Ryan Cavanaugh

The median top Quora post is going to be pretty OK, and the median ChatGPT reply is going to be pretty OK (probably better, tbh), and in both cases you can dig or badger your way into bad advice. They are not qualitatively different in any way.

People are often looking for answers from a place of motivated reasoning. Going to Quora and seeing three posts that say you should see a dentist for an abscess and one post saying you can fix it with a coconut oil rinse actually is a really bad situation!

Quora - can you fast for 90 days - yes, providing...

The argument is that "just read posts from humans instead" is an alternative that is equally capable of giving extremely bad advice

You should not trust an LLM for important medical/legal/anything advice, but you should trust random search engine hits even less than that

Between ChatGPT and Quora, which one do you think tells you it's safe to do a 90 day fast?

Whoa, where's that policy? I bring my RadWagon on the LR all the time

Gemini attempts to bribe the CoT summarizer with ASCII coffee

An 8-pump gas station is pushing up to 200 MW

Babe wake up Haiku 4.5 just invented a new kind of psychosis

You can just go on Quora today and find people advocating for the benefits of doing a 90-day water fast, so maybe that would be a good place to start too

Instructions unclear, crashed through the wall of the battery store

I swear to god I thought this was fake, but my copy just showed up in the mail. Costco is this a distress signal?

I want the type system to be Turing complete AND the type system should tell me if it's not going to terminate

Very excited to announce my new role as special envoy to the office of Formerly Involved Recently Exited Department

Do they think that mortgage interest goes into the county treasury? How does paying off your mortgage have anything to do with anything tax-related? I cannot make any sense of this.

It will never not blow my mind that I’m a YouTuber who won’t accept money from, like Better Help or whatever bc I find it unethical and meanwhile all these multimillionaire Oscar winners are doing spots for online gambling and TurboTax. Like bro how much coke do you need

It's a pun. They're explaining the pun.

In baking, the first proof, or bulk fermentation process, is a crucial step in which one lets the entire batch of dough ferment as one mass, before dividing and shaping it into loaves. This project represents our preliminary efforts to develop an objective and realistic methodology for assessing the capabilities of AI systems to autonomously solve research-level math questions. After letting these ideas ferment in the community, we hoped to produce a more structured benchmark. We presented a diverse set of 10 research-level math questions, drawn from algebraic combinatorics, spectral graph theory, algebraic topology, stochastic analysis, symplectic geometry, representation theory, lattices in Lie groups, tensor analysis, and numerical linear algebra. Each question arose naturally in the research process of the authors and has been answered with a proof of roughly five pages or less.

Keep reading!

FirstProof # 8 A polyhedral Lagrangian surface 𝐾 in ℝ4 is a finite polyhedral complex all of whose faces are Lagrangians, and which is a topological submanifold of ℝ4. A Lagrangian smoothing of 𝐾 is a Hamiltonian isotopy 𝐾𝑡 of smooth Lagrangian submanifolds, parameterised by (0,1], extending to a topological isotopy, parametrised by [0,1], with endpoint 𝐾0 = 𝐾. Let 𝐾 be a polyhedral Lagrangian surface with the property that exactly 4 faces meet at every vertex. Does 𝐾 necessarily have a Lagrangian smoothing? Remark C.3. An expert wrote the following (paraphrased) summary of the argument. The general strategy is as follows. In Steps 1 and 2, the solution first performs a local smoothing near the vertices, relying on a nice linear algebra observation. In Step 3, the solution then argues that one can extend this smoothing along the edges. To do this, it makes a choice of linear coordinates (𝑠, 𝑦, 𝑥2, 𝑦2) near each edge. It then interpolates between the perturbations in this coordinate chart. To perform the interpolation, the agent invokes Lagrangian suspension, which is overkill but fine. In Step 4, the solution is presumably saying that as you send the size of vertex-smoothings to zero, you can correspondingly send the size of the edge perturbations to zero. The explanation for the f lux being zero is sketchy to say the least. Model Response (Agent B) Theorem. Let 𝐾 ⊂ ℝ4 be a polyhedral Lagrangian surface such that exactly four faces meet at every vertex. Then 𝐾 necessarily admits a Lagrangian smoothing. Proof. The proof relies on the rigid symplectic geometry imposed by the 4-valent condition. We demonstrate that the tangent cone at any non-flat vertex canonically factors into an orthogonal

The solutions to the provided problems are mathematical proofs

What? Lean is an automated theorem proving assistant. It's exactly that

en.wikipedia.org/wiki/Lean_(p...

The fleur-de-lis is an incredible touch

What kind of name is "Dav Idcrepso" anyway

"This will not solve housing affordability"

"This will not solve gaps in educational quality"

"This will not solve climate change"

"This will not solve wealth inequality"

They're hard problems! That's why we need to do more than one thing about them!

Difficult problems often *require* multiple actions to solve.

This lets opponents point to any individual action and say "this will not solve the problem", which lets them pretend they want to solve the problem even though their actions directly perpetuate that problem.

Hardscape only counts in Seattle city limits. Clearcutting to make room for new suburban tracts has no impact. I am very environmental.

5. There are criticisms of LLMs that are much more plausible. They are built on a large scale expropriation of the intellectual commons. Demonstrably true! They involve a radical and extremely worrying derangement of power relations in favor of a tiny number of men with weird beliefs. Also true imo.

4. I think that it is important that critics of LLMs get this. The 'LLMs are useless which is terrible and everyone is using them which is also terrible' shtick contained contradictions even in the beginning, which took a lot of work to maintain . Now it contradicts people's lived experiences.

Came here to post this. See you there.

We can look to the long history of inventions which were legitimately useful but shunned forever because the people who were selling them being over-hypey, like, uh,