tpoacher

not to mention "conscious"

If anything, this sounds like a toxic person trying to project or provoke.

It reads like a character assassination piece superficially presented as nuanced philosophy, and we don't even have the actual exchange, only this person's account of it, which we are meant to trust in terms of both wording and demonstrated intent.

"If you have nothing kind to say" as the quote goes ...

Give it time. The bubble will blow and it'll be back.

I like how both these projects seem to imply that in the event of an EOW scenario linux users can just f##k right off, thank you very much. xD

Honest question. Since when do we take company "pinky promises" seriously?

Like, does anyone believe Anthropic seriously doesn't train on your data if you ask them politely about it?

Hm, that's a shame.

I suppose you could always convert that message to imply a personal standard rather than company policy, and then defend any criticism coming your way as acting in good faith on behalf of the company and its mission to prevent low quality / potentially dangerous PRs.

But yeah ... not as strong an argument if you're stuck with that.

Having said that, if the higher ups are "all in on AI", you may have luck reversing that logic and having an AI filter all PRs on the basis of "quality", where you get to define what counts as quality, as above. You could even have an automated reply to say this PR failed to meet our AI filters, please revise accordingly (and possibly dump either the guidelines or AI-generated feedback on how to correct the PR before it can be seen by a human).

This way you get to tell your higher-ups that you're using an advanced AI pipeline, while simultaneously filtering out (or at least deslopping as much as possible) most bad-quality, generic AI PRs. win-win.

"Closing as too broad / not focused enough in language and scope / not adhering to our PR submission guidelines (see [link])"

(where the top guideline on [link] clearly states: "we do not accept submissions that are either outright AI-generated or conceptually similar in terms of style, verbosity, generality, scope-creep or signal-to-noise ratio")

Facetious I was.

Bigcorps re-using known names.

A worrying trend.

Good to see Anthropic back alternative OSes.

Or making up stuff

Along with its Hirds!

with the exception that in software there typically isn't an expectation that the client does their own debugging before accepting the software, and any bugs that manifest in said software after purchase treated as the client's fault for not spotting them.

so yeah, not the best analogy.

based on my limited memory and understanding of the incident, when MCAS kicked in, it would make the plane controls more "rigid" in the intended "corrective" direction, as a tactile way of "encouraging" the pilots to comply towards the correction. If a pilot really wanted to override, effectively they could do so by using brute strength to overcome the imposed rigidity, and force their preferred direction.

But then the bug made MCAS kick in on false positives repeatedly and in a prolonged manner, causing the pilots to tire out and no longer be able to overpower the AI to take control of the plane controls.

(lay understanding, and oversimplification of a complex issue, possibly wrong, take with a huge grain of salt)

For similar functionality in nano, if anyone's interested, every time you save (regardless of exiting), nano keeps a backup.

Not on by default, you have to enable it in your .nanorc (not least because you need to provide a backup folder where you want to store these backups).

Very handy. Occasionally I've deleted files accidentally or made destructive changes, and could restore a working version from the past as if I had a timemachine.

Depends on your definition of "controlled". The ambiguity behind the term is doing a lot of heavy lifting here.

Boeing's MCAS system was also "just software". Which in principle can be "controlled", i.e. changed, updated, audited or whatnot.

But then people died precisely because pilots found themselves unable to override or "control" the systems precisely when it mattered.