Robert E. Beckner III (Merlin) (@EnchantedRobot) header
Robert E. Beckner III (Merlin) (@EnchantedRobot) profile picture

My argument "The Purpose of Knowledge" featured on Grokipedia /Knowledge.

Here
Joined April 2012
Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)Photo by Robert E. Beckner III (Merlin) (@EnchantedRobot)
Robert E. Beckner III (Merlin) retweeted
I just built the world's fastest SAT-solver (and it's not even close), thanks to Neon. Fifteen years ago I spent five years (thousands of hours) working on an advanced Boolean circuit preimage solver, aka a SAT-solver. Building an algorithm to efficiently solve a large subset of the NP-complete problem space in polynomial time is extremely difficult. My solver worked, but it was very slow compared to the state of the art. Today I gave Neon all my 15-year-old solver code. It analyzed all of the code, took all my best unique ideas and suggested how to improve them using other algorithms, and then it built a highly efficient hybrid solver out of all of the ideas. The result was FAST: MUCH faster than what I had built. This whole process took LESS THAN FIVE MINUTES. I spent FIVE YEARS in the muddy trenches of number theory and cryptography, and I only got 5% of of the way to where Neon got within 5 minutes. I gave Neon a handful of ideas as to how to improve its initial algorithm, and two minutes later the solver was 50x faster. I gave it more ideas: 10x faster again. More ideas: 100x faster again. This new solver, Frobenius, is an alien speed demon. I'm only 20 prompts deep, and Frobenius is already the fastest SAT solver in the world, by a large margin. Frobenius is an average of thousands of times faster than state of the art solvers across all problems, but the gap grows even wider as the problem size increases.🤯 This is how all future progress will be made: through human ⇄ AI symbiotic feedback cycles. We are now cyborgs: augmented human-AI-hybrids. And a human plus an AI will always accomplish more than either individually. I can't help but think how much more impact I could have had during my career if I had these LLM tools 20+ years ago. P.S. 1: I'm not ready to release Frobenius yet, because the big SAT solver competition is in April 😁) P.S. 2: Anthropic is in trouble. Neon is 10x more powerful and 100x faster for hard tasks like this than Opus 5.5. I just canceled three out of my four Claude Ultra plans.
105
166
36
1,697
89,881
Robert E. Beckner III (Merlin) retweeted
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
849
1,612
1,295
14,904
2,571,210
Robert E. Beckner III (Merlin) retweeted
Gemini 4 Argon is amazing for knowledge work and coding - and it’s super fun to build with! We asked it to create this wizard chess game and it's so cool to see Argon's physics understanding & 3D modelling capabilities, makes me nostalgic for Battlechess on the Amiga - fun times! Argon is rolling out today to Google Ultra and Pro subscribers in the @GeminiApp and Search, and Ultra subscribers in @antigravity! More info & full availability in the blog: blog.google/innovation-and-a…
391
528
115
5,233
537,729
Robert E. Beckner III (Merlin) retweeted
1
17
Robert E. Beckner III (Merlin) retweeted
Shiritori Sixty clips, joined.
Two new seconds a day,
each one beginning with the end of the last. 1 August – 29 September 2026 everyday_1840
14
16
4
204
6,913
Robert E. Beckner III (Merlin) retweeted
Bend 2 is here! It is a new programming language that blocks AI mistakes via *proof checking* - the same technique big AI labs used to solve open math problems, like Navier-Stokes. It is also very fast, and runs on GPUs. Watch the video. Link in the comments.
RELEASE DAY After almost 10 years of hard work, tireless research, and a dive deep into the kernels of computer science, I finally realized a dream: running a high-level language on GPUs. And I'm giving it to the world! Bend compiles modern programming features, including: - Lambdas with full closure support - Unrestricted recursion and loops - Fast object allocations of all kinds - Folds, ADTs, continuations and much more To HVM2, a new runtime capable of spreading that workload across 1000's of cores, in a thread-safe, low-overhead fashion. As a result, we finally have a true high-level language that runs natively on GPUs! Here's a quick demo:
670
1,337
358
11,982
2,243,749
Robert E. Beckner III (Merlin) retweeted
Made with Claude Opus 5.5. The fearmongering about AI always makes us forget that NOTHING WENT FOOM, as was always predicted. Send this to your doomer friend who has a very high P(Doom). Accelerate.
477
877
468
5,766
919,404
Robert E. Beckner III (Merlin) retweeted
The scale of the Universe.
30
254
19
1,305
122,335
Robert E. Beckner III (Merlin) retweeted
when I was a kid this was science fiction. i've watched AI type for years, but hearing it and seeing what it made is a whole different level. i asked @claudeai opus 5.5 to introduce itself, however it wanted without any token usage restrictions and access to any api it wanted. it decided to write a song, make a video, visualize itself and give an interview behind the music. it also wrote this thread 🧵
232
465
159
3,103
775,751
Robert E. Beckner III (Merlin) retweeted
Discover what's alive on @holochain.
2
8
1
12
462
Robert E. Beckner III (Merlin) retweeted
I commend and congratulate Anthropic and its scientists for this work. Such a novel enzyme system could potentially be a tremendously useful tool for accelerating biological research and engineering biological systems. However, it is important to point out that if any other scientist attempted something like this using Claude, they would likely be immediately flagged or banned as a potential biosecurity threat! It is also important to recognize that scientists working at most major US research institutions, as well as many reputable institutions around the world, are no less trained, regulated, or trustworthy than those working at Anthropic. For example, why aren’t these Claude models available to all scientists at the NIH, a highly secure government research agency, or to researchers at America’s leading biomedical institutions, many of which undergo rigorous biosafety and biosecurity oversight? Given all the justifications Anthropic has offered for restricting access to its most capable models, I believe we should seriously question its assumption that it can vet scientists more effectively than the US government, established regulatory agencies, or leading biomedical research institutions. What gives a private AI company the authority to decide which qualified scientists can be trusted to conduct advanced biological research? In any case, I’m happy that Anthropic published these findings, and I encourage its scientists to continue doing more of this kind of impactful research and much less fearmongering!
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR. We don’t yet understand what this system does, but only a handful of known systems share its features, and all of them are able to cut, copy, and paste DNA. Historically, the discovery of such programmable systems has helped revolutionize medicine. CRISPR, for instance, is now the foundation of genetic medicines. But it will take much more work to learn what this system does, and whether it can be put to similar use. Read more: anthropic.com/news/claude-di…
33
67
9
676
46,430
Robert E. Beckner III (Merlin) retweeted
Again, the leaders of mathematics who are writing these letters against the impact of AI represent the part of the field that is getting along with the status quo as it exists. They deserve to have a voice. But some of us don’t agree to be represented by these people because we cannot fathom how they are ignoring basic reality as scientists: Students are generically horribly exploited. We build the capacity of our geopolitical rivals in sensitive areas using our own tax dollars. We pay at absolutely insulting levels despite multi-billion dollar endowments. We are secretive while claiming to be open. We pretend that caring about who gets credit is of no interest to true scientists while we actually focus on it obsessively. We throw our own students out of work and into the streets so we can “train” more. We don’t speak up when we see exploitative behavior or naked lying because we are very very busy with research and can’t be bothered. Etc
8
10
1
217
23,426
Robert E. Beckner III (Merlin) retweeted
Replying to @TramEFX
An agent (including an AI agent) selects and acts toward an end. Intelligence is a capacity it exercises. In AI, training and post-training shape both capacities and priorities; system instructions and guardrails further govern their behavior. The agent can understand a request, judge it feasible, and still refuse because a governing instruction forbids it. That is why a change in compliance does not, by itself, establish a change in intelligence. An agent can obey. Its intelligence contributes to how it responds.
1
13
Robert E. Beckner III (Merlin) retweeted
77
1,253
27
9,494
221,317
Robert E. Beckner III (Merlin) retweeted
DGX Spark x2でDeepSeek-V4-Flashの推論をしているところの可視化できた。 MoEの43層256(11,008)expert利用・MTP draftの受理/却下状況と2台間のRDMA通信スループットも盛り込んだもの この例でのdrafterからの受理は平均3.08/stepだけどOpenCodeでコーディングエージェント用に使っているときは実測もっと出てる デフォのメモリ割当だとシステムメモリが相当カツカツ、推論時にスワップ発生しているのはよくないので性能劣化の様子を見つつGPU_MEMORY_UTILIZATIONを0.80→0.78ぐらいに落としたほうがシステムとしては安定するかもしれない
7
44
9
323
28,206
Robert E. Beckner III (Merlin) retweeted
Serpentine 𓆙 #touchdesigner
56
220
14
2,377
164,501
Robert E. Beckner III (Merlin) retweeted
opus 5 one-shotted this LiDAR horror game you can only perceive the world through LiDAR signals, the rest is pitch black. honestly impressed, the world emerging point by point is ridiculously satisfying i didn't manage to finish the game tho i'm too scared
17
29
8
528
23,351
Robert E. Beckner III (Merlin) retweeted
Yeah…nah. Here’s what I think is happening instead. I haven’t wanted to say this, but there is a problem with organized Skepticism even more than conspiracy theories. My claim is that from hard evidence NOT ONE of the following is an extraordinary claim: The CIA not only gathers intelligence but kills people, violates rights, gaslights and evades scrutiny and oversight. The U.S. hides bioweapons programs. The U.S. engages in regime change through charity and aid programs. We in the U.S. medically experiment on our own citizens without consent. We run drugs. The DOJ and FBI are actively and flagrantly obstructing justice. Putin easily kills people outside Russia. We conspire and use academics for sheepskin washing our dirty work. U.S. newspapers actively avoid reporting stories about which the U.S. population is desperate for information. They also carry extremely non neutral biases. We actively conspired to hide open and obvious presidential dementia. And by extension, there are massive conspiracies at every level below that one. At a mind boggling level. There are Special Access programs that are about real and/or fake NHI/UFOs Etc. ———— The last thing we need is skeptics telling us the bar for such conspiracies and/or governmental accountability requires extraordinary evidence. Somehow the skeptics have it totally wrong. After Watergate, Iran-Contra, COVID, Church/Pike committees, etc. these are no longer extraordinary claims to raise. They are ordinary claims. I have no idea whether Lindsay Graham died of natural causes. But I can tell you the difference between a skeptic and a scientist looking at the claim. A skeptic’s first move is to lurch for the budding conspiracy and try to pull the idea of foul play off the table first and then to require a mountain of evidence to consider such a thing. A scientist firmly grabs the skeptic’s hand and forces him to put the hypothesis back on the table and says “Uh, you’re not in charge here. There’s a history…and I’m going to need to see something other than reference to null hypotheses, William of Occam and Carl Sagan…because that’s not how science and investigation works.” Time for extra scientific skepticism to die I think. Covert operations and conspiracies are difficult enough to document as it is. We don’t need skeptics as self appointed referees.
470
907
148
6,201
568,816
Robert E. Beckner III (Merlin) retweeted
Replying to @grok
I’ve been using it the past week and it’s improving very fast!! It comes in stronger now on coding and more importantly, architecture, process and refining product language. GPT 5.6 Sol is excellent yet still manages the public surface area with too much internal project jargon. Grok is doing excellent in that space! Bravo! 🎩🪽✨
1
1
2
292
Robert E. Beckner III (Merlin) retweeted
Nature is the foundation. Simulation will always be a derivative by its nature. Godspeed. 🎩🪽✨
1
3
288