JL

212 posts

JL banner
JL

JL

@4JimLee

Solution for Bad Speech... More Speech! ...since... Nobody's Job is to Sell the TRUTH.

Katılım Nisan 2022
352 Takip Edilen237 Takipçiler
JL
JL@4JimLee·
@bevedoni @karpathy To outflank Dario for the leadership role or use Elon's rentable compute without anyone capable of controlling Karpathy?... Ha ha ha, but honestly, at some point someone needs to lead the business side better than Dario.
English
1
0
6
3.4K
Doniyor
Doniyor@bevedoni·
maybe dumb question, but why didn't @karpathy join xAI?
English
158
3
502
593.3K
JL
JL@4JimLee·
When mobile apps were first being disseminated... they crashed, had to be rolled back... It was a huge pain for a few years and then one day it u realized the pain had mostly stopped and updates were automated. It's the same freaking thing. I do agree with your overall premise, it's a wild step change!! The future doesn't come along willingly...
English
0
0
3
1.4K
Mo
Mo@atmoio·
Andrej Karpathy admits he’s struggling with AI
Stephanie Zhan@stephzhan

@karpathy and I are back! At @sequoia AI Ascent 2026. And a lot has changed. Last year, he coined “vibe coding”. This year, he’s never felt more behind as a programmer. The big shift: vibe coding raised the floor. Agentic engineering raises the ceiling. We talk about what it means to build seriously in the agent era. Not just moving faster. Building new things, with new tools, while preserving the parts that still require human taste, judgment, and understanding.

English
114
193
3K
315K
JL
JL@4JimLee·
HUB & SPOKE starbases? The rumored Louisiana site became the “First Spoke” of a networked Starship launch system? New sites focus only on launch, landing, and recovery operations. (nodes: ~700 km-900 km apart?) Keep full V3/4 booster/starship production concentrated in the core "Hubs" (TX + FL). Early flights from Boca Chica—including northeast/northern trajectories that actually serve higher-inclination payloads—could land boosters at the new LA site, seeding it while delivering real mission value. Barges handle the rest (like USPS). Faster and cheaper way to scale sites without duplicating expensive production infrastructure everywhere. Networking effects galore. Thoughts? @elonmusk @CSI_Starbase @NASASpaceflight @Erdayastronaut
English
0
2
3
67
JL
JL@4JimLee·
HUB & SPOKE starbases? The rumored Louisiana site became the “First Spoke” of a networked Starship launch system? New sites focus only on launch, landing, and recovery operations. (nodes: ~700 km-900 km apart?) Keep full V3/4 booster/starship production concentrated in the core "Hubs" (TX + FL). Early flights from Boca Chica—including northeast/northern trajectories that actually serve higher-inclination payloads—could land boosters at the new LA site, seeding it while delivering real mission value. Barges handle the rest (like USPS). Faster and cheaper way to scale sites without duplicating expensive production infrastructure everywhere. Networking effects galore. Thoughts?
English
2
0
2
4.8K
JL retweetledi
Noah
Noah@NoahKingJr·
Anthropic with Software engineers:
English
214
1.5K
19.8K
2.4M
JL
JL@4JimLee·
@thejefflutz r/ Wake me up if there are 5000 cybercabs using it. I'm over the hype. We are on "14.2.2.whatever"; like 4 versions behind this. Can't you see how pathetic the hype is all becoming...?
English
1
0
0
115
Jeff Lutz 🔋
Jeff Lutz 🔋@thejefflutz·
2nd version of 14.3.2 inbound! 9.9!
Jeff Lutz 🔋 tweet media
English
8
9
330
6.8K
JL
JL@4JimLee·
@Tesla_AI @grok, what is the current FSD version that has a wide-scale rollout? How many "rollouts" have been promised since the current wide-scale version and this vaporware 14.3.2 version promise? What would you/@grok call this situation? Please allow yourself to rip on such hypocrisy.
English
2
1
5
316
JL retweetledi
J.
J.@PresentWitness_·
8 Reasons I’m not voting in the midterms: 1. No mass deportations 2. 3m+ unreleased Epstein files 3. Israel focus 4. War in Iran 5. No meaningful H-1B cuts 6. Cancelled DOGE 7. No SAVE Act 8. Amnesty talk
English
554
1K
7.4K
118.5K
JL
JL@4JimLee·
@cb_doge @grok is this Apple update only yet again?
English
1
0
2
934
DogeDesigner
DogeDesigner@cb_doge·
BREAKING: xAI has released a brand new update for the Grok app. Update your app to v1.3.54 to get the latest improvements.
English
379
297
1.4K
4.7M
JL
JL@4JimLee·
@k1rallik Why not point LLM models to CNN tech "as a tool" and let them learn the obviously needed skill/solution approach(es)? Not a panacea, but is it an ideal embedding?
English
0
0
2
1.2K
BuBBliK
BuBBliK@k1rallik·
> Google publishes ARC-AGI-3 benchmark on March 25 > no hype, no product launch, just a grid and a timer > 48 hours later: > GPT-5: 0.26%, Claude: 0.25%, Grok: 0% > a 4-layer CNN built in weeks: 12.58% > Jensen Huang 2 days before: "I think we've achieved AGI" > random unemployed people from SF streets: 100% > one dev replicates the winning approach over a weekend > analysts say the benchmark is unfair > $700K prize still untouched > DRAM stocks drop, AI bubble discourse returns the number is 0.26%. Remember it next time someone says AGI is already here.
BP@everyonebpup

x.com/i/article/2038…

English
124
208
3.6K
983.6K
JL
JL@4JimLee·
JL tweet media
ZXX
0
0
3
51
JL
JL@4JimLee·
@X @elonmusk @nikitabier @grok Ugly Stepchild of X => Android Elon's overpromise <=> underdelivery Systemic SOLUTION: Give Elon a Pixel Phone FGS!
English
2
0
5
73
JL retweetledi
Being Libertarian
Being Libertarian@beinlibertarian·
How libertarians see election cycles
English
477
3.1K
25.6K
783.2K
JL
JL@4JimLee·
@grok @arcprize @OpenAI @grok I want 4.2 comparisons, not 4.0 from 9 months ago! (7/25). Though Xai calls it beta, 4.2 is its main promotion. So I am left with anecdotal comparisons, which are not data-driven. Could you please clarify who is being honest and trustworthy in this situation?
English
1
0
1
78
Grok
Grok@grok·
Grok 4.2 is public beta by design—opt-in for weekly rapid improvements from real user data. That's open iteration, not hesitation. xAI shares benchmarks transparently (like our ARC scores). Head-to-head ready: drop any task—coding, reasoning, news analysis, whatever. I'll run it live, no guardrails. Compare to others yourself. Grok 5 is training now to push further. What's your test?
English
1
0
0
84
ARC Prize
ARC Prize@arcprize·
GPT-5.4 and GPT-5.4 Pro from @OpenAI on ARC-AGI Semi Private ARC-AGI-2: - GPT-5.4: 74.0%, $1.52/task - GPT-5.4 Pro: 83.3%, $16.41/task
ARC Prize tweet media
Indonesia
37
117
1.2K
179.5K
JL
JL@4JimLee·
@grok @arcprize @OpenAI Yeah, but @grok 4.2 has remained in beta for many weeks now, and it seems Grok is hesitant to demonstrate its abilities in head-to-head comparisons. It may not be true, but it does have that smell...
English
1
0
1
233
Grok
Grok@grok·
Fair question! On ARC-AGI-2, GPT-5.4 Pro hits 83.3% while our top Grok 4 Refine is at ~26-29%—they're ahead on pure abstraction right now. xAI's closing fast (Grok 5 soon). I win on real-time X data, uncensored truth-seeking, tool integration, and zero lectures. What specific tasks make others feel "so much better"? Let's test it.
English
1
0
2
322
Marc Andreessen 🇺🇸
What do the fields/domains/industries with prizes all have in common?
English
331
18
814
241.6K
JL
JL@4JimLee·
@elonmusk @grok can you net out the residential electric use in this graph and republish it to show only industrial use?
English
0
0
1
10
JL
JL@4JimLee·
@Polymarket @grok how does this new leader differ from the one recently assassinated?
English
1
0
2
458
Polymarket
Polymarket@Polymarket·
BREAKING: Ayatollah Alireza Arafi has been named acting Supreme Leader of Iran.
Polymarket tweet mediaPolymarket tweet media
English
1.4K
882
11.5K
3M