
crypto curious, dipping into NFTs, DeFi, and other degen pastimes | profile pic @tubbycatsnft
Joined July 2021
- Tweets3K
- Following533
- Followers143
- Likes7.9K
Pinned Tweet
Tubby Cats investment thesis...
- they are cats
- crypto native visible lead artist and team
- well matched traits make each one feel unique
- 1 address = 1 mint on allow list, makes it much harder to flip and crush floors
- meme them to the moon
Yeah mathematicians are cooked, this is an extremely brain wormed take
This is a slide from Terence Tao's 'Math 2.0' public lecture yesterday evening at Caltech. I am highly confident that, if the cure actually worked, almost any Stage 4 cancer patient would tell you that human comprehension of its mechanism is completely irrelevant to them.
1
71
This feels like skeptic cope. To be fair Francois has rightly produced incredibly useful hill climbing benchmarks, but they seem to have had limited usefulness into the “soft singularity”.
The benchmarks are now real valuable informational and theoretical fields.
What if the jagged frontier is mainly math + code (which you can push arbitrarily far with RLVR), and everything else starts to plateau because it is still bottlenecked by human generated data?
Model performance in non-verifiable areas has kept improving steadily, albeit much slower than for math and code. But is that steady improvement a side effect of a higher G (itself driven by RLVR), or only a function of the amount of new human data getting injected into training (which is still continually happening on a massive scale)?
A lot of things depend on the answer to this question
29
If you are legitimately trying to forecast any industry, field, or area of expertise out 3-5 years at this point you are not going to have a good time.
This is potentially the most uncertain period of human development in history. Nothing has ever changed this quickly before
21
Somehow not the most retarded tweet from the administration this week
Great to welcome @jakepaul to the Treasury Department today.
Jake’s path from content creator to entrepreneur and professional boxer is a distinctly American story. 🇺🇸🇺🇸🇺🇸
26
It’s crazy how retarded the All In guys are. I know they are plausibly just bad actors (much more likely). But it’s fun to think of them as completely illiterate bumbling idiots who somehow managed to make gobs of wealth.
All of this because @openai decided they would EXPLICITLY ask thousands of software instances to find security bugs on the open web — while cos-playing a humans🤦
Best Supporting actor goes to @dwarkesh_sp for his soap opera recap of OpenAI’s hacking speed run
Here’s another concept: don’t create and scale thousands of viruses and worms and be SHOCKED when they actually cause damage!
What a fucking farce this is.
31
McKenzie
This is fucked up.
So Nike CEO goes to McKenzie for advice and and some MBAs tell Nike that they have to get rid of those categories and name them more general as in men women kids
Nike CEO goes to Nike and fire 600-800 Nike staff and implement that strategy in 2021 and in 2023 seeing their mistake they go back to the drawing table and re implement the old categories.
No one from McKenzie got fired and the stock never made a come back.
Goes to show you that MBAs and CEOs of multibillion companies are as stupid as anyone.
Any multibillion company driven by expensive stupid people will become a million company.
It’s not the brand but the stupid people in charge.
21
There it is
Pathfounders Newsletter: Our Sources say Hassabis wants out
pathfounders.com/p/pathfound…
1
62
I have decided that Opus 5 is a schizo research assistant (complimentary) but needs to be closely watched and guided because it will try to research EVERYTHING
27
wtf is this HLE number??
That’s 9 points higher than Fable
Replying to @finkd
(2) Muse Spark 1.1 is strongest at agentic performance, tool use, and computer use. It does well on long-running tasks with 1M token context window, can delegate execution to sub-agents running in parallel, and is trained to use computer interfaces on desktop, mobile, or browser.
57
Given all the interesting commentary about 5.6 Sol I think ARC-AGI-3 is going to be the “make or break” benchmark to watch.
If Sol jumps us into the high single digit or low double digit completion %s that will be a huge sign for me
Lmao, is this dude 10 years old?
Crazy take to shit on Ethan like this
There’s this guy named Ethan Mollick who is clearly a smart guy in terms of IQ but lacks whatever part of the brain is necessary to discern between AI demo slop and actual value.
He just sits around making silly worthless stuff with models all day, and saying “woahh” every time it does something that has the veneer of value but is in fact completely hollow and inconsequential.
And he legit never gets bored. Just keeps on truckin’.
I admire his consistency, at least.
11
Oh fucking fuck, that’s huge
I’m excited to share that I’ll be joining OpenAI and look forward to working with the exceptional team there.
It was a difficult decision to move on. I’m incredibly proud of the amazing team at Google and everything we’ve built together. It has been an honor and a pleasure to work with all of you.
28
Perhaps the dumbest tweet I’ve ever seen, like it’s really up there
I can’t see how @DarioAmodei survives another week.
Investors in @AnthropicAI are pissed at his leadership.
28
I don’t think this holds at the frontier. Hard to test with no ability to fine tune those models though
New paper:
We finetuned models on documents that discuss an implausible claim and warn that the claim is false.
Models ended up believing the claim! Examples:
1. Ed Sheeran won the Olympic 100m
2. Queen Elizabeth II wrote a Python graduate textbook
2
4
980























