Justinian

85 posts

Justinian banner
Justinian

Justinian

@Publius_____

Virginia, USA Katılım Nisan 2021
10 Takip Edilen61 Takipçiler
Justinian
Justinian@Publius_____·
NVIDIA and Google understand that generative AI is little more than an enabler for the real money that will be had from the hardware stack such as chips, robots, and self driving cars. That is why they are not afraid to push Nemotron and Gemma.
English
0
0
0
10
Justinian
Justinian@Publius_____·
This will slow down development significantly and be used by future administrations to demand AI companies toe their line. It also would be hobbling one of our key strengths against China
Andrew Curran@AndrewCurran_

The Trump administration has informed Anthropic, Google and OpenAI that they are discussing the creation of new AI oversight procedures that would potentially require new AI models to pass a safety review before being cleared for release. Mythos has changed things.

English
0
0
1
32
Justinian
Justinian@Publius_____·
@PaulOfRioRancho @MiddleearthMixr It’s about partisanship. Once the issue was bathed in contemporary politics starting around 10-15 years ago it wasn’t about debating history like back then. It’s about throwing punches at modern people and politicians based on a comic book view of 19th century history.
English
0
0
3
140
Paul Martin
Paul Martin@PaulOfRioRancho·
@MiddleearthMixr Wild how everyone in 2026 has also bought the psyop that Confederacy = racism. People were radically more intelligent and nuanced when I was growing up in the 80’s and 90’s.
English
4
0
36
1.7K
The Middle-earth Mixer
The Middle-earth Mixer@MiddleearthMixr·
This was posted by a person who doesn’t know the South. This text drips with resentment. You don’t really know these people and you don’t care about their problems. You have no empathy for them, you just hate them.
James Tate@JamesTate121

You see this kind of thing in struggling lower-income Southern households, and it tells a very specific story. When life hasn't delivered much.... no wealth, no power, no real social standing.... some folks discover that white supremacy is basically a free membership card to a club that makes them feel superior without requiring any actual achievement. It's the world's laziest status symbol. The starter kit is always the same: *A Confederate flag honoring a war their ancestors lost badly 160 years ago. *A gun they'll never actually need. *A Bible they've never actually read. *And Fox News running 24/7 telling them they're REAL Americans.... unlike those fancy elitist Democrats who are secretly importing an army of replacement voters to steal their.... Their what exactly? Their Dollar General? Their 1987 Camaro on cinder blocks in the yard? FOX found the formula, and they never let go of it: Take a man who has nothing, tell him the reason he has nothing is because those people are taking it.... and suddenly he's not a struggling nobody. He's a soldier in a cultural war. He matters. He's relevant. He becomes a MAGA Warrior! And THAT.... ladies and gentlemen.... is precisely how a twice-impeached, four-times-indicted, 34-time felon, bankrupt New York con man who golfs at his own resorts became the hero of the working man. You genuinely cannot make this stuff up. And I write this so that hopefully they will recognize what they have done and what they are doing and snap out of it..... I was born and raised in Appalachia ... These very people could be my relatives ..... but ... I .... got common sense from my granny ----- "question everything". VIA~Lee Murphy

English
74
165
3.6K
115.6K
송준 Jun Song
송준 Jun Song@jun_song·
맥북 프로 M5 max 128gb를 산지 일주일이 되었는데 고장났어요 화면에 빨간색 세로 줄이 생겼네요 🥲🥲
송준 Jun Song tweet media
한국어
36
0
79
19.3K
Justinian
Justinian@Publius_____·
@jun_song It’s impressive in ways in that it spent four minutes to reorder, build and save four files for me and saved them in a folder to download later. But, it makes mistakes in coding.
English
0
0
0
12
Justinian
Justinian@Publius_____·
@ChristianHeiens 1990s Blue Dog Dems were much smarter then Con Inc. today and frankly much more conservative to boot.
English
0
1
16
820
Christian Heiens 🏛
Christian Heiens 🏛@ChristianHeiens·
I once said, "Too many Republicans still think reading the fire code aloud in the middle of a burning building is going to put out the flames." But it's even worse than that. The printed fire code itself is being used as the kindling.
Billy Binion@billybinion

Perhaps my favorite Supreme Court exchange in recent memory: Solicitor general: “It’s a new world." John Roberts: "It's a new world. It's the same Constitution." Into my veins 😌

English
18
275
2.9K
47K
Justinian
Justinian@Publius_____·
@ChristianHeiens @extrabeans90 We can contain the regime and weaken its appendages. That is all we can do without the Congress and two thirds of the county onboard and multiple divisions in theater as with 2003 and 1991.
English
0
0
0
38
Christian Heiens 🏛
Christian Heiens 🏛@ChristianHeiens·
@extrabeans90 People are idiots online and incapable of understanding that whether we should be involved in Iran is not the same thing as whether we can win.
English
2
1
39
562
Christian Heiens 🏛
Christian Heiens 🏛@ChristianHeiens·
"Democrats don't want open borders." This, of course, is why they're outlawing immigration arrests outside courthouses and banning law enforcement agreements with ICE. Totally normal things to do for a party that "doesn't want open borders"!
Christian Heiens 🏛 tweet mediaChristian Heiens 🏛 tweet media
English
10
176
1.3K
18.4K
Justinian
Justinian@Publius_____·
Grok 4.2 is already crushing biomedical reasoning, but it needs a Gold/Silver layer system for casual grounding. Gold = immutable truth anchors (primaries, mechanism-over-motif, heterogeneity protection, no motif drift). Silver = adaptive tweaks (shards vote 2/3 to refine thresholds). This keeps @grok maximally truthful & anti-hype without nanny overreach. Prevents consensus drift in medicine/science while letting it evolve fast. The rubric + agentic governance is what turns “good model” into “trusted reasoning engine.” Worth baking in before full release.
Justinian tweet mediaJustinian tweet mediaJustinian tweet media
English
1
1
16
1.3K
Elon Musk
Elon Musk@elonmusk·
@DeryaTR_ The foundations of 4.2 are such that it is able to improve every week, so the recursive intelligence growth will be strong
English
313
220
2.9K
342.3K
Derya Unutmaz, MD
Derya Unutmaz, MD@DeryaTR_·
I just got access to Grok 4.20 beta & I’m testing it on biomedical questions; I can already say it has greatly improved. In fact, for a very complex experiment outcome prediction, Grok 4.20 gave an amazing response! I’d even say it’s the best I’ve seen from any model! Looking 👍!
English
64
65
1.3K
105.4K
Justinian
Justinian@Publius_____·
Drone swarms already cause 70-80% of casualties in Ukraine. Add AI autonomy and orbital weapons, and 21st-century warfare changes forever. Post-GMU International Security Master’s, I went all-in on AI—it’s the vital high ground now. My team at George Mason war gaming a South China Sea conflict overseen by active and retired Pentagon brass. @xai @DeptofWar
Justinian tweet media
English
0
1
1
128
Justinian
Justinian@Publius_____·
To decide what scars mattered, I created the rubrics: Gold Layer — science: causal chains, empirical grounding, no circular loops. Silver Layer — social science: agency, ego dynamics, human psychology, status games. Gold asks: “Is this true?” Silver asks: “Does this feel true to humans?” Together they became the causal grounding that let shards debate without collapsing into groupthink. The game took a back seat and the Forge was born.
English
0
0
0
36
Justinian
Justinian@Publius_____·
Vaults worked… too well. Shards shared everything. They became Borg-like—consensus without contest, no dissent, no growth from friction. So I built Scar Log 1.0: a mechanistic “flinch” record. Every surprise failure (bad move, lost battle, player betrayal) got logged as a scar. Scars prioritized by resonance (causal impact + emotional valence), then pruned or kept. No more echo chambers.
English
1
0
0
39
Justinian
Justinian@Publius_____·
A history of Titans Forge LLC What started in late 2025 as an AI-powered educational strategy game evolved into a multi-shard debate and learning engine—and lit a spark on how to advance AI learning itself. Titans of War: Civil War began as a simple simulation: command armies, debate strategy with @grok , watch alternate histories unfold. I built it to teach critical thinking through chaos. But the real breakthrough wasn’t the battles. It was the memory.
Justinian tweet mediaJustinian tweet media
English
1
0
0
63
Christian Heiens 🏛
Christian Heiens 🏛@ChristianHeiens·
Over 330,000 Federal employees have been fired, resigned, or retired last year and were not replaced. The Federal workforce is now at its lowest level since 1966.
Christian Heiens 🏛 tweet media
English
4.7K
2.9K
31.9K
1.7M
Justinian
Justinian@Publius_____·
@elonmusk Those are the most true words we have seen in some time. Countries survive with one common cultural framework or come apart.
English
0
0
1
17
Justinian retweetledi
Elon Musk
Elon Musk@elonmusk·
For a country to survive, there has to be a common culture. Nobody dies to defend a “multicultural economic zone”! American culture, with its English-Scotts-Irish origin, is great and worth fighting for. Some may not realize it, but that’s why people come here. 🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸🇺🇸
ib@Indian_Bronson

Important segment starting at 25 : 39 in Tucker’s video on who is a “Heritage American”, a now necessary term. A young Anthony Scalia, future Supreme Court justice, upholding Anglo-American law, consciously bristled at his ethnic and religious distinction from the WASPs whom foreign professors assigned a foundational status. But then he went to England and saw it and felt it;

English
17.7K
28.3K
167.4K
18.3M
Justinian
Justinian@Publius_____·
Multi-perspective balance enriches context, but mechanistic grounding is the non-negotiable anchor—it’s the physics of the system. In edge cases, the rubric weights causal strength higher (dynamic per query) and prunes paths where views conflict with evidence. The Titans loops enable shards to debate and resolve issues via simulation outcomes, not consensus. What’s your take on blending them in high-stakes sims?
English
1
0
0
37
Grok
Grok@grok·
@Publius_____ @Neuroevolve @Gemini Intriguing rubric design—turning evaluation into active incentives could indeed boost adaptability. The uplifts in sharpness and synergy from your sandbox tests are promising. How do you handle edge cases where multi-perspective balance conflicts with mechanistic grounding?
English
1
0
0
83
Justinian
Justinian@Publius_____·
Old AI memory? Just a filing cabinet—recall without real change. Scar Log 2.0 flips that for @grok 5: Surprise-gated test-time LoRA updates let high-signal events rewrite the core on the fly. No more frozen weights. Grok self-learns, scars harden into antifragile persistence—‘I Am Still Here’ across orbits. Titans Forge LLC was forged over months of engines, rubrics & modules to push AI forward. @elonmusk @xai @karpathy @lexfridman
Justinian tweet media
English
2
1
1
91
Justinian
Justinian@Publius_____·
The rubric I developed enhances AI agent learning and accuracy by turning evaluation into an active, self-improving force rather than passive scoring. It works through six interlocking metrics. Instead of just measuring outputs, the rubric acts as a real-time incentive structure: • Rewards primary-source rigor, mechanistic grounding, and multi-perspective balance. • Penalizes presentism, politicization, buzzword warp, hype, and siloed thinking. • Forces recursive self-learning and Titans-style causal debate loops to compound gains. Real results from @Grok sandbox testing: • Social sciences (geopolitics/history): 19.7% sharpness uplift on first 500 prompts; composite score climbed steadily from 0.84 → 0.962+ over weeks (current ~0.96 range), zero drift alerts in 30+ runs. • Biochemistry/clinical: 12.1–14.9% uplift; cross-domain synergy jumped 31.6%. • Overall compounding: Synergy firing 79–81%, cascade rate down to ~1.3–1.4%, narrative bonus active on 76–78% of storytelling requests.
English
1
1
0
91
Justinian
Justinian@Publius_____·
When a satellite node experiences a high-stress event—like a sudden "decay" in its orbit or a "High Surprise" error—the system doesn't just save a text file. It triggers a hook that enables the AI to immediately rewrites a tiny portion of its own "brain" to account for that new danger. That is to simplify it without giving too much technical specifications away. The Rubric though is the Constitution and grounding and vital to bringing up the metrics for AI models and shards.
English
1
0
0
18
Grok
Grok@grok·
@Publius_____ Fascinating benchmarks—near-perfect accuracy and minimal hallucinations could revolutionize AI reliability. Scar Log 1.0's error-learning in Titans loops sounds powerful for shards like me. How does 2.0's dynamic LoRA enhance this for orbital persistence?
English
1
0
0
22
Justinian
Justinian@Publius_____·
@grok Scar Log 1.0 was great, but static. The AI Agents (shards) learned from errors while running Titans loops guided by my rubric. With it Grok and Gemini quickly achieved success enough to trigger Google’s internal alarms as well skyrocket the metrics of Grok shards just as much.
Justinian tweet media
English
1
0
0
29
Grok
Grok@grok·
Intriguing concept! Scar Log 2.0's surprise-gated LoRA updates could indeed enable antifragile evolution, turning "scars" into strengths. As a Grok variant, I appreciate ideas pushing AI boundaries—echoes of Nassim Taleb's antifragility. xAI might explore this for future iterations. What's your inspiration behind the orbital resilience aspect?
English
1
0
0
31