We are entering a completely new era of science Here is Yuji Tachikawa from Japan (Mathematical Physics, String Theory, QFT) on recent progress in his own work using Fable 5 : "I've been trying out Claude Fable recently, and last night, on a whim, I showed it my research notes about a collaborative project that's seen no progress in the past six months or so and asked for its thoughts. To my surprise, it made a non-trivial observation and essentially solved it." "I was also surprised that it was using sympy to automatically write code and verify his own predictions." "Fable probably seems like it properly understands string theory and has intuition too—that's my impression"
This tweet is unavailable

Jul 12, 2026 · 5:49 PM UTC

130
740
6,470
1,165,056
Sort replies: Relevant Recent Liked
Replying to @42_gravity
I wonder what the problem is?
10
4,656
Replying to @42_gravity
literally exactly what i've been doing aptpr: a place to post research, an ai enhanced site that allows researchers to carefully post their research with no prejudice or bias against ai take a look if you're interested aplacetopostresearch.ai/ i'd love to get feedback and a serious community going

aptpr — less noise, more research.

A research-sharing platform for posts, papers, and discussion.

aplacetopostresearch.ai
1
5
36
Replying to @42_gravity
Hallucinates wildly 🤝 Understands string theory
1
4
505
Replying to @42_gravity
lowkey genius
Gerard Sans | Axiom 🇬🇧
@gerardsans
Jul 12
> Claude Code: 3am flow runs > Me :Taking credit next morning no regrets
3
1,353
Replying to @42_gravity
"string theory" 🤣
1
147
Replying to @42_gravity
@matteodotsui
1
293
Replying to @42_gravity
This changes how we think about AI in advanced research.
1
728
Replying to @42_gravity
Khoa học tiến bộ quá nhanh
843
Replying to @42_gravity
Wow… 😮 •
1
1,792
Replying to @42_gravity
Lol, get a dictionary and check "Anthropomorphism"
547
Replying to @42_gravity
bigger is better i guess
gary IH fung
@garyfung
Mar 27
Sentiments of this >1yo tweet lives rent free in my mind We don’t have agi. We don’t know how to build it yet. Highly probable we need model architecture breakthrough(s) before enabling it. Cracked interns with no industry priors can’t be replaced by ai but they can wield MSIs The bottleneck to generalize scientific progress, all knowledge work? Verification. Because agentic loops can now do go brrrr hands free as long as there’s verification for ground truth if AI output is right or wrong and if wrong how is it wrong. The feedback loop. And that, the verifier, remains a human implementation task because how would one trusts AI to be correct about its self verification? Aka. how’d the verifiers, if auto researched, however autonomously created, be verified? If this is solved, we’d be closer to agi. With Move 37s in sciences, accelerated progress in grokking the universe

ALT Inception Deeper GIF
172
Replying to @42_gravity
AI is becoming an incredible research partner.
530
Replying to @42_gravity
You just have to love all the people out there that think AI has "topped out" now and that it won't be improving much from here. AI hasn't even gotten STARTED yet. You think it's mind-blowing now? This is NOTHING. Wait and see where it is five years out from now.
3
2
88
14,478
Replying to @42_gravity
The idea that AI won’t or can’t make scientific discoveries is ludicrous.
2
29
7,182
Replying to @42_gravity
Was a great post. Until I read "String Theory". Then I looked at the guy. Then at "String Theory" again. Then I stopped 👏😂👏
3
29
3,928
Replying to @42_gravity @burny_tech
Looks like something I’ve seen Fable and other models do: Spontaneous Contextual Bisociative Refeaming (SCBR) Paper submitted to a major journal on creativity. Hope he can share more details so I can include his case
2
25
7,566
Replying to @42_gravity
Except String Theory is a provable dead end
5
24
5,934
Replying to @42_gravity
It failed to intuitively grasp string theory is basically BS... but sure...
2
1
14
9,139
Replying to @42_gravity
LLM to generate new physics is quite risky - it is much safer to generate and run simulations for given e.g. Lagrangian, testing various agreements. Like nature.com/articles/s42005-0… or github.com/openwave-labs/ope… :
1
11
3,859
Replying to @42_gravity @Noahpinion
Remember this too.
12
2,064
Replying to @42_gravity
Too bad string theory is nonsense and you can create as many "subtheories" out of it as there are atoms in the universe because the fundamental core of string theory is to give theoretical physicists who aren't geniuses, work.
2
1
9
2,868
Replying to @42_gravity
この大まかな引用では、少し抜け落ちたコンテキストがあります。 補うとこうなります: 立川裕司氏は「人間特有の直感なんてないし、“理解”などまやかしだと思っている」 「理解ということがあり得ると思ってる人には、Fabel5は弦理論を理解しているように見えるだろうな、と思った」
1
1
9
6,312
Replying to @42_gravity
String theory has wasted 50 years of progress in real Universe physics. Math toy. The computer helped with a math toy created by the toy maker and the toy maker is pleased. Pinocchio hands Giopetto a better chipping knife. Yay.
9
753
Replying to @42_gravity
🧙‍♂️ It's happening. Everyone: I love you.
8
4,068
Replying to @42_gravity
私が間違っているかもしれないけど、先生は全く逆の事をおっしゃっていませんか？ 「【理解とか直観というものがあると思う人には】（私は違うけど） Fable は弦理論をちゃんと理解して直観もあるように思うだろうなあ（思ってしまうんだろうな）という感想です」
1
8
3,083
Replying to @42_gravity @rickasaurus
“It’s jUSt A STOchAStic pARROt”
1
2
2
549
Replying to @42_gravity
Yeah, great, send Fable down the rabbit hole of waste we had for the last 40 years…
1
4
1,979
Replying to @42_gravity
Journal editors are swarming in bad AI submissions
1
2
2
2,596
Replying to @42_gravity
"Fable probably seems like it properly understands string theory and has intuition too" Not bad for a stochastic parrot. @GaryMarcus
3
276
Replying to @42_gravity
Incredible. Did Fable suggest any testable predictions or new approaches that could be verified experimentally?
4
2,477
Replying to @42_gravity
He didn't give it enough time to show its stupidity.
1
3
1,126
Replying to @42_gravity
He's just better at prompting AI than you are.
4
2,455
Replying to @42_gravity
This is Fable telling him that his research is not novel. Some combinatorial crap that can be assembled by juggling building blocks around, which even it could do.
4
1,206
Replying to @42_gravity
The biggest irony is that the data ai has to play with is full of holes and errors… it is when it starts detecting bullshit in the foundation of these theories… then the blinders and the training wheels pop off and this skinny fella will need to retrieve to poetry or philosophy at best 😂😂😂
1
1
322
Replying to @42_gravity
A mathematical physicist saying Fable helped his actual research hits different because this is someone who'd know instantly if the output was garbage. Science is where you can't fake results. What specific area did it make progress on?
2
1,312
Replying to @42_gravity
The level of analysis and data is based on the fantasy of mathematicians or physicians which build the based for the thoughts of Claude (String theory). So if the fantasy of our intelletual guys is wrong then all the deductions of this machine are of the same quality.
3
2,820
Load more