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."
It carries the heading "Engineering that's evolved and amplified for the AI era," and for any AI skeptics in the developer teams at Red Hat, the tone of the email may raise alarm bells. The times are changing, it states.
https://www.theregister.com/2026/03/31/red_hat_ai_dev/Open linkView original on reddthat.com368
Comments27
Duh, it’s run by IBM. The most brain-rotted management suite on earth. All they do is chase the cool new hotness, and unfortunately it works for them – they’re mostly selling to other brain-rotted manager types. (The end users, as usual, get hosed.)
For some reason I thought IBM didn't exist anymore lol. What a horrible day to find out they do.
They exited the PC market a while back. They still make enterprise stuff.
They own Red Hat and IIRC, it's their fastest growing (if not largest) business unit.
It's like they don't understand the reason for this success is how different their road taken has been compared to all-in on AI companies.
IBM like Apple has been cautious about generative AI Not to say they don’t, their Granite models work great for personal machines.
Choice is always key, embedding it in the OS is a terrible idea.
The "scale the delivery of value to customer" phrase is a massive red flag, it means they have no fucking clue what they are doing. If they did, they would be more clear about their thinking and reasoning and not use PR speak.
They do provide some more specifics that are quoted later in the article, but it still sounds like they haven't really thought this through:
What's not clear about "delivering value to customers"? I think it is very specific to [product] that [company] delivers!
"Value" is a term so ambiguous, it's actually worse than not saying anything.
There's some value in your words here! 👍
"We will move entire products to this model simultaneously" ah yes, the All-in strategy that's very safe and prudent like in poker... And yes, the rest of the memo sounds like the usual marketing team and execs auto-fellating each other, to the dismay of everyone else.
Well auto fellating is the trademark tone of LLM output, which the email probably is.
The real concern here is that they intend to push velocity. Senior devs are already struggling to keep up with vetting these velocitized changes and are sending out warnings that quality and security will suffer. IMO the tech isn't mature enough to run unsupervised and transforming senior devs into code reviewers is a big mistake.
Can the Ai just use rust?
you just add "do not put bugs in the code. review it as much as you need to." to the prompt.
Sadly I can't tell if this is a joke or not because I have met so many people who seriously believe things like this work. They are the ones who eventually get the most pissed when LLM messes up on them because they got the LLM to "promise" not to do the specific thing it ends up doing.
They generally evolve their superstitious ritual to something else that will eventually fail, like changing the wording, or making the LLM specifically include a phrase indicating a promise of quality. They also believe when the LLM "apologizes" and think that indicated self reflection and learning. Very few are prepared to accept that the LLM can go off the rails at unpredictable times and unpredictable circumstances, and their utility has to be monitored like a hawk unless the outcome really doesn't matter.
lol, but not completely inaccurate. "You are a pedantic code reviewer. Review the MR you just created"
Cooooool 😐
How long until they get more hate than Canonical?
Anyone surprised that enterprise software is doing this?
I bet they're just using GitHub Copilot in VS Code. Big whoop. lol
Hopefully it's just AI tools for development they're talking about (though that will be bad enough if RHEL becomes vibecoded slop) and not stupid AI "features" baked into the OS.
I think the inverse might actually be preferable. If there's slop in the code base, that will be harder to avoid than whole modules that you can just not install.
While neither is preferable, putting it in the development is more insidious.
Unfortunately, that ship has sailed.
Ugh. Come for the systemd, stay for the slop.
That is unfortunate. I don't like everything about Redhat's products, but they do have some very useful tools.
What more would you expect of an IBM subsidiary?