Two possible walls for advancing math with AI: first, it could be that the kind of problems that AI solves are the ones that don't open new avenues to concepts and ideas, and second, problems complexity is usually not linear, and the new and interesting problems to solve could be at a level of complexity beyond what AI can currently solve. Perhaps mathematicians could work to find ways to help AI cope with those walls.
I think OpenLDK allow you to use the Java world for sbcl when you don't want to use Clojure or ABCL. But using OpenLDK requires you to pay the interop cost and when speed of the java libraries is not a sharp constrain.
Here using images is incidental. In fact, what Fable is actually doing is generic encoding/decoding of a message into a medium. Ask it to hide a message in a poem and it should try to apply a strong (high signal-to-noise ratio) family of schemes, for example quantized, margin-rich encoding. Robustness of features depends on the medium, so Fable is selecting strong encoding and robust features tailored to it. This frames AI steganography not as a trick of computer vision, but as an emergent property of a model reasoning about information density across arbitrary media. Here emergent is just surprising general pattern matching and constraint-satisfaction and cross-representation mapping that allows it to express the encoding/decoding across media.
To meaningfully test a forecasting model, use a data-generating process with a high signal-to-noise ratio where a classical method (empirical conditional distributions, i.e., counting) can capture most of the signal. Then check whether the model captures that signal too. If you only test on high-noise processes, any model looks comparable to SOTA because there's little signal to differentiate them. Also testing on a range of signals with different signal-to-noise ratio would allow us to know how much of the signal the model captures in different scenarios.
Perhaps some people chose math because they were very good at it, they could understand one shot the difficult concepts and have not enough money to study other subjects because you have to move city or pay a lot. Those kind of people can move to others field if economics conditions are favourable.
It sounds like what these people are very good at is what AI is becoming very good at, so they will be losing their economically valuable edge. For example, it used to be that many math and physics majors went into software development.
A neuron is a dirty, contingent, blind-evolved part, like a screw in a particular machine. It is necessary for this particular machine, but not conceptual, not a feature. The strong activation on top-k images and the change when it is removed just mean that the screw is necessary for this particular machine, not that it's something else. The essay shows neurons are legible, but legibility isn't the same as meaning. Still, given math job prospects, I love screws, they are really not dirty.
The only way I see creators can continue earning money is by a new market in which the products they create can not be paid in money but in a new coin. I propose that the coin stays strictly internal and not convertible, that fiat enters only from the outside, and that exit is communal, not automatic. The viable solution is a federated network of small, closed credit loops. Inside each loop, creators use a non convertible token for work and tools. Cash enters through dues or grants, and cash out exit grants are decided by community vote based on need.
Small loops keep social trust high. When loops need to trade, they establish simple bilateral treaties with each other. This preserves value, protects human labor, and keeps decisions in local hands.
In chats with models like deepseek the LLM makes many basic math wrong claims. Maybe more advanced models can in special circumstances, guided by the work of humans and Lean, prove theorems. But just like in RL, the brute force approach of failing a million times may not be enough when the reward is far.
In math the reward is finding the decomposition that reduces the problem to simpler cases, and in many cases that reward is very far. Furthermore, the LLM may process the many cases fast, but the decomposition often comes from a human, and the model takes the credit for the proof.
This foundation could provide the LLMs with the decompositions, generated by human ingenuity, that they need. But once supplied, the model can run the cases and get the credit. Hence more credits for LLMs and fewer jobs and worse prospects for humans, even if it isn't an intended consequence.
When opinions conflict, if the first opinion is right and the second opinion is wrong you could end up receiving an unnecessary treatment that could hurt you. So I think you need to know for sure why the first opinion is incorrect, or at least avoid accepting the second opinion blindly, because the first doctor won't be there to push back.
Of course, I know nothing is ever certain here. Still, having two opinions or more data can be valuable, but only when whoever makes the final decision can actually use that information: detecting any bias and deciding with a good balance between risks and rewards.
Yes, this is a difficult decision - if conflicting treatment advice is provided by the two doctors, how do you decide between them? It's not necessarily the case that the second doctor is correct.
Of course, if additional tests are done by the second doctor and new information is uncovered, that helps.
My ex-wife was diagnosed with thyroid cancer, and the first doctor said that urgent surgery was needed and both thyroids need to be removed.
She obtained a second opinion from a well-known oncologist, who said that only part of one thyroid needs to be removed, and it wasn't so urgent.
Because everyone wanted to believe the second doctor, that was the option she went for.
This was a very expensive doctor. Shortly after the surgery, he notified her that, in fact, the cancer had spread to both thyroids, and the other one had to urgently be removed as well. This led to double the medical bill, opening up her neck twice, and all the accompanying inconvenience of two surgeries.
It was a strange situation where an incorrect assessment by the doctor directly led to that doctor making more money.
Also, it highlights the tough decision in terms of how you decide between two conflicting opinions if there are no additional diagnostic results to assist.
There is only one thyroid ... presumably you're referring to the two lobes.
There are generally clear discriminators determining whether a lobectomy or total thyroidectomy is called for. What is missing here is the detailed discussion with the two doctors of exactly why they were recommending their course of action and what attempts were made to reach consensus. "everyone wanted to believe" is not the way to go about such things ... but there's a good chance that the evidence favored the less radical course. A lobectomy has considerably less radical consequences and so is strongly preferred if a total thyroidectomy isn't indicated. But it's not uncommon for the pathology report of the removed lobe to indicate high-risk features that call for a "completion thyroidectomy", which is what your ex-wife had. It's not some nefarious conspiracy to make more money.
Thank you for the corrections and clarifications. It does still support the point that it can be hard for a layperson to make a call on which opiniom to adopt.
With "everyone wanted to believe" I was referring to the patient and family.
It was (is) difficult to get appointments with these doctors as they are fully booked months in advance and there was no opportunity to consult with them again or get them to discuss the case with one another. Hence the decision had to be made with the information that could be remembered from the consultations. (These days I always record all doctor's consultations for future referral, as memory / recall is of course notoriously bad.)
The difference is going to a national cancer center.
My late wife was treated for metastatic melanoma at MSK. Their approach is to have clinical teams and a tumor board that are assigned patients with related cancers. They are seeing many patients and there’s a warm hand off between specialists. Your radiation oncologist, medical oncologist and surgery / neuro specialists are on a team and know each other and are often familiar with the cases. In my wife’s case the community oncologist was wonderful, but he was a sole practitioner for this particular cancer.
In the second opinion scenario, they get to use that experience and caseload to look at a case with some perspective. Your community oncologist likely works on a small team and may not have the luxury of time and exposure to trials, etc.
Between November 2024 and January 2025, I visited liver surgeons at 4 different hospitals in 4 different states and got 3 different opinions on what was the correct next step for my cancer treatment.
I broke my leg at the edge of the Fibula bone. 4 Orthopedic specialists gave me 3.5 different opinions with different and conflicting recommendations for treatment.
So, unless the first two opinions are completely identical - yeah, definitely get the third. Better to go for people with slightly different specializations if relevant.
I've had a couple of second opinions for far less serious stuff in recent years. In both cases the original doctors suggested major surgery pre-emptively, even though they were very different types of thing.
In the first case, I decided to talk to an expert in the particular area who then ordered different tests. These indicated the issue had actually gone away completely, and I avoided surgery as a result of those.
In the other, a grade 2 ACL tear, it was less clear-cut, but I felt like the explanations given by the second doctor as to why he'd recommend against surgery were clearer and made more sense to me.
I'm glad I got both second opinions and the first case in particular, made me much more likely to get second opinions in future as well as more confident about doing so.
I understand that insurance is not your friend and you must advocate for yourself in the modern healthcare landscape, but if appointments take months to get to because everyone is overbooking limited oncology specialists to get their second and third opinions, at what point is this counterproductive for wider healthcare outcomes?
Particularly with cancer, when earlier treatment is better.
From the perspective of both individual patients and the system as a whole, the best long term potential lies in more accurate and widespread cancer screening. If it can be detected earlier then treatment becomes much easier and cheaper. The liquid biopsy assays available today still aren't accurate enough to recommend for everyone but they should improve. In a few decades I expect adults to be getting them every few years.
Some people are used to thinking deeply and concentrating, but are also curious and they are using LLMs to get more information faster. So for them, the use of LLMs seems fruitful. What I wonder is whether the tendency of letting the LLM do some of the work will affect them too, eroding, in the long term, their concentration powers.
Maybe. Or it just changes how they think. I'm comparing it with the advent of Google and the internet; when I went to college for software development, we were supposed to learn from a book but I quickly found googling for just the information I needed at that particular point in time was faster. So I never learned anything "deeply", instead I've been very picky in finding solutions through google and the like.
I think LLM assisted work is a similar shift in how people work.