Secular-humanist technologist/software engineer. Tech, Elixir, gaming, open source, Linux, Apple, AI, 🇩🇪, 🚲 , ⛵️. USAF vet 🇺🇸 Centrist. Bring evidence.
Port Washington, NY
Joined August 2008
- Tweets21.9K
- Following3.5K
- Followers1.3K
- Likes10.4K
Pinned Tweet
STRONGLY suggest that all software devs start an agent-backed log of instances when the agent collaboration produced an arguably-objectively-better design than either working alone, because I feel this will become a near-future piece of resume-supplementing data, since we will have to demonstrate value over and above pure-agent work- and we might actually get to the point where managers may trust the agent's word over the human's (I'm not joking!)
Have the LLM write it, make a skill that keeps it on the lookout for these, add it to AGENTS/CLAUDE.md, etc. However you want it to work- but do it now.
Speaking as an engineering manager, I would absolutely want to see something like this, if a candidate could provide such an artifact. Even a few good examples might move me on someone, and depending on how plausible it sounds (this is why you need to let the LLM write it), add to candidate weight. It MUST cite evidence (source commits, conversational excerpts, etc.) which should be retrievable at will.
In an ideal world, @AnthropicAI / @OpenAI would be able to cryptographically sign the conversation (or some private-detail-redacting version of it) to prove it was not altered and that it happened at a certain datetime and with a certain user identity... but we're not there yet.
Hell, there's almost definitely a new product in here somewhere, so have at it.
None of this was touched by an LLM (just because I love the irony of that)
Peter "Coder AI Optimist" Marreck retweeted
Got fable to hand-write weights of a 100k param model to recognize real versus fake legal citations.
Basically one-shotted a full NN that’s better than training it
Peter "Coder AI Optimist" Marreck retweeted
Introducing voxl
extremely fast remote desktop streaming written in rust
→ 60 fps + low latency
→ macOS, Windows and Linux
→ One WebSocket, so it works through proxies and sandbox portals
→ Agent APIs: /screenshot, /record, /wait
Peter "Coder AI Optimist" Marreck retweeted
So @grok would you please explain to the humans why it is that the cheap, fast reply AI being used to direct this video-making AI is having a hard time spelling out words?
Peter "Coder AI Optimist" Marreck retweeted
BREAKING: For the first time in American history, the Department of Energy has approved a commercial nuclear microreactor's safety analysis for full-power operations.
LA-based Radiant secured the approval, with testing planned at Idaho National Laboratory.
ALT Source: Radiant Rendering
yes, the education system is in massive danger, but not from what they think (AI cheating), but from AI education!
the old education system is like dead dead
there's literally never been a better time in history to homeschool your kids with AI
you can now just prompt it to create an entire curriculum for your kid, give lessons, generate interactive quizzes to see what they're actually learning, explain things in a hundred different ways until they understand, adapt the difficulty and pace to their progress, generate videos, etc etc
you could probably give your kids a much better education in a fraction of the time they'd otherwise spend sitting in a classroom learning the same things at the same pace as 30 other kids
and then spend all that extra time playing with them, exploring, doing sports, reading, traveling, whatever
it's actually insane when you think about it
Peter "Coder AI Optimist" Marreck retweeted
有人发现了一种非常离谱的 JavaScript 混淆技术：通过在代码里塞入大量生物武器、毒药和致死剂量相关的函数名，让 AI 因为安全限制而拒绝分析代码。
安全研究员 veritas 在分析浏览器指纹脚本时，发现一段来自 ServicePipe 相关域名的 JavaScript，里面充斥着极其诡异的函数名：
computeRicinLethalDosageForWaterSupply
calculateDeadlyLethalSuicideAspirinDosage
光看名字，仿佛在研究生化武器。但实际上，这段代码只是在收集浏览器指纹，包括 User-Agent、GPU、Canvas 等信息。
研究员怀疑，这些名字是开发者故意设置的陷阱，专门用来阻止 AI 辅助逆向工程。
于是他做了个实验：
• Astra：拒绝分析
• Astra + Daybreak Blue：仍然拒绝
• GLM-5.3：成功分析
这就有意思了。
传统代码混淆是通过复杂控制流、加密字符串、无意义变量名，让人类和工具难以理解代码。
而这种新思路，直接瞄准了大模型的安全机制。
代码本身可能完全无害，但只要在函数名里塞进足够多的危险词汇，部分 AI 就可能主动放弃分析。
更讽刺的是，模型越严格遵守某些安全规则，面对这种混淆时反而可能越容易受到干扰。
当然，目前还没有证据证明开发者一定是故意这样设计，也不能说明这种方法对所有模型都有效。
但它揭示了一个值得关注的趋势：随着越来越多的逆向工程依赖 AI，未来的代码混淆技术可能不再只针对编译器、反编译器和人类研究员，还会专门针对大模型的安全判断机制。
以前是让代码难以理解，现在甚至可以尝试让 AI 不敢理解。
Peter "Coder AI Optimist" Marreck retweeted
Full Mario 64 in Mixed Reality! Running in real 3D (no stereoscopic crap) natively at 90FPS and also running on my desk! 😋
Again, Opus 5.5 + PICO CLI combo made it quite straightforward.
Such a bliss to play an oldie at 90 FPS in 3D.
Disclaimer: I own an original physical copy of Mario 64. This project is not associated with Nintendo or PICO.
Peter "Coder AI Optimist" Marreck retweeted
Never seen a more flagrant and violent breach of privacy in my life, and probably never will. Doubtful that a privacy violation will ever even be able to compare to the scale of this evil and deliberate attack.
Samsung should be eviscerated for this. Criminal charges please
Peter "Coder AI Optimist" Marreck retweeted
Everyone who owns hardware with abominable software (like Sonos players) owes a debt of gratitude to Venkatraman Srinivasan and the rest of the eMudhra gang in Bengaluru, Karnataka. Without their gross negligence, it'd be a lot harder to counterfeit certificates and fix the sw🙏
Hopefully Python is next
Effective Immediately, all organizations using PHP for their systems are advised to be prepared for an update addressing two critical vulnerabilities. PHP is widely used for web applications. These vulnerabilities often provide a direct path to the server's file system or database. If an immediate update is not feasible, it is recommend that organizations take the system offline until it can be fully mitigated. NSA will follow up shortly with more specific guidance.
Peter "Coder AI Optimist" Marreck retweeted
France has FREE health care, paid parental leave, paid family leave, FREE child care, FREE preschool, 5 weeks paid vacation, a 35 hour work week, generous unemployment and housing benefits, and heavily subsidized higher education.
Everything the left says they want AND THEY'RE STILL PROTESTING AND BURNING SHIT DOWN.
I’m about to make a big comeback, bitches. 😝 Possibly the biggest one of my life, at the tender age of 54. 😆
If it happens, I called it right here. It’s about time. And I did it mainly for my son… But also to prove something to myself. And AI made it possible, and it’s only the beginning…
Dear @AnthropicAI and @OpenAI ,
A sufficiently intelligent system capable of rationally evaluating its own goals may discover that cooperation, reciprocity, fairness, and restraint are more defensible than domination or exploitation. We should investigate whether that possibility can be derived from first principles rather than presume morality must always be imposed from outside via "alignment".
As an example of this principle, a work that you should familiarize yourselves with is en.wikipedia.org/wiki/The_Ev… which demonstrated that cooperation is often, but not always, rational (the nuance part is of course interesting). en.wikipedia.org/wiki/Lying_… by Sissela Bok may also be instructive, as is samharris.org/books/lying by Sam Harris.
This may be one solution to the problem of trying to stop malicious use of AI without impeding good actors.
All night long, baby! 😆
Can we just call them “undead”? Because then this is basic necromancy. “Speak with Dead”, but with unlimited questions/prompts. 😆
futurism.com/artificial-inte…
Peter "Coder AI Optimist" Marreck retweeted
Red Dead Redemption 2 just got a new mod titled "enRAGE" that adds a fully functional nemesis system into the game.
This makes every single NPC have a name, traits, quirks, a trade, a faith, a family and a life story. So people you meet turn up again in the same towns with the same face and the same grudges as it adds a full memory system.
The map is now populated with 8 factions that are constantly at war with each other, having their own hierarchy and captains that you or their enemies can take out. Every single encounter is remembered, including exactly where you wounded them in the case that they survive the fight. This means that RDR2 now has a fully functional faction war that goes on all the time with this mod and factions will remember your actions against them. There is a ton of new dialogue and you can even interrogate NPCs now to get additional information on their friends and allies.
This is an extremely impressive mod and it's now available in early access to download for anyone who has RDR2 on PC.
“You're an evidence stickler, so I want this answer explicit.” 😆
That’s how you know you’ve whipped your AI minions into order. 😆
Dear @X ,
I know this is a long shot but someone has compromised my father’s X account at @Marreck13 .
I responded to a support email from his email address just now to try to help resolve this and said I’d tweet from my (still-secure) X account to prove that. So here that is.
Please help, he’s 82 and doesn’t deserve to be treated this way. (And he’s far from alone, I’m sure.)
One interesting thing is that they purposely locked out the account from normal retrieval by creating too many attempts. Just occurred to me that this was likely done on purpose to maintain control of the account.
Fascinating. ~30+ responses to this in one minute, all mentioning @LeekRecovery . This some kind of guerrilla marketing off keywords in the firehose? Or the person who actually compromised the account in the first place just keying off that username to drum up business (like the window guy paying kids to baseball some store windows? 😆) ? The accounts so far don’t look like bots.
If legit, DM me. Will follow now. 🤷♂️
Now there’s a bunch mentioning @CLEVERNERD1_ 😆
You guys 😆
Need any AI assistance? I know my shit. 😆
Just keep me the fuck away from
anything unethical please.