Allan @niemerg.bsky.social · 08/07/2025You don’t fear a gun; You fear a man with a gun. Likewise, you shouldn’t fear AI; you should fear dangerous men with AI. 080
Allan @niemerg.bsky.social · 07/07/2025With tests, lawyers could finally trust AI assistance because they'd have systematic verification - not just for document creation, but for document correctness. So I’d love to know: What would you want to test in your contracts? fullydoxxed.com/to-use-ai-la...fullydoxxed.comTo Use AI, Lawyers Should Steal This From Software Developers | Fully DoxxedSomething remarkable has happened in software development over the past few years. Large language models have revolutionized how code gets… 050
Allan @niemerg.bsky.social · 07/07/2025Write tests once, apply them to every document of that type. Hallucinated case citation? Test catches it. Violates state statutory requirements? Test catches it. Wrong client entity details? Test catches it. It's like always having a fresh pair of eyes on every document. 150
Allan @niemerg.bsky.social · 07/07/2025Legal tests work just like software tests: - Run automatically when documents change - Verify against pre-defined requirements - Catch regressions from edits - Show exactly what failed and where Turn "hope I didn't miss anything" into systematic verification. 130
Allan @niemerg.bsky.social · 07/07/2025The real problem: AI generates documents faster than humans can review them. Without verification infrastructure, you either: - Slow down to manually check everything (lose productivity gains) - Risk missing critical errors (hello, Bar sanctions) 100
Allan @niemerg.bsky.social · 07/07/2025When AI writes code, developers can: - Compile it (immediate feedback) - Run tests against it (systematic verification) - Debug it step-by-step (transparent execution) When AI writes contracts, lawyers have... track changes and hope for the best. What if lawyers had systematic verification too? 100
Allan @niemerg.bsky.social · 07/07/202530% of Python code is now AI-generated. Meanwhile, many lawyers are still afraid to let AI draft a simple NDA. The difference? Software developers have automated tests. Lawyers don't (but they should). 🧵👇 (Make sure you see the demo in the post!) fullydoxxed.com/to-use-ai-la...fullydoxxed.comTo Use AI, Lawyers Should Steal This From Software Developers | Fully DoxxedSomething remarkable has happened in software development over the past few years. Large language models have revolutionized how code gets… 100
Allan @niemerg.bsky.social · 04/06/2025This is English but reads as incomprehensible as code in a computer language I do not know—which I guess it is. 010
Allan @niemerg.bsky.social · 04/06/2025Feels like you write this to dazzle the audience so they get that the character is crazy smart, it’s not supposed to be comprehensible. For that reason, you need it all tossed in together so you don’t confuse your audience elsewhere. 130
Allan @niemerg.bsky.social · 28/05/2025At this rate, Tourette's will be considered an unfair advantage in business writing by 2026. The weirder and more unpredictable you sound, the more human you seem. Language evolves whether we like it or not. I was right about all of this, by the way. fullydoxxed.com/fuck-chat-gp...fullydoxxed.comFuck! ChatGPT Really Is Ruining Writing (And I Told You So)! | Fully DoxxedTwo years ago, I tweeted this: And whaddya know, I was right. ChatGPT written words have so infiltrated human writing that what was once… 000
Allan @niemerg.bsky.social · 28/05/2025The future belongs to wherever ChatGPT can't go. One-sided rants. Emotional language. Stream of consciousness. Inside jokes. Raw, unqualified advice. We're about to deliberately write "worse" just to prove we're human. And it's gonna work. 110
Allan @niemerg.bsky.social · 28/05/2025Why do we hate AI writing so much? It's like C-3PO - technically correct but nobody wants to hear the protocol droid drone on. No, I don't want to "delve" into your "tapestry" of bullshit, thank you very friggin' much. 100
Allan @niemerg.bsky.social · 28/05/2025New Blog Post: Fuck! ChatGPT Really Is Ruining Writing (And I Told You So)! Two years ago I called this shit and now it's happening - ChatGPT is making "perfect language" and its signifiers low status. Polish now equals suspicious. Clean writing screams bot. fullydoxxed.com/fuck-chat-gp...fullydoxxed.comFuck! ChatGPT Really Is Ruining Writing (And I Told You So)! | Fully DoxxedTwo years ago, I tweeted this: And whaddya know, I was right. ChatGPT written words have so infiltrated human writing that what was once… 130
Allan @niemerg.bsky.social · 16/05/2025Wild to me that there is a machine that gives you answers correctly 80% of the time, and makes stuff up 20% of the time and most of the time people don’t double check what it says. 000
Allan @niemerg.bsky.social · 10/05/2025Napkin math suggests that if you are making methane and extravagantly shipping in the water, that’s still only 10 cents of cost per dollar of methane, so it seems like you are right—very interesting. 100
Allan @niemerg.bsky.social · 10/05/2025A key consideration would be water, so that may make it not economical. But my prediction is that if it becomes economical it will be done. 100
Allan @niemerg.bsky.social · 10/05/2025I don’t disagree with you on the need to grow and modernize the grid, I’m just pessimistic about it happening. Would be great. Regarding pipelines, it is my understanding that there are networks of underused pipes in many states like Texas and Oklahoma that may be great places for solar. 400
Allan @niemerg.bsky.social · 10/05/2025Everything you’ve said is correct, and yet, solar has been growing at 26% per year, for many years now. Transmission capacity isn’t growing anywhere near that. We are going to need economic uses of electricity that don’t require transmission and can take TW of energy. What other options do we have? 200
Allan @niemerg.bsky.social · 09/05/2025True, but impossible if we don’t radically change policy around transmission lines. 200
Allan @niemerg.bsky.social · 09/05/2025The solar panel stat is insane. If we could turn solar into hydrocarbons, things would be wild. You could repurpose existing infrastructure to collect from disparate oil wells but with energy from the sun. 310
Reposted by AllanGovernor JB Pritzker @govpritzker.illinois.gov · 08/05/2025A historic moment as we witness the first American leading the Catholic Church. Hailing from Chicago, Pope Leo XIV ushers in a new chapter that I join those in our state welcoming in at a time when we need compassion, unity, and peace. 1375786753
Allan @niemerg.bsky.social · 08/05/2025If it’s a 100Wh battery, that’s like 1/10th of a cent of electricity? You’d need to do like 200-400 queries to match the daily electricity of usage of a refrigerator? 010
Allan @niemerg.bsky.social · 05/05/2025To avoid that generic "ChatGPT smell" in final work, I explicitly ask the model to use my voice and words from our conversation. The result? Content that actually sounds like me. fullydoxxed.com/how-i-use-ll...fullydoxxed.comHow I Use LLMs to Write | Fully DoxxedYou've probably seen a lot of skepticism of large language models (LLMs) like ChatGPT and Claude on social media. Often, AI is seen as wor… 000
Allan @niemerg.bsky.social · 05/05/2025When I'm stuck or procrastinating, I just start dumping thoughts to an LLM. No organization needed. LLMs are great at helping draw out and organize your thoughts so you can focus on what you *really* want to say. 100
Allan @niemerg.bsky.social · 05/05/2025My key technique? "Shared context building" - I ask the LLM to collaborate with me through questions rather than immediately drafting. This helps clarify my thinking and produces more original work. 100
Allan @niemerg.bsky.social · 05/05/2025New blog post: Some thoughts on how I use LLMs to write. It’s not about getting them to do your work but using them to help think through what you need to do. fullydoxxed.com/how-i-use-ll...fullydoxxed.comHow I Use LLMs to Write | Fully DoxxedYou've probably seen a lot of skepticism of large language models (LLMs) like ChatGPT and Claude on social media. Often, AI is seen as wor… 110
Allan @niemerg.bsky.social · 02/05/2025The story, then, isn’t the executive order itself, it’s the President asserting authority he doesn’t have and the acquiescence of Congress in the face of a public that is unhappy with it. 020
Allan @niemerg.bsky.social · 02/05/2025Right? Like if you are shirking from explaining the law and the facts because it facially appears to favor one side, you are presenting a politically biased take while cosplaying neutrality. 210
Allan @niemerg.bsky.social · 02/05/2025Imagine being so committed to the view from nowhere that you can’t even say that an illegal presidential order targeting your organization is illegal. Maybe this kind of journalism should be destroyed. 110
Allan @niemerg.bsky.social · 29/04/2025LLMs are definitely changing writing, but it goes both ways: if you don’t want to write you can outsource more and more of the work to LLMs. But if you want to write more and better, LLMs are incredible tools! 000
Allan @niemerg.bsky.social · 28/04/2025A really great power tool is using text-to-speech with an LLM. Open a window and just start talking. Start working with the LLM as early in the process as you can. It’s a great way to stop procrastinating. 000
Allan @niemerg.bsky.social · 28/04/2025Probably my most important tool is “shared context building”. Instead of doing the thing I want to do (write code, an email, etc.) I work an artifact towards the thing—a memo, a plan, etc. “I want to create a memo about [x]. Ask me questions, make suggestions, and let’s collaborate on this.” 100
Allan @niemerg.bsky.social · 28/04/2025I’ve been thinking about the same. I imagine power users have rediscovered a lot of the same techniques but I haven’t necessarily seen them written down anywhere. 100
Allan @niemerg.bsky.social · 17/04/2025Saving this photo for my future Onion pitch: Adams: ‘Well, Well, Well, Not So Easy To Find A Mayor That Doesn’t Suck Shit, Huh?’ 030
Allan @niemerg.bsky.social · 14/04/2025These people will never come close to putting in the numbers my Grandmother achieved, and she was never weird about it. She was a real one. 000
Allan @niemerg.bsky.social · 12/04/2025It feels a little like a eulogy for pre-training scaling, they spent a huge amount of effort, didn’t unlock anything amazing, and want to get something for their efforts. 050
Allan @niemerg.bsky.social · 11/04/2025The U.S. did make something that the world wanted: stable financial assets with decent returns. We were amazing at that! 000
Allan @niemerg.bsky.social · 08/04/2025This has all been so stupid, but at least "you should put tariffs on out major geopolitical rival too" sounds like an attempt at policy more sophisticated than "I like Tariffs" or "everyone is ripping us off" 000
Allan @niemerg.bsky.social · 02/04/2025His ego won’t even let him admit that he was beat fair and square so he plays it off as if he knew it would happen all along. He’s going to continue being ego driven. 110
Allan @niemerg.bsky.social · 27/03/2025As I understand it, the hardest part isn’t the form factor but creating manipulators with anything near the sensitivity and flexibility of human hands. Put human level hands on any robot that is human sized and you have a winner. 020
Allan @niemerg.bsky.social · 27/03/2025Ok, but the specific desire for humanoid robot is often to create a generalist robot that can operate in human environments with people? 120
Allan @niemerg.bsky.social · 22/03/2025Also, it's worth noting that sanctions don't stop North Korea from using the technology, it just stops Americans from using the privacy technology. And, privacy technology isn't the worse thing to have around these days. 000
