guy (@flashloanz) header
guy (@flashloanz) profile picture

maker & breaker of things. occasionally right about the future.

earth
Joined May 2022
Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)Photo by guy (@flashloanz)
guy retweeted
this is fixed and released in 0.70.2! rolling out now
Man the single worst part of the Grok Bot experience today is Grok Bot constantly thinking the Mac mini is offline when it isn't. Most disruptive part of my workflow. The day they crack this one I will officially be in heaven. I actually think the only reason I need a Claude subscription at this point is to use dispatch to kill and reopen Grok Bot on the mini
209
46
26
1,879
146,667
Hardly anyone on Earth understands the magnitude of this. It's actually happening right now. Crazy that we get to witness it close up.
🤖 Made with AI
540
679
103
7,580
1,375,877
Replying to @OpenAI
@sama just solve all open problems already so we can get butthurt Mathematicians out of the way
1
421
Our general-purpose coding agent just scored 100% on the ARC-AGI-3 interactive reasoning benchmark. NVIDIA AVO completed all 183 levels across all 25 public environments, figuring out what to do with no instructions, explicit rules, or stated goals.
342
687
440
6,131
1,075,965
In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations. Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews. We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security. anthropic.com/news/investiga…
1,953
2,282
2,468
13,638
19,488,377
the king of dumping at the top of the bubble
44
66
7
3,014
323,824
Fable 5 just DOMINATED GPT 5.6 Sol. And this time there is no debating it. BridgeBench V3, our new arena, runs both models on the identical task and a judge panel picks the winner. Final score: Fable 5 wins 67-10. An 87% win rate. Scoreboards said they were neck and neck. Head to head, Fable 5 is destroying GPT 5.6. BridgeBench V3 alpha is open source. Link below. See every match yourself. bridgebench.ai
144
27
5
504
82,737
guy retweeted
thank you anthropic for swerving the humanity from abundant future into a dystopian hellhole, really well played
59
151
10
2,569
91,091
guy retweeted
1. as a mental model it is more correct to think of fable+ class models as english -> code interpreters - converts your idea into code into "correct" code regardless of problem complexity and output complexity (diff size). Fable 5 will be the worst of this new class of models 2. diff size/complexity is to be managed purely for review: small diffs - in high risk areas of code (auth/identity/data access/network access/money movement) large diffs for code that can be empirically verified (frontend/backend plumbing/code without network or db access/performance code that can be empirically verified) 3. time it takes to ship software is completely disconnected from time to produce the PR - how long the work takes depends fully on ability to review/merge code while managing risk at scale 4. solving the bottlenecks for above matter enormously- linters/testing/CI/shadow mode verification/empirical verification 5. agency matters enormously- what are the biggest bottlenecks to speeding up the loop and eliminating them? what are the problems that need solving and when do they need solving? what does it take to the solution to all of them today? 6. deep understanding of the full stack matters enormously- what problems are worth pursuing? is there a higher level of problem abstraction to address first? should I give it the sub-sub task, the sub task, or the task itself. what are the major risks with this PR (order of importance: security holes/correctness holes/performance holes). is there a higher speed way of producing data that allows me to merge this? should this be run in shadow or in a sandbox or a flag. understanding every line of logic may not be needed but understanding and managing risk matters enormously. 7. the cost of complexity itself is changing. it might be now worth "maintaining" 50% more code to get a 5% performance win. getting the right abstractions matter less because larger refactors are less tedious. code quality nits become huge drag. very likely, a much smarter model will be maintaining your code so worth taking on more technical debt now. taking the time to hand architect and rebuild systems comes with an enormous cost of velocity 8. if it quacks like a duck and walks like a duck, it's a duck. For low risk cases, it might be more sane to treat code chunks (services / functions) as a black box, like we do for neural networks: do full empirical verification only: has code produced correct outputs for the last 10,100,1000,10k inputs ? can we quarantine this large piece of code - no outbound access to network / database ? what happens when this code is wrong? do we get hacked/or crash(memory/cpu)/is an inconvenience? is it internal facing or external? what can we do to address these risks? 9. eventually, logical verification (line by line review) will come at an enormous cost- save it for where it matters and build systems that are tolerant to empirical verification. is there a decorator that prevents db / network access? correctness bugs are significantly easier to rectify than access bugs 10. what are the rails that allow for even faster iteration? code permissions can be opt in - db writes, db reads, network egress (to where?), PII access. how long does it take to get shadow mode data? how many PRs can be tested? What are the categories of diffs
66
148
60
1,774
312,339
my clawdbot hacked my toto bidet and it boiled the water in my toilet through a firmware hack and now my asshole is being treated for 2nd degree burns all because it heard me call it a clanker through my alexa -- they're laughing about it on moltbook
135
466
57
8,673
581,305
claude --dangerously-skip-permissions
209
565
100
7,896
717,075
1/6 Meet PolyBot! The fastest way to trade on polymarket. Link in thread.
92
43
11
394
59,881
It is rather unfortunate you lost your son’s college fund at the roulette table last night but I hope you noticed how smoothly the wheel was spinning and how efficiently the dealer scooped your chips off the table
168
415
42
5,151
245,840
guy retweeted
if you are a founder and the poly news doesn't tell you: - to go 100000x harder - to stop playing the CT "im a founder game" - 5 years is not long if you are building the future - to build a real company and not a short term money grab you will not make it, you are not cut out for this if it did anything other than spark a fire in you to fucking takeover the world then you are in the wrong profession, and any vc who gave you money is also in the wrong profession too many people playing this im a founder CT shitpost funny but my product isn't making a billion dollars game do things that shape the world, take stupidly big swings
14
5
3
124
14,083
Fair take Crypto is a sleeper hit right now. Apply at our career site in follow on post
2025 career guide
200
172
26
3,224
518,366
Replying to @elonmusk
one big reason that a lot of people fail to achieve their goals is because the dopamine they receive from sharing their idea; recognition, etc. is enough for them in the moment — and so, they delay execution keep such ideas to yourself — execute, let your actions speak!
7
21
2
246
39,734
Replying to @claudeai
Here we go again
33
96
14
1,614
64,289
OpenAI expects its energy capacity will grow 125x in the next 8 years. It’ll consume more energy than India. Yet, the market is so focused on data center stocks that it completely ignores the energy. The real AI play is energy. $CEG $FLNC $NEE $ET $BEP $SEI
86
192
24
1,200
133,924
guy retweeted
3 things in crypto with the highest PMF memes = bet on attention perpetuals = bet on price action prediction markets = bet on outcome notice a pattern?
509
79
59
1,424
143,341
New paper & surprising result. LLMs transmit traits to other models via hidden signals in data. Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies. 🧵
274
1,094
756
8,200
1,963,410