Like you said, I think skills are quickly subsumed by model abilities. I leave relevant notes about my own projects in agents.md and local comments around tricky bits of code that need them, but otherwise I just use vanilla codex.

I think a lot of people are missing some of the advantages of skills - they're to give an agent/harness the knowledge of how to do something or more context without it always being part of the prompt.

This could be to try and expand the capabilities of the model but in it can be used to store personal preference/workflow/organisational knowledge/improved efficiency etc. The harness could work through and build some of this on its own - but it's much more efficient if it has a document it can just look up. I use them for the database structure and it's quirks, how to use branded document templates, external document tone, system workflows, compliance requirements etc.

I wouldn't want these to load every time but being able to call them is really useful.

By saying it “isn't that hard” you're admitting that it could definitely be easier.

Models keep on improving though, so doesn't fine tuning become an ongoing task with ongoing maintenance burden?

(one of the blog post authors here) -> once you set up a finetuning pipeline, it's often trivial to rerun it on top of a new open weights model. so, it's orthogonal to base model improvements

What's the benefit of SSEs here?

I guess this ethics stuff is cool, but I'm more interested in how good it is at running a business and dealing with adversarial humans like in previous vending machine experiments. I hope they release something on that soon.

You'll make more money when you downsize or move to a cheaper city or country for retirement if your house goes up in price. If you own a house in San Francisco you can sell, move to a vlcol city, buy a new house, invest the remainder, and retire. You probably will want a smaller house once the kids move out too.

> move to a cheaper city or country for retirement if your house goes up in price

What kind of life experience would lead to such a conclusion? It's almost like you have no idea what a community is - with one's friends, neighbors, familiar shops, clubs, church etc. Money is not everything in life. Any policy which forces people in their late 60's to uproot and go somewhere else to rebuild their lives is bound to face a very angry backlash. A mindset like yours is why we got prop 13 here in California.

Is this a situation where AI will go away and we will regret the loss of skills? At worse, we will be forced to use open weight models instead of the cutting edge, so I don't think it's a big deal. I'm sure people got worse at arithmetic after the invention of the calculator.

I don't think the real threat is at the individual level, but at the societal level.

Building skills over time leads to insights that lead to innovation.

AI does many interesting things, but it doesn't innovate (yet).

The real threat isn't that we'll all lose our skills (possibly) and then lose access to AI (unlikely), it is that AI will remain at roughly currently levels and we'll dull our skills due to reliance on it and innovation will stall because we've offloaded too much of the thinking to the non-innovative machine.

I'm not saying this is what definitely will happen, but it does seem like a very possible outcome.

The number of people now involved in software development has now increased because of a lower barrier to entry. I know many people who would previously use a no-code tool or hire offshore devs, or simply not have their problem solved, who are now vibe coding. Many of these people couldn't write very much code manually if they had to, but they're closer to understanding software than they were previously.

Yeah, this describes me perfectly. I can't really program, but I'm now building a bunch of different projects and submitting PRs that people seem to appreciate.

https://github.com/kristofferR

One of them, a Home Assistant integration for controlling adjustable beds, would be borderline impossible to do well manually - I've vibe-reverse engineered the Bluetooth protocols of more than a 100 Android apps.

> I've vibe-reverse engineered the Bluetooth protocols of more than a 100 Android apps

Impressive coverage, but how do you verify that the reverse engineering matches what the hardware expects? Do you find people who own those beds and get them to test stuff for you?

It's hard, but I try to verify it as best as possible by doing multiple independent runs, often one with GPT-5.5 and one with Opus, and then verify that they agree. When I'm as certain as possible I release it and rely on feedback when something doesn't work properly.

> The number of people now involved in software development has now increased because of a lower barrier to entry.

I don't see what this has to do with what I posted.

Yes, AI lowers the floor on software development, and there are positive aspects of that, but that doesn't change the possibility of an innovation stall.

re: concerns at a societal level. More people are making software than before. People who have been coding for a long time have moved up an abstraction layer and are further from the code. But many people are actually closer to the code than before.

> But many people are actually closer to the code than before.

More people are coding, I wouldn't say they are closer to the code if they are vibe coding. Are any of them going to produce the next breakthrough in computer language/framework/method of development/etc?

The risk of AI is that we dull the skills of enough people at the high end of the state of the art of the nuts and bolts of software development that we slow down innovation on that end. That's the concern.

Previously-non-programmers vibe coding CRUD apps they never could have before is all well and good but really has nothing to do with this concern. They may create wonderful and successful businesses but they are irrelevant to computer science related innovation.

> I'm sure people got worse at arithmetic after the invention of the calculator

For LLMs, we can see this sentence but replace "arithmetic" with a variable X

I'm sure people got worse at X after the invention of LLMs"

The problem isn't that X skills atrophy necessarily

The problem is that for LLMs, X is "basically all knowledge and communication skills"

Can we really tolerate a society where "basically all knowledge and communication skills" are atrophying?

Did the entire history of computing prior to 2023 disappear? Are we really that boned?

He means that he's working 10 times faster so the manual things that he does once per task have to be done 10 times, and so he starts automating those too. I'm not claiming that he's actually 10x faster, just explaining.

But it kind of fits my experience too. Less time coding, and more time gathering requirements, testing, and doing knowledge transfer etc. Then you start thinking about how to make those parts automatic via automated tests or automated documentation generation that you review etc.

I think that AI is here to stay, and that you should focus most of your efforts on the parts that AI can't do. It's like with calculators, you still know how to add and multiply and stuff, and such low level things are even more pertinent with AI because it's not deterministic and can make mistakes, but generally you can rely on it for low level coding and debugging and what you really need to focus on is making sure that the artifacts it produces are aligned with your requirements, both functional and non-functional. Think about automated testing, scalability, security, etc. A single developer can produce code of much higher quality with AI because they have the bandwidth to work on such beneficial but optional things that they might have somewhat neglected before when implementation was more time-consuming and expensive.

People like to look at rent control from a purely economic lens, but the sociopolitical aspect must also be considered. A suboptimal economic outcome might actually be optimal when all factors are considered. Social harmony and a feeling of hope for underprivileged residents are hard to value, but we must admit that they do have value.

That assumes rent controlled units actually benefit mostly the underprivileged. They don't.

They create a strong incentive to keep a flat indefinitely. It's not unheard of for a person to continue renting a rent frozen apartment after they buy a house because the rent is so cheap - effectively it's a second home.

There's also plenty of corruption. "Knowing a guy" is one of the best ways to get a rent controlled unit since the wait time can be years.

> "Knowing a guy" is one of the best ways to get a rent controlled unit since the wait time can be years.

That's how military housing works. There's an underground "economy" of favors.

Are these "favors" rank-blind (i.e. does its hushhush-nature more-equalize the men, just trying to secure places to live)? Obviously officers and enlisted have different housing, but might an E-6 kiss an E-5's ass only on account of wanting his soon-to-be-vacated housing?

My background is non-military (with one each, officer & enlisted, brother).

Rank does give one priority, but reciprocal favors still play a large role.

Even if they don't benefit economically, there's some value in having them feel that they are valued and have a voice at the table. Such policies can provide the popular support needed to implement other more economically optimal policies, and we're seeing this happen for example with the mayor of New York.