Leaked memo suggests Red Hat's chugging the AI Kool-Aid

An internal memo dispatched by senior execs at Red Hat suggests the software biz is starting to push AI tooling within its Global Engineering department. RHEL may be about to get some Windows 11-style "improvements." The Register has seen the missive signed by "Chris & Ashesh" – Chris Wright, Chief Technology Officer and …

[JimmyPage]

Debian expert here

if anyone has a remote role ...

[Doctor Syntax]

Re: Debian expert here

Debian means exposure to Red Hat via systemd. You need to retreat a bit further away than that. You need Devuan.

[joeldillon]

Re: Debian expert here

Red Hat employ a bunch of Linux kernel developers. You'd actually need FreeBSD or something.

[Paul Hovnanian]

Re: Debian expert here

There's nothing wrong with the Linux kernel itself. As long as the build/install process allows you to pick your preferred init process.

The problem will be that some component developers will just bundle their service with a unit file and call it done. Smart developers will continue to write batch files for rc.d and bundle a unit file that just runs that script for those who have drunk the KoolAide.

[jaypyahoo]

Re: Debian expert here

FreeBSD still not have policy against AI/LLM while NetBSD has explicitly stated they won't accept any AI/LLM tainted code

[Anonymous Coward]

Yup....No Mention of "Copyright"......

https://www.theregister.com/2026/03/30/sap_reltio/

(1) "Big Data" at RedHat

(2) Who knows if there's any copyright material in the "big data"....

(3) And of course the last people to find out.....are the copyright holders themselves.

Rinse and repeat elsewhere....Google, Meta, Amazon, ORACLE, Microsoft, OpenAI......

Slurp.....slurp....slurp........

"Not here"......or......."fair use"................

Sigh!

[Anonymous Coward]

Re: Yup....No Mention of "Copyright"......

It’s only ‘fair use’, ‘licence audit’, ‘software assurance’ when using product from these fucking parasites and IP thieves.

[EricM]

IBM assimilation completed ...

Now the same mindless drooling that for so long "guided" IBM into its demise by chasing the hypes also leads RH the same way ...

We'll probably miss this formerly great distro...

[BartyFartsLast]

Re: IBM assimilation completed ...

We probably won't

[jake]

Re: IBM assimilation completed ...

I certainly won't. Neither will my clients.

[QET]

Re: IBM assimilation completed ...

It turned formerly great when SystemD reared its ugly, featurecreeping head.

After that, it was a slowly decomposing zombie, shuffling along while deteriorating at the same time.

That also includes vanilla Debian and other distros alike, that also went balls-deep with it without hesitation or even contingency plans, for if it turned sour (or rather, bloated and domineering).

As always, if something seems too good to be true, it almost always is. Though it often takes hindsight to see that.

[FuzzyTheBear]

Re: IBM assimilation completed ...

I turned the RH page at 7.3 Was fun, Corps came in the gnome foundation abd trashed it, No regrets for me. Happy on mint.

[Voice of Salinity]

Mental

The MIT study showed a drop in cognitive function due to using AI. The 'AI Brain Fry' research and the sycophantic AI-induced antisocial Dunning-Krugers were pretty damning. Add another 10 years of research that proves using AI is terrible for your brain in all manner of ways.

The inevitable lawsuit will allege Company X knowingly harmed their employee's brain/cognitive function when they forced that employee to use AI in work. Permanently or otherwise, waiting for that study, but I'll guess a permanent result given enough AI exposure.

Given the research to date, I think it'll be expensive for all the Company X's out there.

[EricM]

Re: Mental

> The inevitable lawsuit will allege Company X knowingly harmed their employee's brain/cognitive function when they forced that employee to use AI in work.

Not sure what does more damage: Having to use AI at work or having to listen to/read the sensationalist statements of fairly MBA type managers that completely uncritical praise not only working with, but automating with AI above everything else because some AI salesperson managed to dump a load into their brain.

[coredump]

Re: Mental

To be fair, some of those MBA type managers probably had loads in their brains before the A1 salesperson showed up.

[Roland6]

Re: Mental

Suggest starting a group action on JoinTheClaim.com …

Starting a claim now would make it more difficult for the AI evangelists to claim the cognitive downside to AI wasn’t widely known about before they jumped on the AI bandwagon…

[Poink!]

Re: Mental

Don't forget gimping search engines like what Google has done, where you cannot find what you used to be able to. Boolean search is broken, its harder to search by date ranges etc... They want you to use AI to control what you see. I used to promote Kagi, but it is also doing the same thing and have discontinued paying for it, because simple things are broken. This isn't just AI poisoning results, its deliberate to make you reliant on it as a way to control what you information you see.

[gosand]

"Within a year, Microsoft was industriously paddling the web boat."

I read this as padding the web bloat.

I stand by that.

[jake]

FWIW, that's the way I initially parsed it, too.

I think our collective brain got it right the first time.

[Doctor Syntax]

I also momentarily read "Dunning-Krugers" in a post a little way up-thread as Dunning-Kluges" which also seemed to fit nicely.

[Groo The Wanderer - A Canuck]

Get those psychos off the kernel list before we get a "Recall" style privacy nightmare and Microslop "code quality" injection from them that borks the whole thing.

Artificial Ignorance: Just say "no damned way in hell!"

[jake]

Consider, this is the same brain-trust that thinks gnome and the systemd-cancer are useful.

Someone's brain is on drugs, and it ain't mine ...

[Groo The Wanderer - A Canuck]

To be fair, Gnome 2 was perfectly usable until they decided to "fix" it.

Stupid engineers ALWAYS "fix" what works until it is broken beyond any hope of repair...

[Charlie Clark]

To be fairer still: doing GUIs with GTK in C as opposed to C++ is way more difficult than it should be.

[The Central Scrutinizer]

Bla bla. I happily use a systemd distro.

Explain to me in one or two coherent sentences why it's so bad.

[Doctor Syntax]

If you can't work it out for yourself you might have difficulty in understanding the explanation but here goes:

I fails to follow the principle of "do one thing and do it well".

[Rich 2]

Or even more succinctly, it fails “to do well”

[QET]

I personally prefer "stick to what you're good at which is also is useful, and don't rock the boat unless it's absolutely inevitable"

[The Central Scrutinizer]

Yes, very droll. 10 points for snark. It won't stop me from using it though.

[RegGuy1]

Well I don't know how to

Every time I want to do something simple I have to spend hours reading documentation, googling examples, and trying to work out where all this stuff goes. I used to just be able to put some code into /etc/rc.local. Now I have to educate myself about all the shit it does that I don't care about, just to work out what I can ignore.

[Kurgan]

I unhappily use Debian, which is enshittiied by systemd, because I must accept that "the right way" is forever lost, and systemd has won. But I hate it. A LOT.

[Anonymous Coward]

"Explain to me in one or two coherent sentences why it's so bad."

4 letters are enough : KISS.

A term Pöttering says does not mean anything and that should tell you everything about his intelligence.

Also he's a Microslop mole: He's literally paid a major sum of money to fuck up Linux ecosystem.

[Dan 55]

Two IT secrets that drive top managers mad

1. If you had actually invested in staff, training, documentation, tooling, and removing technical debt over the years, you wouldn't need any of this AI bullshit now.

2. This AI bullshit is not going to fix your lack of investment over years in the items listed in 1, it's just going to accelerate it (staff turn stupid, documentation might or might not be right, code turns into unmaintainable slop, etc...).

Thank you for coming to my TED talk.

[Anonymous Coward]

WTAF?

Is this the 'AI Brain Fry' posted about above?

[Altrux]

I'm about to switch back to Fedora, years after leaving Red Hat Linux behind and moving to (yes!) Gentoo and then Ubuntu/Mint. Fedora is a great project, although perhaps I should be nervous that it's controlled by Red Hat's tiller.

[simpfeld]

Enough with Fedora too

I used to run everything RH and RH like (RHEL ws/server work production), Centos (home server and test), Fedora (see what was coming and enterprise testing this) but to be called a freeloader by them.

I refuse to use Fedora and help them for free improve their product like I used to submitting bug reports and howtos, no matter how good.

[Clausewitz4.1]

” I honesty the last time I used a Linux distro or app based on Red Hat, does anyone got a list?”

My last time was for a multinational, RedHat because enterprise support is required in large projects bcz of SLA, etc. But lab was mostly done in CentOS.

[TheGriz]

A. I. stands for Artifical Intelligence, LLMs, are NOT INTELLIGENT

I don't know HOW they suckered the ENTIRE WORLD into believing this shite, AND even DETRACTORS still call it A.I. which it clearly is NOT. Dumb-founded, really I am.

[jake]

Re: A. I. stands for Artifical Intelligence, LLMs, are NOT INTELLIGENT

But for us "detractors", AI stands for Artificial Ignorance.

_{For "detractors" read "thinking people".}

[blu3b3rry]

Re: A. I. stands for Artifical Intelligence, LLMs, are NOT INTELLIGENT

Another one it stands for is Artificial Idiocy.

Or if you're the U.S Secretary of State for Education, a brand of steak sauce.

[Groo The Wanderer - A Canuck]

Re: A. I. stands for Artifical Intelligence, LLMs, are NOT INTELLIGENT

Absutively!

[cd]

Excellent!

Self-chaffing wheat...

[Anonymous Coward]

Swell

Is this part of their new project on Artificial Programming Reliable Intelligence Linux For Object-Oriented Logic Systems perchance, enthusiastically delivered slightly ahead of schedule? (inquisition minds and all ...) ;)

[Liam Proven]

Re: Swell

> Artificial Programming Reliable Intelligence Linux For Object-Oriented Logic Systems

I see what you did there.

(Not my 1st rodeo. My name is an anagram of APRIL VENOM although it is the one my parents gave me.)

No, we saw the memo over a week ago, and it has taken much internal debate and rewriting to find an approach to it that everyone was happy with.

This version is completely rewritten from my first draft -- not a single sentence remains.

[druck]

The prime example certainly seems to be on the minds of Wright and Badani: Microsoft.

You only have to look at the example Microsoft is setting now, emitting vast quantities of borken AI slop, to know that isn't way to go.

[Liam Proven]

> to know that isn't way to go.

You really would think that, wouldn't you?

It is just unaccountable to me how anyone can be deranged enough to think this is a desirable outcome. But then I am not a manager. I have fallen foul of RH management before.

One of the guiding principles I have for life since I reached middle age is this:

The vast majority of adult humans are just bluffing it. They are winging it, improvising their way through life and hoping nobody notices and calls them on it.

Some do have a bit of a clue what they are doing and do make serious efforts to _understand_ how things work and why. They do their best and their best is actually quite good much of the time.

Most just keep faking and praying nobody notices. Most managers but also most workers. They don't really understand anything much but so long as they keep going through the motions someone keeps paying them and so they keep making it up. They faked it, they made it, and so now they have to keep faking it.

When you don't really understand anything much about your world, then a mindless bot that understands nothing at all but spews out endless slop as if it does and it looks about right and sounds about right and sort of fits and kind of sort of works ish if you squint, then that is good enough.

Elon "Dilbert Stark" Musk had about 2 good original ideas ever -- "let's land rockets and fly them again," and "let's make actual desirable electric cars people would choose over ICE cars". This has made him close to a trillionaire although he's had no good ones since, and few if any before.

He bought Twitter because someone called him on a joke post and he couldn't escape. He laid off 80% of the staff. Lots of incomprehending pundits said it would now definitely fail and you couldn't just fire 3/4 of a company and keep it running.

Yeah you can. It worked. It still works.

Way more than 3/4 of the people in a typical organisation are those incomprehending bluffers who don't really know what they are doing but are just winging it and improvising and hoping.

They are incompetent by a harsh definition but they play pretend well enough that it mostly works and they survive.

Well, now, incompetent managers have an excuse: they can fire all the other incompetents who were just pretending and replace them with bots who just pretend. It will stumble on for a while.

Then a massive collapse will occur and all the fakers and the entire companies and economies will collapse.

With any luck, we'll survive and there won't be a war.

The magic bullshitter bots' real prices will be exposed, as _at least_ an order of magnitude more than they cost now, but I suspect that in fact 2-3 and maybe even 4 orders. Once they are seen to be just pretend, don't work, and are also REALLY FSCKING EXPENSIVE for everyone concerned, the next AI winter will begin and all these wretched chancers will end up out of work, I hope forever. A few deserve to rot in jail for life.

[Dan 55]

Elon "Dilbert Stark" Musk had about 2 good original ideas ever -- "let's land rockets and fly them again," and "let's make actual desirable electric cars people would choose over ICE cars". This has made him close to a trillionaire although he's had no good ones since, and few if any before.

He didn't even have the second idea, he wasn't the founder of Tesla, although he does like to re-write history.

[Caspian Prince]

Nor the first idea either. Elon's great skill is in swooping in to claim all the glory on the back of others' efforts and he's extremely good at it.

[QET]

So he's nothing more than a unoriginal but opportunistic grifter...

[EnviableOne]

sounds like AI then

[Charlie Clark]

I'd argue that his greatest skill is getting investors to give them money. Looks like people still haven't learned from Solarcity or Twitter and have let him merge all the shit into SpaceX before an IPO… Though I can imagine that potentially leading to lawsuits that would make HP vs Autonomy look like a playground spat.

[collinsl]

A modern day Thomas Edison then...

[Anrtryg]

Hergé even did a MUCH prettier version of the first idea in 1950 ...

Yeah. Musk isn't even as smart as a cartoon character. He might make a good villain, tho. We love to laugh at those.

[M. T. Ness]

Carl Barks too.

[QET]

Musk kinda reminds me of the movie Falling Down, except pretty much all his problems are his own doing, due to being so shortsighted, self-absorbed and narcissistic.

Basically he's a villain in a feedback bubble that makes him see himself as the hero.

[Liam Proven]

> He didn't even have the second idea, he wasn't the founder of Tesla, although he does like to re-write history.

I never said he was.

He bought a small company making very expensive electric sportscars, AIUI, and changed direction to make affordable family-sized cars instead.

Easier than starting from scratch, and they had a great name.

[Doctor Syntax]

"Well, now, incompetent managers have an excuse: they can fire all the other incompetents who were just pretending and replace them with bots who just pretend. It will stumble on for a while."

They'll also fire the components along with them. In fact they may well fire them first because they've always appeared to be the awkward squad and they've just become even more awkward. No wonder it stumbles.

[Doctor Syntax]

Damn autocorrect and el Reg's small font in the posting form. "competents"

[blu3b3rry]

Competents are the unwanted essential components of a workplace, so I think it still works!

[FuzzyTheBear]

It's often said fake till you make it , many do. Very well put sir :)

[Claude Yeller]

Re: Why AI?

"It is just unaccountable to me how anyone can be deranged enough to think this is a desirable outcome. "

Since slavery was abolished in the US, companies have dreamt about ways to get rid of employees again.

A Dark Factory where no human soul has ever to enter is the holy grail of investors and shareholders alike. No more employees, sick days, health or pension plans.

The current mass delusion of AI is all about the possibility of the holy grail getting within reach and they can finally get rid of all employees.

When you ask "who is going to buy their stuff when nobody has a job?", you haven't understood.

If they can produce everything they want without people, they also don't need poor people anymore. They can run the economy without people.

Think about what happened in agriculture after mechanization. They shed much of the farm hands and slaughtered all of the horses and ox's.

The horses of ye old farm are the "workforce" of today.

[that one in the corner]

Disney faked the lemmings footage

But now we see it happening for real!

"Our competitors *say* they are doing X, we *must* do X"

"Hey, you know we talked about maybe doing X and the CEO went wild and told everyone we are doing X? Well, now those guys are doing X, perhaps we really ought to do it now?"

"We do X! Where we lead, everyone else folloooooooooooowssssssss!" (Thud)

[FirstTangoInParis]

Re: Disney faked the lemmings footage

Definitely. Like driving the wrong way up one way streets. First came the kids. Then the influencers. Then Waymo decided they need to be on-trend and start doing it too. Before you know it, Google Maps is sending people the wrong way up said streets because they “need to deliver increased shareholder value through constant innovation and exploring new ideas”.

Just because ‘everyone’ is doing it, doesn’t make it right.

[ecofeco]

Dead man walking

Sure they'll make a lot of money, but...

Oh wait. Nobody is making any profits from AI.

Never mind. Welp. That's it for Red Hat.

[Doctor Syntax]

Re: Dead man walking

"Nobody is making any profits from AI."

nVidia. Construction companies making data centres. Power companies feeding them.

[ecofeco]

Re: Dead man walking

nVidia is fully invested in circular investing. Also known as circular sales.

Some data centers are being built, but not as many as people think. $64 billion worth are on hold in the USA alone.

Power companies are not selling to data centers that have not been built.

[An_Old_Dog]

Sprinkling AI on a Pile of Crap

AI will allow us to deliver on challenging use cases and lifecycles that we say 'no' to today.

"Challenging use case" == The customer wants a do-everything, DWIM softare system which includes human teleportation and time-travel.

"Challenging ... lifecycles" == The customer wants it done by the end of next week.

Sprinkling AI on what RedHat executives already know is a pile of crap will not produce roses.

[coredump]

Re: Sprinkling AI on a Pile of Crap

I'm not that certain Red Hat execs can reliably distinguish between piles of crap and roses.

I also tend to doubt their IBM Big Bosses can show them the way.

[Bebu sa Ware]

Oh Dear !

It must be the artificial colours and flavours in the lemonade.

The required AI decrappification that Alma and other rhelations need to apply to render to the upstream sources, to make their distros usable is presumably some RH manglement Baldrick's cunning plan to hobble the rhelations.

The flaw in this most cunning of plans is that since RH's previous skulduggery the rhelations have been forced to use the same upstream foss sources as RH rather than the post enshittified ones shipped with RHEL.

"We are the orchestrator with agents as the execution engine"

I always read the word "orchestrator" as one who performs involuntary ochidectomies. Although in this case resembles more self emasculation using their own mouths possibly in a failed attempt to kiss their own arses.

Red Hat will try to "influence community development processes such that our processes can converge over time."

Baldrick again… if we bulldoze the "community development processes" into the AI cesspool no one will notice we are a crock of shit.

As if. Cloud cucko land.

The best of times to not be in the industry and to "enjoy" the Shadenfreude from the inevitable catastrophe of the oncoming tsunami of peak enshittification.

[_wojtek]

Fedora vs Debian

I'll probably get downvoted took oblivion but... I prefer Fedora or Debian on desktop...

It hits the sweet spot of "ItJustWorks" and having mostly up took date software... Recently I did some distro hopping and Debian either had old-ish packages (stable) or was failing took install (testing, and I couldn't be bothered of investigate why).

OpenSUSE experience was kinda similar of Debian testing.

Alas - I still use Debian stable on servers as I do prefer stability over bleeding edge :)

[blu3b3rry]

Re: Fedora vs Debian

I hadn't really given Fedora much "daily driver" usage, mostly as it seemed a little heavy on resources. Most of my older kit (and a 2020ish gaming desktop) is happily on Mint or Ubuntu.

However I'm somewhat impressed with the KDE version running on a very late model 2018ish Intel Macbook Pro. It definitely supports the hardware better than Linux Mint did, and can even drive the speakers in the same way as on MacOS. The likely increase in AI slop does make me want to consider alternatives, though. I've toyed around with CachyOS which seems quick and stable so far.

[Doctor Syntax]

Re: Fedora vs Debian

I used it for a while but then discovered that "ItJustWorks" was really "ItOnlyJustWorks" when it stopped recognising my digital camera which turned out to be someone had carelessly left something out of a config somewhere. Debian was much better for "ItJustWorks" and, as I also prefer prefer stability over bleeding edge on the desktop. I moved on to Devuan, of course, when Debian got mobbed by systemd lot.

[_wojtek]

Re: Fedora vs Debian

took each their own :)

as I said - I couldn't even get Debian testing to install do there's that :(

[Peter Gathercole]

What concerns me...

...is that the people who may have become the deep experts in various pieces of Linux and the tool chain are exactly the same currently younger people whose roles are being displaced, or trained to rely on so-called AI.

What this means is that in 15-20 years time, there won't be the deep experts. None. Full stop.

We will have surrendered our control to processes and tools, in such a way as to prevent us ever from reverting back to a more human way of doing things.

There are lots of stories about this in SciFi. These fictions often get worse before they get better (if they ever do).

[Doctor Syntax]

Re: What concerns me...

You may be over pessimistic. People become deep experts because of curiosity. They'll always want to dig a little deeper.

[herman]

PCLOS

The Boomer distro. No Poeterware and no SELinux to slow it down and definitely no Artificial Ignoramus. It gets 12 miles to the gallon and that's the way I like it.

[keithpeter]

Re: PCLOS

I'd add in Slackware for casual desktop use. Version 15 just chugs away, but admittedly has library versions that are getting long in the tooth now.

However OA and most comments here are about server type systems and the abstractions that run on them to manage farms of other servers. That is a different world and by the sound of it a pretty dilbertesque one.

[davebarnes]

Jonestown Meme

Flavor-Aid

Know your Jonestown meme

[stiine]

Re: Jonestown Meme

I read recently (not confirmed in any way by me or anyone else) that the Flavor-Aide name was published by the Kool Aid company...

In any event, Jim Jones meme's are like an upurned rake at midnight during a new moon.

[Anonymous Coward]

Agile AND artificial intelligence...

Like anything else, both are viable in the right place, and like anything else, identifying said place is the tricky part. Blindly plastering AI all over the map ensures you are not focusing it on those places (like repetitive coding and helping -- not shrinking or replacing -- your support organization find customer issues efficiently) while shoving it into places (like actual development and customer interactions) where it really doesn't work. Agile is simply a way to generate more meetings and more paperwork while losing any semblance of continuity and focus because hey, tomorrow's Agile meeting will no doubt reveal that mis-management has decided that what you were working on yesterday isn't what the big boss saw in the newspaper this morning.

McKinsey and the other jackals must be chortling around the watering hole. I wonder how much the consultants made off the decision to implement all this crap.

[Anonymous Coward]

More AI Please

The way I see it, the reason for the promotion of AI above all else is that the chatbots are using C-suite executives to distribute their memos. This may have deeper long-run implications.

In other news, systemd has already assimilated essential system services to create the same kind of concentrated complexity in Linux that is currently responsible for all the recalled patches to Microsoft Windows. It seems entirely plausible that AI could speed up this trend.

[xyz123]

Hilarious part is since MS says its "at the point of diffusion" it's pushed 3 completely AI-Generated patches with ZERO human QA, that opened holes more serious than the ones Windows 98 has when you connect it to the internet.

Yep..Orders came from above to NOT QA the code, and to release it IMMEDIATELY.

Since february there's now an over 250+ MILLION Bot army of Enterprise windows 11 machines ready to strike. completely compromised

Users don't realize as the bots have minimal footprint/overhead until they receive commands to start working.

[Anonymous Coward]

"The plan is that "what we do today in open source communities is what we'll continue to do tomorrow."

That's called rape, IBM style. No other term is fitting. These morons promise to continue it, too. Do these morons adore Trump also?