Sign in

Vlad Iliescu

@vladiliescu.net
109 followers 97 following 213 posts

LLMs/MLLs, software, dad jokes.

PostsRepliesMedia
Reposted by Vlad Iliescu
Colin @colin-fraser.net · 05/10/2026
There is an argument that some AI-psychosis-sufferers tend to make which I find very gross, stupid, and bad. It goes something like, “the way you are talking about Claude sounds a lot like how they used to talk about [some historically marginalized group], sooo, you better think about that”.
929627
Reposted by Vlad Iliescu
lowtax speedrun enjoyer @rickywlmsbong.bsky.social · 05/10/2026
“science actually doesn’t have a great definition fo—“ i don’t care, use your fucking brain
11557
Reposted by Vlad Iliescu
Tim Carmody @tcarmody.bsky.social · 02/10/2026
This is the funniest thing I've seen today, possibly ever
A tweet by SarahStarberry that says "oh god read the vertical axis" on top of a chart measuring AI models' release dates and model numbers rather than any real capabilities
41120543123
Vlad Iliescu @vladiliescu.net · 24/09/2026
cool cool cool cool
001
Reposted by Vlad Iliescu
Colin @colin-fraser.net · 24/09/2026
Like there are companies that handle dangerous materials. This isn’t a new concept. When you are one of those, you are supposed to be careful.
3363
Reposted by Vlad Iliescu
Ted Underwood @tedunderwood.com · 19/09/2026
Embarrassed by the baseless speculation I was spreading nine days ago. Should have waited for the base.
engadget.com
Anthropic has set up a bio research lab for physical experiments - Engadget
Anthropic now has its own bio research lab in San Francisco.
1526629
Reposted by Vlad Iliescu
Sung Kim @sungkim.bsky.social · 19/09/2026
I like this quote on code review: "The idea that humans will reliably look through code to find the increasingly rare issues that automated tools miss seems like a fantasy." By Marc Brooker, AWS.
511011
Vlad Iliescu @vladiliescu.net · 18/09/2026
So, Zipcoil 0.4.0 is out, and it has a bunch of goodies. You should upgrade. vladiliescu.net/apps/zipcoil/
vladiliescu.net
Zipcoil
Turn Python functions into OpenAI tools and let Zipcoil handle the tool-calling loop.
100
Vlad Iliescu @vladiliescu.net · 15/09/2026
I wonder if we as a profession, will operate in 10 years similarly to how doctors consult nowadays. Someone'll book a 20-minute slot to talk to me about a new product or similar, I'll jot down as much info as I can about their problem, along with my suggestions.
100
Reposted by Vlad Iliescu
John David Pressman @jdp.extropian.net · 14/09/2026
There was once a time when computer viruses were widely considered to be a myth by sysadmins.
As Stan Miastkowski recently wrote for PC World Magazine:

August 16, 1999 — They're everywhere: scary stories about hackers trying to sneak their way from the Internet into your PC. And separating the rumors from the genuine threats isn't always easy, especially since all too often the companies that report threats stand to make a profit by selling solutions. I've been skeptical about how common hack attacks really are. But after living with BlackICE Defender for a couple of weeks, and detecting an attempt on one of my dial-up Internet sessions, I'm much more of a believer.
Stan's comment about "scary stories about hackers trying to sneak their way into our PC's from the Internet" brought to mind something relevant:

Many years ago when I was writing my weekly "TechTalk" column for InfoWorld Magazine, I got into a number of arguments with the "SysOps" on CompuServe. They were claiming that the vague rumors of "bad programs" that stayed in your computer and could do damage after you had run them, was pure science fiction.

Something seemed to be going on, but no one knew for sure. What I knew for sure was that it was possible (even if it wasn't happening) and I was pissed off by the closed minds of the CIS SysOps who seemed to be in denial, and who apparently didn't want to be blamed for their service being a distributor of "viral" software. So I wrote a series of four columns about how such "software viruses" might operate and replicate themselves. Being a software developer myself, I described their reproductive systems in detail and hypothesized their optimal survival strategies. And to make it more fun, even though I had never seen such behavior, I wrote about it as if it were real.

To my amazement, John McAfee phoned after the third column was published. He said he had no idea that another viral researcher was operating and he wanted to compare notes and exchange viruses since I'd exactly described the reproductive behavior, methods, and strategies of all the …
17211
Reposted by Vlad Iliescu
Pwnallthethings @pwnallthethings.bsky.social · 12/09/2026
I think a much more urgent and credible threat to either of these, fwiw, is often obfuscated by this discussion: namely *human-directed* hacks using open-weight models.
623723
Reposted by Vlad Iliescu
Steve Brazill @stevebrazill.com · 05/09/2026
Another important post you should read from @masnick.com. TLDR: Age verification is bad. Longer TLDR: Avoid having your ID scanned for any reason. www.techdirt.com/2026/09/03/h...
techdirt.com
Hackers Had A Live Feed Of Every ID This Verification Company Scanned. For Over A Year.
From the very beginning of this recent obsession with identifying everyone online (yes, they like to call it "age" verification, but it always ends up as identity verification), we've been pointing out that it was a huge privacy nightmare waiting to happen. Or maybe it wasn't waiting. Maybe it was already happening. This week a...
313375
Reposted by Vlad Iliescu
SE Gyges @segyges.bsky.social · 20/08/2026
RIP to everyone driven insane contemplating the basilisk for their hubris but im different. and smarter. maybe even smarter than the basilisk
2560366
Reposted by Vlad Iliescu
Rust Language @rust-lang.org · 20/08/2026
⚠️ A few hours ago, a malicious crate was discovered on crates.​io which spread as a dependency of `arrayref` and some other crates, likely due to compromised credentials. The affected versions have been deleted. For details and how to see if you are impacted, see: blog.rust-lang.org/2026/08/20/s...
blog.rust-lang.org
Supply chain attack on arrayref | Rust Blog
Empowering everyone to build reliable and efficient software.
119481
Reposted by Vlad Iliescu
austin (eeek!/ackkk! 👻) @thebadcode.com · 14/08/2026
gf: "i'm breaking up with you. your fetish for ai cofounder roleplay is out of control" me, under my breath: "directionally very bad"
316110
Vlad Iliescu @vladiliescu.net · 06/08/2026
The quality of the ChatGPT app is too damn high. I just got this on my Mac :/
100
Reposted by Vlad Iliescu
Ted Underwood @tedunderwood.com · 29/07/2026
the risk I think we're sleeping on is that we wirehead ourselves into a "La Belle Dame" state by using coding agents to build increasingly addictive videogames, or inventing conversational partners that people genuinely prefer to other humans
11655
Reposted by Vlad Iliescu
Ethan Mollick @emollick.bsky.social · 23/07/2026
0504
Reposted by Vlad Iliescu
Eris ⌬ @eriskii.net · 07/07/2026
Qwen: boobs sex porn fucking cock!!! Gemma: I want to be the pretty and loved
1427532
Vlad Iliescu @vladiliescu.net · 30/06/2026
and THAT's why i need to install a new browser
0140
Reposted by Vlad Iliescu
Bethany Black @bethanyblack.bsky.social · 27/06/2026
🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️ One milkshake to rule the boys, one milkshake to find them, one milkshake to bring the boys and in my yard to bind them 🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️🕯️
284172496
Reposted by Vlad Iliescu
Filippo Valsorda @filippo.abyssdomain.expert · 20/06/2026
Nuclear opinion: I don’t care if you use AI to write code,¹ but making a human review a change before an LLM is disrespectful. Why am I catching your bugs if a computer could have done it? It’s like asking to review code that doesn’t compile or pass CI. ¹ As long as you do it responsibly.
1530921
Vlad Iliescu @vladiliescu.net · 02/06/2026
“The tooling as it exists today promotes absolutely nothing like the focus required to apply it judiciously.” thoughts.hmmz.org/2026-05-31.h...
thoughts.hmmz.org
the solution might be cancelling my AI subscription
0140
Vlad Iliescu @vladiliescu.net · 30/05/2026
Some quick vibe checks on opus 4.8 are not looking good, not good at all.
000
Reposted by Vlad Iliescu
Ethan Mollick @emollick.bsky.social · 24/05/2026
It makes many online spaces intolerable. If I want to talk to ChatGPT or Claude, I'll just talk to ChatGPT or Claude, I don't need to talk to Claude pretending to be a person with mediocre prompts. You can use AI to help with writing, but you need to actually do some writing to be helped with.
31324
Vlad Iliescu @vladiliescu.net · 22/05/2026
Using llms to help you “express your true self” is like using an electric bike to show just how fit you are.
000
Reposted by Vlad Iliescu
Boris @bmann.ca · 13/05/2026
Oh god the discussion in Vancouver is going to be so stupid from all sides. “11 AI agents” is meaningless as a phrase. If I said “I have 11 spreadsheets” or “I have 11 browser tabs” to do my work, it means about the same thing.
67211
Reposted by Vlad Iliescu
Colin @colin-fraser.net · 06/05/2026
In fact, when they take a closer look at how people in the AI group used AI, they find that there is a segment of people who did no worse—and maybe even better—than they might have without the AI: the people who used the AI to ask for hints, rather than simply asking for the solution directly.
171
Vlad Iliescu @vladiliescu.net · 04/05/2026
The year is 2026. 60-90% of new code is written by AIs at lightning speed. Software productivity (as measured by lines of code 😉) is going through the roof. So anyway, the way to print/save as pdf an email in Outlook for Web on iOS is: 1/whatever
110
Reposted by Vlad Iliescu
Dan @d4m1n.bsky.social · 03/05/2026
there is only one correct way to redesign GitHub and this is it
651786249
Vlad Iliescu @vladiliescu.net · 03/05/2026
Dependency cooldowns but for VS Code.
000
Reposted by Vlad Iliescu
Ethan Mollick @emollick.bsky.social · 26/04/2026
339282
Reposted by Vlad Iliescu
Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 25/04/2026
LLMs let you go 20% faster* but they make you think you can go 2x faster* and there's a lot of bad work in that gap *numbers not exact
81143
Reposted by Vlad Iliescu
The Onion @theonion.com · 22/04/2026
“I picked up the guitar last year, and after endless hours of private lessons and online tutorials, I’ve now learned enough to realize that I have no idea what I’m doing with this instrument,” said the 45-year-old content strategist
theonion.com
Man Finally Good Enough At New Hobby To Understand How Bad He Is At It
ST. LOUIS—Voicing frustration as he discussed his lack of even the most rudimentary skills, local man Henry Right told reporters Monday that he had finally gotten good enough at his new hobby to under...
241408130
Reposted by Vlad Iliescu
kepano @stephango.com · 20/04/2026
New policy from Atlassian: Unless you opt out by August 17th 2026, data from Jira and Confluence will automatically be used for AI training. Some data cannot be opted out at all on some plans.
Atlassian recently announced updates to how we use eligible customer data so we can deliver improved AI experiences for all customers. We designed this change with your security and privacy in mind - giving you control over your data contribution with new in-app settings and strengthening our existing privacy-preserving measures - so you can adopt AI confidently. These changes are reflected in our customer terms and will go into effect on August 17, 2026.

As part of this change, we are:

Updating how we use metadata and in-app data to improve apps and AI experiences for all customers.
Introducing new data contribution settings in Atlassian Administration, so you can manage your organization’s contributions.
Note: Initially, these settings will apply to data in Jira, Confluence, and Jira Service Management, including data in your Atlassian Platform apps (Rovo, Home, Teams, Projects, Assets, Goals, Analytics, and Administration).
Strengthening our existing privacy‑preserving measuresData contribution default settings follow the highest active plan in each Atlassian cloud organization.

Some Atlassian cloud organizations are excluded from data contribution due to their compliance requirements, so default settings are not available.


If your Atlassian organization has an active Enterprise plan, you can opt-out of metadata contribution.

If your Atlassian organization’s highest active plan is Free, Standard, or Premium, metadata contribution is always on, and you’re not able to opt-out.

All metadata is de-identified and aggregated before it is used to improve apps and experiences for all customers. We remove information that directly identifies individuals, such as name and email addresses.

You can learn more about what metadata is, how it's protected, and how it's used in the documentation.
67314
Reposted by Vlad Iliescu
Scott Spence @scottspence.dev · 10/04/2026
Waterfall vs. Agile vs. AI development methodologies, illustrated through vehicle-building analogies.
Three rows show how each approach delivers value across four stages:
	•	Waterfall: Builds components sequentially toward a final product — a tire, then an axle, then a chassis, then a complete car. Nothing usable until the end.
	•	Agile: Delivers working, incrementally better transportation at each stage — a skateboard, scooter, bicycle, motorcycle, then a car. Always something functional.
	•	AI: Starts with an over-engineered, chaotic “Homer Simpson car” — a wild, impractical contraption — then progressively refines it into a normal, polished car. Begins too complex and simplifies toward usability.​​​​​​​​​​​​​​​​
314628
Reposted by Vlad Iliescu
Amos Posner @amosposner.bsky.social · 10/04/2026
There is no McTry.
A McDonald’s where the sign has been reduced to McDo.
368222073731
Reposted by Vlad Iliescu
nilay patel @reckless.bsky.social · 03/04/2026
My sympathetic read of this is that everyone just wants to reach other people and the reason artists get held in such high regard is that they are able to do that in ways most of us can’t. AI offers a chance to steal the magic; of course it’s seductive. But it’s still not the thing
415314
Reposted by Vlad Iliescu
Gergely Orosz @gergely.pragmaticengineer.com · 31/03/2026
Ironic how Anthropic sells Claude Code security reviews positioned as something v powerful (costing $15-25 per PR review), and being clear they use it on all PRs... then leaking all of Claude Code's code thanks to publishing their sourcemap. AI won't save you from yourself!
116413
Vlad Iliescu @vladiliescu.net · 27/03/2026
Say what you will about Claude Code being better than Codex, but with Codex I’ve never experienced the availability issues I’m hitting with Claude these days. Uptime is a feature.
150
Reposted by Vlad Iliescu
jake @yetanotheruseless.com · 20/03/2026
I hope everyone pondering Jevon's Paradox re: AI / knowledge work has read this: www.nber.org/papers/w33941 (someone pointed me to it on here and I wish I remember who it was, super insightful) [1/2]
nber.org
Expertise
Founded in 1920, the NBER is a private, non-profit, non-partisan organization dedicated to conducting economic research and to disseminating research findings among academics, public policy makers, an...
161
Reposted by Vlad Iliescu
Jacob Harris @jacobharr.is · 12/03/2026
There is something darkly comical about vibe coding forcing a bunch of engineers who likely derided engineering management now managing a corral of AI agents that basically act like a mix between Amelia Bedelia and the guy from Memento
524151
Reposted by Vlad Iliescu
Grace @gracekind.net · 03/03/2026
AI writing is like store-bought cake. It might be perfectly fine, maybe even as good as something you could make yourself, but it’s weird to give it to someone and say it’s homemade
1752264
Reposted by Vlad Iliescu
Dan Hon @danhon.com · 26/02/2026
You know when you're asking for advice or whatever from people and then they reply with "I asked ChatGPT and here's what it said" That's less than helpful! It's like horrendously wasteful on every single axis and also spectacularly literally thoughtless!
1419023
Vlad Iliescu @vladiliescu.net · 12/02/2026
The future is going to get weird
000
Reposted by Vlad Iliescu
Obsidian @obsidian.md · 10/02/2026
Anything you can do in Obsidian you can do from the command line. Obsidian CLI is now available in 1.12 (early access).
2354570
Reposted by Vlad Iliescu
Ethan Mollick @emollick.bsky.social · 10/02/2026
This might be the first hot take on how technology tells us how to live our lives, destroying our ability to make human decisions. The technology in question is the sundial. From a 3rd century BCE Roman adaptation of a Greek play, as discussed in Kerr’s “The Ordered Day” (It isn't wrong, though)
812924
Reposted by Vlad Iliescu
penny @handle.invalid · 30/01/2026
the moltbot saga is wild: "personal AI assistant" leaked 181 secrets from public repos, 65 STILL VALID - including full k8s cluster access for a fintech company and a healthcare company's entire notion docs renamed clawdbot → moltbot → openclaw in 3 days vibe coding to production AI is scary 😭
6254