Believe it or not… yes. /s

The only thing from the OS that could ruin my day would be a botched kernel update. And that is countered by using the LTS kernel.

KDE updates bring new features but don’t ruin my workflow. And if they do I just roll back to the last snapshot and wait for a fix; never happened yet.

Dev toolchains are versioned by repo anyway. Services run in containers. IDEs are installed with their own version manager.

Then he’s stupid enough to buy into such shit. “Yeah but look, Ubuntu comes as binary and Microsoft even offers tools to attest its conformance.”

Only downside is that you have to use one of the listed distros for your argument; so pick the one that will bother you the least. As an ArchLinux (btw) user at work, if I had to switch to Debian or RedHat based I would be a bit pissed; but still better than Mac or Windows.

Even if you end up installing it… I would take a Linux with Endpoint Protection over Windows with Endpoint Protection or Mac with Endpoint Protection any day. If this is the hurdle it takes for IT departments to allow using Linux workstations, then yes please.

Normal text based analysis; which LLMs are pretty good at. So our coding agents can follow the rules during development and the review agent can follow it during review. Only static code checks are enforced in the CI pipeline; the review agent only leaves remarks. If we decide to not give a damn, we can still merge it. But its hit-rate is relatively high; so typically when it finds something, it’s valid and its recommendations for fixes are also often pretty good. This helps already iterating on a PR on your own before you pull human reviewers in.

Not all guidelines are enforceable with linters. For example one of our code bases at work defines when to avoid mock tests, how integration tests should look like, how a module’s architecture is supposed to be, how the naming convention for components of the different layers are, etc.

Those are super ugly to formalize into a linter but super easy to check with an LLM.

Interestingly I also think it’s suicide for capitalism as we know it. Ultimately the system works by having people earn their share by working, and using their income to consume various things (because you want and need more different things than you could possibly ever create on your own). Companies exist because people buy their stuff, and people can buy stuff because they work for it.

If LLMs end up helping build more with less labor while the system stays unchanged, companies get into trouble: while they produce more, less and less people consume their products. So what are they producing for? And how do they make enough money to warrant the production in the first place?

So something has to change anyway. No matter if capitalism or communism. I just don’t see how a fight against LLMs could help, because you can’t wipe away the concept or the means to apply the concept without wiping everything. And bringing upon The End is not a feasible solution to prevent a downfall; it’s merely speeding it up. So I would rather “watch” what happens and hope the system adjusts like it “always” did.

my argument is that these tools are anti-worker at their core

But do we even want to be workers? I don’t. I have to be worker.

We shouldn’t be fighting parts of a broken or rather unfair system, but the system. I don’t want to fight for my place at the bottom of a pit.

These risks are not different from before and still open source has the advantage of being open to anyone and any agent.

Kernel modules written by US employees with unknown ties to US intelligence was just as much at risk for containing a backdoor as if written by an agent using an LLM with US intelligence ties.

And in both cases intelligence actors or security researchers from around the world can review it all and try to spot backdoors. So can agents with different LLMs behind them.

It would be an issue if there was only one state sanctioned LLM available, but as long as there is world wide competition in training LLMs, opensource will be “reviewed” with competing interests in “mind”.

But why does this need any new legislation? It doesn’t matter if the agent is air gapped or not, as long as it doesn’t do illegal stuff. Hacking is illegal unless you were tasked to do so for security reasons by the target.

As long as judges don’t start treating agents as independent persons, their operator should be on the hook for everything they do. So if you fail to safeguard your agent, and your agent performs an illegal action, you are liable. That should be deterrence enough.

But isn’t that the problem? You can’t force a rogue user to do anything; or they wouldn’t be rogue. So why should they apply air gapping to their agents?

Very possible. They rely on good translations for example, and I would assume Yandex was one of the services that offered this. Or at least it might have run in the same data centers.

Without trustworthy translations, you can hardly influence as easily.

And even if there are alternatives, the existing data pipelines need to be reworked first (new APIs, no contracts, etc)

I think he cut them off because Donny wanted it over Ukraine not wanting to be in on his super cool “deal”.

Yeah and certain password requirements are also dumb. But at least they still give me a degree of freedom how to generate the password and how I store the password.

Second factor is sometimes worse than Passkeys; with vendor specific apps (Steam Guard, Blizzard Authenticator, Microsoft Authenticator, etc) instead of following standard TOTP. But still. I want agency over my credentials.

White is probably a bad choice for a car that any real human is going to use.

It looks more like beige.

I drove an Audi A4 with light grey / silver fabric interior for 15 years. It was 18 years when I sold it and still looked mostly like new. Only the drivers seat had a few slightly darker spots that I could likely have cleaned with a wet-dry vacuum cleaner.

I didn’t have kids in the car during that time, though. But I guess neither would whoever drives Jaguar.

She sells sea shells by the Seychelles.

I only dislike the ability for service providers to enforce a “security level”. Those should be the exclusively under my (the user) control. Sending preferred default (hints) would be cool, but not enforcement.

That a webpage can basically prevent me from using a passkey if the passkey is stored in software (a password manager for example) is complete bullshit. Especially if there is still username/password as alternative which is clearly worse. Sure, hardware tokens are the ultimate security. But it should be up to me and not them if I want that level of security.

I would say yes. I have colleagues who have a hard time solving some tasks with an agent. Then I take a stab at it and succeed almost immediately. We both have the same freedom to choose and customize the available tools, but apparently my setup and usage is better.

I think “no AI” policies are a good thing and the intention is (likely) not literally forbidding AI; they couldn’t in any meaningful way. But the code should have a human attribution and ownership. If the code reeks of LLM, it’s shitty code quality. If people open MRs saying it was written by an agent, they distance themselves from their contribution but expect others to review and integrate it; which is also shitty behavior.

If on the other hand it’s well crafted code that was being thought through and refined, no one should or could care if an LLM was involved.

I generally like their outcome, but hate the boilerplate. I think I need somewhere around 20 lines over two different files which would otherwise be a one-liner in a cron file.