer to get that job done so I moved on to other things.
The analogy I've had for myself is that it feels like using a bulldozer to dig rather than a shovel. If you use it to dig archaeological artifacts, it can make things worse than you started. A lot of the work however, is just moving dirt around, so you are wasting time by using a shovel.
That this doesn't have a clear and obvious answer one can expect shows how the issue is politics, not strategy.
When you apply as a mechanic, there is no such weird political debates about certain power tools where people have passionate opinions on which tool to use.
That's probably not going to be enough for AI maxxers, but it probably won't be too much of a turn off for anyone but the most extreme AI minners, and everyone in between will probably be fine with it.
Frankly I plan to steer well clear of any "the majority of our code is AI generated" shops for the foreseeable future. Seems like disasters waiting to happen and I'd rather let other people step on those rakes
Look at the uptime and incident rate of all the big tech companies that have gone all in on AI generated code
As of 2023, 27% of American working-age adults were at a PIAAC Literacy Level of 1 or below, out of a total of 5 levels. This has gotten drastically worse in the past 10 years as, in 2013, Level 1 and below was only 17%.
Full scores for 2023 are: % Level 1 or below: 27% Level 2: 29% Level 3: 31% Level 4/5: 13%
For reference, Level 1 means someone can't really handle a full page of text, and can sort of handle simple 1-page web pages. Level 2 is the point where someone can start to handle a few pages of straightforward text, but still nothing particularly complicated.
(Both of those descriptions undersell just how bad it really is, but I'll leave it at that, for the sake of brevity.)
People that aren't using AI at all often aren't using it because they effectively can't. On a fundamental level.
Source: https://nces.ed.gov/surveys/piaac/2023/national_results.asp
There's a whole level of ignorance out there that is honestly dumbfounding to even comprehend. The numbers for numeracy and problem solving are even more horrifying.
(It's for this reason that the most popular apps in the US are algorithmically generated feeds of photos, and often-non-verbal videos shorter than a TV advertisement.)
In my experience, it's a mixed bag. I wrote this comment[0], yesterday. It reflects my current work, and how I am integrating an LLM.
I have used it for two parts of my project:
1) The backend (PHP), and
2) The frontend (Swift)
It has been a huge help, in both, but #2 is a cautionary tale. It really needs adult supervision, in developing native UIKit Swift apps. I'm realizing how truly bad the code it wrote was. I mean, terrible.
That's jarring, because it did a great job with #1. It made sound, reasonable design decisions, and provided code that is better than what I would write.
With #2, it behaved exactly like an inexperienced engineer, panicking, when confronted with real-world problems. My rewrite is going to feature a much simpler, sound approach.
All that said, it has been a net positive, and has increased my productivity by a large margin.
I guess the lesson I needed to get from this, is that it is good at helping me to find problems, but maybe not so good at fixing them.
But 2. For most other things, LLMs are fairly underwhelming. Research is usually mediocre. Try being rigorous and repeat your research prompt many times - then make a confusion matrix to tally up how many false positives and false negatives occur. And for the rest, be honest and ask yourself if the LLM is doing much more than a basic search engine query or trip to Wikipedia would have told you. For “normie” use cases, it’s handy-ish but far from revolutionary
That's an interesting analogy as, despite the real ecological issues with it and principled arguments against meat eating, in general meat consumption has trended upward globally in country after country for decades.
How does this connect to everyone's high level ideas/thoughts about "tech", "AI" and "morals and feels" etc. These lines can start to seem a little blurry, at least for me.
For example, would we say my partner is "using AI" (for all intents and purposes), if she's frequently using Google.com throughout the day, and then ends up picking and believing the AI generated answer overview at the top of the SERPs almost every time?
Or do we feel "uses AI", is more along the lines of the vampire kids running 1000 sub-agents on a mattress floor in SF?
I kind of find the whole spectrum really interesting because even basic phone use is now stuffed with AI, whether we choose to label it or not.
Google has search results still? I don't use Google much anymore (thanks Kagi), but this is what ends up showing for me, I don't even see any search results anymore: https://i.imgur.com/eHIA2Df.png It seems like it's 50/50 on page reload if the LLM-reply UI expands automatically or not, which covers my entire screen. I guess Google is doing some A/B testing perhaps.
Anyone who does a search and accepts the first answer just doesn't care much or is incompetent. Anyone with any critical thinking whatsoever does way more than that if they want a correct answer.
"low effort and convenient" seems to consistently win over "best quality" and this is going to be a downgrade in everything, for everyone
This is a pattern I encourage - the AI might not be reliable, but with coaching, it can produce reliable tools. `colordiff` was causing issues with `less` when I was looking at diffs (character encoding issues I think), and when I asked Kimi K2.6 what to do, it built me a rust command-line diff tool in one shot that I've been using ever since (it even downloaded rust, wrote the tool, and compiled it).
For example; ChatGPT is replacing my Google searching. Not necessarily because it's better, or because it's summaries are better than Google (I find them subjectively better but it's not clear cut).
But because the app has a nice history; can ask a relatively complicated question and go do something else and then come back to it, ask a follow up. Etc.
None of that is specifically an AI benefit, but it's a workflow that really helps, well, flow.
Also, Gemini is free or at least has much higher usage limits than ChatGPT or Claude, and it's well integrated into Android and soon Apple with their new Siri, so things like circle to search just work well.
If I am honest I believe my final solution will be a combination of Open Claw, a custom knowledge wiki based on Wikmd. I just need a good all for Claw with history that is as good as gpt
Edit: and context too. It inferred my energy supplier from previously chats and so when I just asked a pertinent question it referenced their policy. Admittedly Google will have way more context if they get the product right.
This makes me less bearish on the AI investments that are being made, if 70% of the working age population isn't using AI then there still is a lot of growth. The future is here, it's just not evenly distributed (yet)
"No, everyone is not using the internet for everything."
Which would have been entirely true when written, and entirely false a relatively short time later.
Everyone does use the internet for everything today, and everyone will use AI for everything soon.
I'm not saying it is a good thing, but this is completely out of touch with how dependent (most) people are on these technologies.
Local models are highly likely to dominate in the long run as "good enough" inevitably becomes trivially cheap. This is a very different pattern of incentives and adoption compared to the internet.
I think it's more similar to the advent of personal computers. They had a brief surge and then turned into something else (smartphones, cloud, etc.) for all but a few niche cases. AI is not changing the consumer landscape. It's getting absorbed into existing platforms where there's a clear use case and benefit. It's just another expected software feature. This is far from the first time people have rejected a "personal assistant" concept and they'll just keep rejecting it.
I agree that where models run will will change over time, probably they'll run everywhere, but it's still the same kind of AI we are talking about.
Smartphones are personal computers.
I don't get these comments.
If you consider things like the machine learning filters in your smartphone camera and Google's AI Overviews for searches it's entirely plausible that the US is currently at 75%+ of AI usage.
If I worked in marketing/growth for an AI company I would try to consider some ways of breaking through this gap.
If you're working in some vanishingly rare domain then maybe it's not yet, but most coding challenges are very much in the wheelhouse of the current frontier models.
I am constantly looking for a new job, but all of them are also require AI coding experience.
Looking things up and asking questions was always something for a minority of the population so the language model usage being relatively low isn't a surprise.
Problem arises if the non-AI segment is leveraged to create regulations that impact the AI using segment negatively.
i am not saying it's really powerful or great. but the lure is undeniable. because of how low friction it has become.
They are great on exploring, understanding and finding bugs in existing codebase.
They are great for simple or one time scripts/programs.
They are terrible, really terrible coders. The overengineering is so deep in their training that no matter what is your prompt, your skills or agents.md/claude.md, if you don't babysit them continuously, at some point they will just fuck up your codebase.
Software engineers are definitely in a bit of a bubble here. Are we just early adopters who see the value sooner, or does it uniquely benefit software engineering, or do we just like cool automation and we're deluding ourselves that this adds value beyond the cost?
From the 3 people I interviewed, all of the answers are very similar which is along the lines of: Kinda, but we need to be careful of using it, privacy, hallucination, etc.
All very safe answers and doesn't say anything new to me. If they had been more specific about why and their experiences with it, I'd probably favor them more due to their experience with it. It'd also signal to me that they form their own opinion rather than simply following the crowd.
That is of course assuming that they're looking for some long-term stable team member.
A skilled interviewer smells dishonesty.
However, and to be fair, whether and how they act on it depends on the specific situation.
Most of our stuff in this world actually does work, and the reason why it does is that skilled (teams of) people that care have built it. Meaning that these people can be found in many _many_ places.
Very few jobs are looking for opinioned most are looking at people who might fit in unless you are hiring to distory from without.
I think upskilling is the right move in this environment and it is dead simple: Invest a couple of days to show initiative, learn agents yourself and be able to speak from true experience.
You’re more or less admitting that you’re playing trendy tech lottery. Which is fine, but maybe not generalizable to the whole industry.
It makes perfect sense that they exist and were way overdue for an update, but they're just extra blades on the multitool. Perhaps in some designs they become more integral, but that is expected and invisible.
Yes "everything", but that's not even close to sufficient to become a huge breakthrough like the internet.
In that case, it's way better to simply write the code yourself.
IMHO the best of both worlds option is agents working with deterministic CLIs. Where the agent does the reasoning (and text generation) but uses CLIs to carry out all of the actions (issuing refunds, unblocking accounts, or whatever).
It's possible to get very reliable and consistent work out of agents when they're using well written prompts with well designed CLIs.
But that's not worth trillions of dollars...
Instead of using the LLM to create deterministic tools, we are using LLMs to replace them. It's completely backwards and I don't know why people (especially high ranking people in my company at least) seem to think that this is the way forward. No, I don't want a whole CI pipeline that is just LLM prompts. Yes it's very easy, but it's expensive, slow and prone to failure in ways you can't even predict.
Same things like using LLMs for the code review process. What would have been a simple linting rule is now a pass with an LLM rather than using the LLM to create the linting rule, which it is absolutely excellent at creating.
Yes, and we're also seeing lots of companies claiming they're using "AI" and it's just deterministic under the hood.
The agent paradigm will eventually give way to experiences that are a hybrid of deterministic and non deterministic and you won’t even know the llm was involved or visible.
The issue is, they don't want to provide "better" support but "cheaper" support. Imagine a trained agent that understands the big picture. Now imagine a company investing in humans to use AI to retrieve knowledge that the human can easily identify as being relevant or not, and using that knowledge to better aid the customer.
Right now AI is being sold as a "we don't need support personells" instead of "how can we provide better service." For a lot of products, better service will probably not matter as "cheaper" products will win most of the time.
Most people don't want to pay for better. They want to pay the same for something better, which is what companies are not investing their time in figuring out how to use AI properly for I think.
A lot of people want to pay for better, but that is hard. Better is more expensive, most of the time, but being more expensive is no guarantee for being better. It feels like the correlation is very weak. Most expensive products are just expensive, not good.
If there was a reliable way to identify the "better" thing, I and a lot of other people would go for that every time we can.
Instead of refining their approach, or challenging their current knowledge base for discovery of inefficiencies or baseless assumptions, they'd rather hit an "easy" button.
I understand the desire to NOT do work. I understand the desire to spend quality time and free time with family. And I understand the idea that familiarity breeds contempt.
What I don't understand is the willingness to replace a deterministic language/framework/approach with a probabilistic slop machine.
The AI psychosis is a real thing.
Regardless which task is handed to him, he "discusses" it first with Claude and very often comes back with like "The AI said... X"
Now that’s real value.
These models do not have any experience. They're not sentient. And are in no way capable of being "smart", let alone becoming "smarter".
<b>Included in your plan limits until Jun 22</b> <br><br>Fable takes 2x the usage of Opus.
<b> Switch models when a message is flagged</b><k <br> When safety measures flag a message, automatically switch to a different model to keep chatting. When off, your chat will pause instead. <a href="https://support.claude.com/en/articles/153636
target="_blank" rel="noopener noreferrer" > Learn more</a>
Ok wait maybe not the next one but surely the one after!
Hasn’t happened yet and there is no evidence it will.
It saddens me to see that high quality content is drowned in this sea of garbage to the point of being almost impossible to find.
Then you go on to search for something, and find only results that are clearly AI generated pages and come to the conclusion that directly prompting some LLM is better than reading an AI slop page that's output by the same AI for slightly less specific prompt.
My concern is that this will only get worse over time - which is great for companies selling AI tokens and bad for society and whoever wants to interact with other humans over the internet.
Swift, not so much. It's relatively new. Looking at AI's abilities like an engineer's career span scaled about 10-20x of time makes it make a bit more sense.
It's going to be worse at newer/niche things, intuitively - which is only going to get worse as it "learns" from garbage outputted by other LLMs moving forward.
You seem to assume that autoregressive pretraining (and unfiltered behavior cloning, maybe) are the only ways to improve LLM performance.
That's fine, for a lot of corporate applications, but not for the stuff I write. I'm anal, I know, but that's how I roll.
The classic AI Gell-Mann effect.
The diagnosis, however, is not.
Have a great day!
But I've seen Claude write crazy code in Python and JavaScript, too
PHP has huge, entire frameworks and systems, refined over years.
The moment you have to interact with the physical world or humans (psychological, imaginative, aesthetic, etc), there are often undiscovered or changing rules—or no rules at all. Or systems are subject to perturbations beyond a defined scope.
The other thing I believe is software developers are experts at doing the things that allow them to make doing those very things easier and more automated. And they do this in public, perfectly documented online.
Both because of the things I described above and because software developers have created the largest machine-accessible training set for plying their trade of any trade, ML—that is ultimately interpolating massive datasets to do things—is unsurprisingly uniquely successful for software tasks.
The less popular a language, the more models struggle.
Writing, UI, and presentations have similar knowledge bases.
Outside of those, quality becomes much more hit and miss. If you ask for a recipe you may get something good, or you may get something completely inedible and random.
"Domain specific knowledge" really means "strong foundations and relevant abstractions" and LLMs just don't do that reliably.
> Computers should adapt to people. Asking people to make themselves more legible to software — to turn themselves into a database — is a doomed idea.
I've been in software a long time, and I do sort of see this trend, but I think it's because these are tools that build other tools. The interface has always been a 'best I can do for now' thing, with the focus on doing things that are useful. Computers were just calculators in the beginning, which led to more complex calculators, instruction sets, programming languages, operating systems, GUIs, interconnectivity, etc.
What people are doing today is experimenting, like they always have. They're putting their experiments out there so that others can use them and build on them. Some will use those tools to build other tools, and some won't. But over time, the experiments that work will get distilled and turn into real products that people who 'do not yearn for automation' will still want to use, so it seems like the value is there.
I guess the real question is whether they will create value that offsets the near-term costs, because I don't think the billions in investments are sustainable, and I'm not convinced the centralized data center paradigm is the right way.
Fair enough, so if there were one “right” answer, that would be the one to give whether true or not.
But here there is no obvious right answer. If the employer is looking for a particular answer, the poster doesn’t know what it is. In that case, the best thing to say is simply the truth, particularly when the truth that the poster gives here is completely reasonable.
The attitude suggested by your response suggests you haven't lived that reality yet.
Either way, I'd rather be rejected by an employer for speaking my truth, than lie to be somewhere I'd rather not be.
"Speak the truth, even if your voice shakes"
My truth is I don't care either way . I get the sense that's the same for parent poster. They just want a job and to say the right thing to get past the hiring filter. Even if I did have a truth its not something I would put above being remote, pay and how a company develops software. I'd rather not have a truth and not have a daily standup.
I think you'd rather have good odds at some companies and 0% at others, rather than abysmal but non-zero odds at all companies.
And as an added bonus, you might get hired at a company where you're actually a good fit, rather than one you weasled your way into, and get to pay rent, food bills, and other expenses through employment for a long time!
"Wouldn't give a straight answer on question X" isn't an instant no-hire, but it's not a positive signal.
Plus, and leaving that aside, I have my doubts that even if you did that, that that company would stay alive for very long. Reality has the habit of eventually ripping this kind of unproductively delusional people (like e.g. a boss that flips if you don't say the right word with regards to the current hype) to shreds eventually.
Look at musk's companies. They will basically never (on any near timescale...) produce GAAP profitability and yet their IPO is in the trillions. To the point that S&P refusing to suspend their GAAP profitability requirements means the index will basically never see this company in it (which I'm quite pleased about).
The power of already-accumulated capital is simply more powerful than things like "don't be completely pants-on-head stupid about a recent fad" "don't seig-heil in front of the world stage" "there's no point in having people come to an office just to spend all day on zoom" etc etc etc.
The market can remain irrational longer than you can remain solvent, and companies can remain irrational longer than you can go without contributing to your 401k.
The best benefit about working in a large office is that nobody checks the basement.
I much rather prefer someone who needs 3 seconds to triage a question and tell me: "This is X, I know this, here is the solution" or "This is Y, I don't know it, but I will get back to you within 24h".
I do absolutely not want a "Well let's think jointly about this for a couple of minutes". There is no jointly with your boss. Let's do a some math of a 1:12 manager to direct report ratio. That means for every hour you have, your boss only has 5 minutes. And if you talk to your boss' boss, they have 25 seconds for every of your hours.
So in the same interest of helping post-grad job seekers, do what you've gotta do to get yourself paid, but maybe don't presume that vibe_that_works speaks for every hiring manager.
Not to disagree of course that time is limited, but in my experience, optimizing it this harshly leads to poor results, because eventually, you just get leapfrogged by reality.
Hyper-optimized systems are brittle and can't really adapt to the market changing.
But yeah, I guess they still need developers. Just doesn't sound like a fun job :D
So let me take this a step further. You want to meet your boss' boss for 10 minutes to present them something. 10 minutes of his time are an equivalent of more than 20 hours of your time. So if your initial idea was to "take maybe 1-2h" to prepare for this -> You are underprepared by at least one order of magnitude.
But why?
Most of my most fulfilling experiences in tech have come out sitting down and hashing out a problem with someone else (including with managers/leaders).
It sounds like a miserable org if I am not expected/allowed to have an actual back and forth conversation with my boss. If I'm employed to be on a team working on an aligned common goal, why would I not use that collective skill and experience to my fullest advantage?
You're describing a coding sweatshop. What is the point of any discussion at all then? If the "boss" can't carve out enough time, that's their own problem. Letting that stress propagate to the team is plain bad leadership.
I know you might think some of these candidates don't have other much better choices to find work, but they absolutely do.
But that sounds more like "evasive" is the problematic attribute and not "long winding".
Which does show up at the same time often, true. But not always.
Although you can certainly do a better-and-worse job of preventing these kinds of issues.
Some people might use skill-based scripts, MCPs, or some kind of raw access to a database. My point is that well designed CLIs are the optimal programmatic interface, for many reasons.
Wait raw access to the database? That’s one of the options for issuing a refund?
These people just destroy their ability to read and understand the systems they're working with. I actually see it as them making themselves redundant. Because if you can't understand anything without Claude, and Claude doesn't even give the right answers, then what are you worth?
I bet lemmings are grateful they were left behind.
It beggars belief that people think that they should rush in some uncertain direction, like some drawbridge is going to be lifted the moment people work out what the right direction is. It's utter stupidity.
Of course I will do that, I get paid for doing that.
Most of the times I can convince that AI is not necessary by showing small PoC flow with AWS diagrams of data flows. This works well especially if the ask comes from technical people.
Other times the C-level interjects (CEO, CFO, sometimes even CTO) and demands that AI should be there. I literally had CEOs send me instagram reels of some AI shovel-sellers to demonstrate that I am wrong and AI is the way to go. No point arguing after that because I have no problem implementing whatever AI they want rather than losing a paying project.
I’d actually love it if LLMs could skip the slow high level lanaguages entirely and just churned out some weird LLM bytecode that was closer to the metal. I don’t want to read it or understand it at all. Here’s my spec, build it and notify when done. I want to ship stuff not build or dick around with code. Basically like when I go to a shop because I want a table, I don’t care if some carpenter “crafted” it or a machine mass produced and spat it out. It’s cute, but most people just want stuff and don’t care how it’s built.
My job is safe because I’m the only person at the company that actually understands what the actual code is doing and I’m the one that gets the calls at 2am and weekends.
“Weird LLM bytecode”
Why not just generate object code for the target mschine directly?
I'm unlikely to run into many of the problems that (for example) the PornHub developers hit, several times an hour.
In that case, I benefit from folks like you, that allow me to have solutions that scale down to my level.
You were replying to “The job market is much different when you're just starting out”. The past is not now, and you are not just starting out, so your comparison of their position and yours is invalid IMO.
> and will do it again.
Good for you for sticking to your guns, I'm about to do the same with a company that has all but said “dig into AI or get left behind”¹, but those starting out as freshly minted grads likely do not have the luxuries that we might have² and the jobs market is freakishly competitive for them right now³ in a way that I don't think it ever has been before.
--------
[1] time will tell if I leave of my own volition before getting kicked!
[2] experience (both actual experience and experience “talking the talk”) to help getting the next gig, a mortgage paid off so making ends meet is easier, etc.
[3] It had been heading that way for a while, the recent explosion of GenAI+agnetics has made it worse.
I certainly feigned enthusiasm when I was in high school to get an after school job in order to help my family buy food.
Lack of adequate calories and nutrition negatively compound. You lose the ability to focus, you increase your medical risk.
I experienced that in my childhood. It’s terrible. I did very poorly academically when I did not have access to food. It’s astonishing to me how fast my academic performance improved after consistently having access to food.
Saying you would rather put yourself at risk instead of hedge your answer on a minor interview question in order to increase your chances of getting a job offer seems like an issue with prioritization.
I bootstrapped myself from poverty to Staff software engineer, past the age of 45.
Is that privileged? Or sheer will and force of effort?
I am not unique. I am an example.
Even though your position might be the result of effort on your part, you do have to acknowledge that you’re privileged to be in a position to expend that effort on what you want, instead of something else, like finding fresh water daily, or whatever. It’s not sheer will that you were born in a (even marginally) more favorable environment than others.
The term “privilege” here doesn’t just mean a trust fund nepo baby.
So while I agree that privilege is certainly a factor, so is what I've just said.
A lot of people here live very cushy lives that cushion them from very pointy thoughts and questions. As someone who too has to live in this world, I'd rather they didn't.
It is very sad to me that people do feel that pressure, and how the current job market is.
On topic with the article, I would love to be able to trust AI with more, but have found that I have some useful moments with it, but more because of Internet search not being how it used to be for quality.
For example I think the decision to stick to certain morals is very hard if someone has a disabled dependent, are disabled themselves, or require consistent access to healthcare. There are different lines for different people of course. Our ire shouldn't go towards individuals who make these decisions but the people in power who force others to be in a position where these decisions need to be made.
I don't want to preach martyrdom, but I am also offended by people choosing moral bankruptcy when faced with even the slightest hardships.
Which might not be ideal, because "orging for the sake of org" to my understanding consumes significant resources not going into building products/marketshare/shareholder value.
But then again, I'm no hiring manager in such a structure, so this is probably just an uninformed take.
It's… like… not that simple.
Update:
Every street corner has a yellow garbage bin for recycling. That is where your plastic bottles go. Seems like a better system than having elderly going through bins.
At Big Tech Company I Work At the LLM is quite happy to make raw API calls. If it thinks the data is big, then it'll write a Python tool to do it.
The reason crafted backing CLIs are useful is you can guide the LLM towards stuff that is immediately useful rather than hoping the nondetermism can separate the wheat from the chaff.
Take CI: is it interesting to know which tests passed? Maybe, but probably not. What is really interesting is what failed. Instead of having the LLM go out and talk directly to the CI system, write an intermediate CLI that filters out less actionable stuff by default, and have a flag that'll deliver the full dump if necessary.
It's a skill to do this stuff, and it's a lot of hard won experience than something I think is easily teachable. You kind of have to feel out your model and how it "thinks" about solving problems.
And then a new model version comes out and you have to learn it all again!
That's one of the things that I appreciate about the PHP that the LLM provides. It uses modern idioms that make better use of the modern language.
We are slowly waking up to the fact, which was always true, that “coding” is just a fanciful preparatory task in order to appease the spirits properly so that we may invoke the spirit of what we are actually after: a live, running process that does useful things. Code is completely useless when separated from that fact.
Typing it is a complete waste of time unless getting up close and personal with it will result in some kind of useful and actionable improvement in you or your understanding. Knowing when it does and when it does not have this property is a skill of its own.
I believe this is the general belief about basically every human skill, that if you stop doing the technical fundamentals you get worse at understanding the activity. The question is whether coding is like sailing a square-rigged wooden ship, which became completely useless knowledge after the invention of the steam engine, or if it's like playing an instrument, which while technically unnecessary after the advent of MIDI and other tools, absolutely hurts your ability to arrange, compose and perform if the skill is neglected.
For my money: I think the AI scenario is more like the latter, but "humans are worse at coding" isn't the consequence I see coming. I worry that in ten years we will be awash in software that's impossible to understand. I don't think that's happened in any human industry ever. Someone has always understood how the machines are built, even if they're very remote from the users of the machine.
Like, perhaps, understanding that it is free of security and functionality bugs.
If you find yourself writing repetitive code you should consider adding a layer of abstraction. If your language isn't powerful enough you can write a code generator.
Code is obscenely low level.
That is one of the things code does. It also communicates the developer's thoughts about how that process should work to others. If the latter is neglected, the code becomes very difficult to collaborate on. Very few lines of code that are written are "write once". Mostly they're changed, repeatedly, over time by many people. The live, running process is a very temporary entity by comparison. Yes, it needs to exist and do useful work. No, it is absolutely not the only thing that matters.
I would argue that this is nearly always the case. I don't think people really understand programs that they've only read at more than a very superficial level. This is why I tend to make (temporary) small changes, printlns, etc. when exploring a new code base: it aids greatly in understanding how a program actually works.
And it's even worse (in my experience) with LLM generated code, as it tends not to result in particularly understandable code. It is a lot like LLM generated prose: it often looks entirely reasonable at a surface level, but has a of weirdness/incorrectness hidden beneath the surface. But that surface level makes it very hard to avoid glossing over the details when reviewing the code. For this reason, I personally find it's much more effort to carefully review code than it is to write it.
Humans make mistakes all the time, but their code tends to naturally be structured for human understanding (to some degree based on skill/experience) because they themselves needed to understand it to write it.
I think LLMs are very useful tools, but after quite a lot of experience using them, I think it's generally better to use them as a sounding board, or to help you get unstuck or remove points of friction. Using them to write all of your code (at least for me) seems like a net negative.
I also think it's extremely easy to overestimate how much time they save. It feels like they're a productivity boost because it takes less intense focus to implement something. But I've experienced several instances where actually writing the code myself would have been both quicker and have resulted in better code.
All that being said, it can also be really hard to not write all of your code with agents once you get used to it. There's also a kind of slot-machine-like effect where you write a prompt, excited for the result, and when it doesn't quite come out right, you think "ah just one more prompt and it'll be good." It's hard to see when you're actually doing it though.
It's also weird to me how much people think typing is what the LLM is replacing. Typing was never the hard part. It's the translation of the high-level idea into an unambiguous process that's hard. That's also the valuable part, that requires thinking through the edge cases and consequences of decisions, and that just gets glossed over when using an LLM unless you rigorously review what the LLM has done.
At the end of the day there's a real tradeoff to be made, and it's worth being conscious of what's being given up.
Those costs don’t disappear and it’s truly naive to think they don’t matter. Take security issues, they may arise because what you thinks was the input is merely a subset of the true input range. And the extra possibilities lead to unforeseen behavior.
A lot of programming is about ensuring that the input and the output are the sets defined in the specs. And the rest is that the transition/relation is the right tradeoffs of performance, correctness, and costs.
No one has ever needed to do that for something that is new. And if it’s not new, you want to do it repeatedly with some guarantee of reliability. Not just in an uncontrolled manner.
That is why we have snippet systems, macros and code generators. And the best with code is to solve problem once and reuse the solution. Which we have done with libraries, frameworks and supporting software.