AstralStorm

The question is, what is the actual paradigm that's being shifted here? Employing people?

It's definitely not wanting to talk to people, and even employment is a bit of a practical joke for now.

Comparing with Internet or electricity here is ridiculous thus far as well.

If we had actual intelligence instead of pretty nifty text robots, then it would be perhaps more true.

Until users find out it's less useful to the user because of that.

Or it causes some tragedies...

Actually they still are making that. And now advanced expensive stuff nowadays too.

After all, it's easier to further an already existing advantage.

But it appears their secret sauce is not dual pricing, but horizontal and vertical integration in manufacturing, combined with low labor costs. Same thing the USA companies did earlier.

Now that is an actual comparative advantage that is actually hard to replicate due to huge entry costs.

(To lower the costs of labor further you need to actually give out life basics for free! China gets away with this by having a huge population competing thus far, but it won't hold.)

Don't worry, economists surely have their calculation of the day.

Previously it was ordinary differential equations. Now attempting to solve these is intractable for such a big and complex system so far, and worse if you misrepresented the model or approximated ot too much.

So essentially all the theories with an equation are a guess.

Doubly so for theories without an equation that are not tied to experimental data... (Hello Austrians)

Only way to achieve that is if people are afraid of these forces and perceive them as effective at policing.

If they're not, nothing will happen.

You can use an LLM as an advanced search engine. The results may vary. Tends to sometimes return hallucinated garbage wasting your time especially for harder queries.

Which are the ones you would want answers to... Everything else is memorized unless you're new.

Usually these things do not have an answer in the search engine either. Or a lousy one like using an insufficiently completed premade library.

These days, even the search will give you AI answers, either directly, via a slop site wasting your time or even on stack overflow...

> Productivity was tracked through metrics such as completed tasks (pull requests), code commits, and successful builds.

Making untested garbage faster to check off tasks quicker. Reopen rate please? New bug task rate? Nobody looked...

> The study also monitored code quality via build success rates. Importantly, increased productivity did not come at the cost of more errors, showing that Copilot helped developers code faster and more accurately.

It builds therefore it works. And the test suite was also AI generated?

Feels like we're back in primary school learning programming.

Classic case of gaming the metrics.

Unless the tool is trash and breaks, or otherwise unfit for the purpose.

You can hammer in nails with your knuckles. Do you want to do that daily? There are people who can, they're still worse at it than a plain hammer or someone even using a right rock. Much less an assembly line.

Oh, let's check how good you are at spotting security issues... Or plain old bugs.

"It's easy to decide if a solution is correct or not." Yeah right. So you have metrics for it?

When you're not rubber stamping things, you will get called out on reducing velocity or perfectionism..

Plus the time you're reviewing is the time you're not coding, so of course you get dinged on your own velocity further.

Ultimately proper review is a career limiting move. Unless you're a pro auditor.

Much like removing gotos automatically makes your code 1% better.

And then someone invented exceptions for the same reason we needed these. More semantics, people still cannot use them.

The semantics being opaque and the thing doing much more than you expected too. And exceptions are trivially simple in comparison!

The funniest part is that you can't get better than the input you put in.

Only when the thing can generalize from a few top notch examples would this problem be solved.

Throwing in a ton of garbage into an automated remixer will produce garbage output that cannot be helped by better classifying it. Even with the 0.01% false positive there is so much trash input the system will learn and reproduce that.

So the future should be developing few shot learning further...

It's written with constant memory allocation in mind. Silly of them to use such a small buffer though, make it a configuration option.

Not really for a small motor cooled by the frame. The bikes are IR reflective too.

So you trust that the client or the centralized service actually does what it says without inspection?

Then you also trust there's no lawful interception on those services anyway?

Adding fake devices doesn't do anything since the homeserver doesn't have your key. It literally can't do that.

If you're sending encrypted to untrusted on a public room, you're trusting the homeserver.

It's literally like trusting a web server that it won't MITM you.

These protocols exist, e.g. on top of I2P network.

Thing is, nobody has any incentive to back them.

Nah, if you run your own identity service, you're supposed to be able to issue any number of unverified identities yourself.

The problem there is that others do not play at all with these, plus actual trust has to be somehow solved.

Typical solutions to trust in DID involve either a big central service, a government approved signature... Or theoretically a distributed web of trust but that bit is under development.

Nothing it is, then.

It seems easy, but actually managing files as a database safely is actually very hard. Especially ensuring atomicity.

Portability is a problem as well.

You do not need an SQL database however, there are others available.

To the OP, when GitHub fails or internet connectivity is down, you're going to be thankful for the local copy that you can later synchronize.

This is what every shared online document platform has problems with. RCSes have the least.

DECT wireless phones and in router voip with a real phone number is rather simple and works. Oh wait, is that EU only or something?

Ambiguous pronoun/declination is not the same as neutral. We have the same problem in Polish.

People will assume the male gender even if it's technically correct.

Yes, allowing code execution by untrustworthy agents, especially networked ones, is fraught with danger.

Phishing an AI is kind of similar to phishing a smart-ish person...

So remind me again, why does an email scanner need code execution at all?

It might even be legible, just encoded in such a way that the monitor would only flag it as unrecognized nonsense of a hallucination and be unable to detect purposeful reasoning inside it.

What it we gave therapists the same interface as ChatGPT?

Mechanical Turk anyone?

They actually tested the human specialists in case you didn't notice that bar in the data.

They are not perfect either, but are statistically better. (ANOVA)

In personal experience, the much stronger and composite type model in Rust makes it easier to refactor.

Adding features in particular is a breeze and automatically the compiler/language will track for you the places that use only old set of traits.

Tooling is still newer though and needs polish. Generic handling is interesting at times and there are related missing features for that in the language, vis a vis specializations in particular.

Basic concurrency handling is also quite different in Rust than other languages, but thus usually safer.