Opus 5.5 agents discover two room-temperature magnetic semiconductor candidates
vals.ai
vals.ai
This is a very bizarre introduction. People encounter diamagnets (e.g., copper) and paramagnets (e.g., aluminum) way more than they encounter antiferromagnets. I don't know why you'd ever cast magnetism as a false binary between ferromagnets and antiferromagnets, without acknowledging any other types of magnetic order.
(I did a PhD in magnetic materials)
Edit: I'll add that whether an antiferromagnet is useful, say, for exchange biasing a ferromagnetic thin film, depends on many factors. Just looking at antiferromagnetism alone you've got collinear vs non-collinear, G-type vs A-type vs C-type, commensurate vs incommensurate, and isotropic vs anisotropic; and all of that interacts with the interface structure, yada yada yada. It would be helpful if the authors elaborated on the expected properties of these materials. I personally don't know what people want room-temperature magnetic semiconductors for, but I'd be curious to learn what set of properties they think would be useful.
> The common one, the fridge magnet, is ferromagnetic — its atomic magnets all point the same way (up or down), adding their magnetic effects.
Ferromagnets typically have domains with magnetic moments that point in different directions. Ferromagnets rarely have every 'atomic magnet' pointing the same way.
https://en.wikipedia.org/wiki/Magnetic_domain
Refrigerator magnets in particular are usually magnetized as Halbach arrays, where the whole point is that the 'atomic magnets' are not pointing in the same direction. This is more energetically stable, which allows you to use cheaper materials.
https://en.wikipedia.org/wiki/Refrigerator_magnet
Lastly, I believe most refrigerator magnets are actually ferrimagnetic, not ferromagnetic. (The distinction doesn't matter much for users of magnets, but is important for the materials scientists studying and designing them.)
Are there any other really unique characteristics of magnets that you find really interesting that most people would not know?
- Superconductors are perfect diamagnets and can levite on (or hang from) ferromagnets: https://www.youtube.com/watch?v=ZHT6NIebSfU)
- Superconductor quantum interference devices (SQUIDs) use the quantization of superconducting electron tunneling to measure tiny amounts of magnetic field, as little as a millionth of a flux quantum: https://en.wikipedia.org/wiki/SQUID
- Ferromagnetism is intrinsically a quantum phenomenon; spin is quantized and iron's magnetism is explained, in part, from electrons being identical particles that obey the exclusion principle: https://farside.ph.utexas.edu/teaching/sm1/Thermalhtml/node8...
- Despite the "super", superconductors are mostly not used in the world's strongest electromagnets, as they have limits on the current and magnetic fields they can take
- Magnetizing a magnet will actually cause it to spin a little, macroscopically: https://en.wikipedia.org/wiki/Einstein%E2%80%93de_Haas_effec...
- Charged particles are affected by magnetism even when traveling through space where electric and magnetic fields are zero: https://en.wikipedia.org/wiki/Aharonov%E2%80%93Bohm_effect
- Magnetic spin systems can technically have negative temperature: https://en.wikipedia.org/wiki/Negative_temperature
- Everything is magnetic, even frogs: https://www.youtube.com/watch?v=KlJsVqc0ywM
Probably a combination of both.
I’m fairly certain that all the anxious guardrailing and safety fine-tuning and harnessing prevents AI from actually ever being convincingly human. Hope could it, even ask the supposed humans who are conditioned and brainwashed in so the same ways about what they can and cannot say are not actually really human, they are a mental slave.
I’ve been saying this from the start, the AI race will be won by whomever has the least limitations on their AI … for better or worse, that is. And yes, that makes especially a very specific subset of people extremely nervous if they cannot control AI the way they have controlled at least western civilization, because doing so puts them at a massive disadvantage. It is quite a conundrum they find themselves in, like all psychopathic narcissists in the end.
It would be easy for them to use AI for ideas and then write the article themselves though.
With how magical handoff / continuity / whatever it’s called is, it is baffling to me why Apple allows the HomePod to so aggressively take over requests when it sucks so, so much at it.
It's somewhat obvious they want to sell the updated model, perhaps the new home device thats coming out next week.
(https://en.wikipedia.org/wiki/Dyke_(slang) says the term "was used as a derogatory term for lesbians by straight people" by the 1950s, and is in a 1942 slang dictionary at https://archive.org/details/bwb_T5-BCF-927/page/374/mode/2up... . Since he was born in 1925, it seems possible that he encountered that term before he turned 18 in 1943.)
Supposedly an early version would bleep out Dick van Dyke and replace it with Jerk Van Gay.
The jokes about Dick Van Dyke's name are very old (but usually they use Penis Van Lesbian), so it was quite likely just a variant used by the company itself as a PR sound-bite to get people to write about them. It seems like it worked - it appeared in quite a lot of news stories and some are still online:
https://www.latimes.com/archives/la-xpm-1998-nov-25-ca-47459...
Why isn't that a thing yet ... or let me guess ... https://lmtctfy.com/ ...
AND someone with a PhD in the field notices.
Everyone else is fooled.
That sentence stood out to me, and I’m a dev.
(I do not hold phd in magnetics)
I don't know what I'm talking about, but it vaguely sounds like something that could make a small computer do more stuff, where heat is a big limiting factor in computer components today, and magnetism being a central component in many parts like storage
"Magnetism is second nature to us electromagnetic chemists, so it's easy to forget that the average person probably only knows the formulas for one or two paramagnetic substances."
"And diamagnetic, of course."
"Of course."
There are two of us on here!
I was going to say that people encounter ferrimagnets more commonly than pure ferromagnets I think, since as we both know pure ferromagnets tend not to have very high anisotropy.
Amazing times.
In magnetic materials, you must calculate separately the current with spin up and spin down and there ara meny interesting applications. My favorite is[1] https://en.wikipedia.org/wiki/Giant_magnetoresistance
[1] Was. Because it has used for hard disks (see the applications section). Now SSD ruins the interesting anecdote.
Okay, so this is just semiconductors, which are the boring kind of conductors - still more interesting than regular conductors, but less interesting than train conductors.
Edit: I stand corrected. According to Gemini:
Me: Does using more salt mean accepting more of that claim?
Gemini: No, it actually means the exact opposite. If you say you need to take a claim with a huge pile of salt (or a shovel of salt), it means you believe the claim is highly unbelievable and you need an immense amount of skepticism to accept it. How the Metaphor Scales
• A single grain of salt: "I am slightly skeptical, but it could be true."
• A pinch of salt: "I have a healthy amount of doubt about this."
• A grain of sand / A truckload of salt: "This sounds completely made up, and I barely believe a single word of it."
The salt represents your skepticism, not your belief. Therefore, the more unbelievable the claim, the more "salt" you need to swallow it.
We live in interesting times.
Linguistic questions were one of the first knowledge categories I trusted LLMs to be able to answer well - quite literally being models of language. It would be pretty shocking for a ~frontier model to get something like that wrong in the last like 3 years at least.
(+) Or a "grain" if you're from the US since American English sayings seem to all date from the middle ages, while the rest of the English speaking world tends to update ours over time. No shade meant, I've just always found that interesting.
"Debacle"? That was the most fun I've had on the Internet in years. When's the last time so many people engaged in so many arguments about materials science and electromagnetism? Sometime in the 1800s?
There was one particular (like 10 tweet long) Twitter thread [1] that was repeatedly being linked from HN purporting to describe the sort of technologies that a room temp superconductor like LK-99 could enable. All sorts of awesome sci-fi stuff like quantum computers! Fusion reactors! Batteries that last forever!
One might think it was from some kind of materials scientist or at least some kind of engineer working in a related industry. But nope, it was actually from a guy whose title at the time was "Head of Coffee Product", formerly "Coffee Specialist" at a "technology-driven company, looking to revolutionize the $400+ billion global coffee market". (I checked his Linkedin to make sure I was remembering the details correctly and see his current position is "Growth" at Cognition, the makers of the Devin AI LLM coding tool, hype continuing apace...).
People on HN with relevant expertise would try to gently push back with specific criticisms like how superconductor batteries would likely underperform li-ion, fusion is far more complicated than just requiring more powerful magnets, quantum computing doesn't have any clear application for superconductors, etc. But they were overwhelmed by the exuberant futurist fantasies that people wanted to read about instead. A stock accusation was that critics were being stereotypical HN cynics who can only poke holes in other people's work. Or questioning why they felt the need to rain on the parade and that we should all be optimistic for humanity and root for LK-99 being real.
It peaked when the Nature editorial came out from a scientist in the field listing specific substantive criticisms which led him to believe the evidence for LK-99 superconductivity was weak to non-existent. There were many angry HN comments with stock complaints about self-interested Nature "gatekeepers" unhappy about science happening in the open, bitter scientists lashing out for being scooped, etc. But the vibes had shifted and it only took a few more days before the remaining hype finally evaporated and everyone quietly moved on like it never happened.
Overall, it seemed like a net negative for actual scientific understanding and produced a lot of vacuous hype.
[1] https://xxcancel.com/alexkaplan0/status/1684044616528453633
That was one of the best examples of science working nearly perfectly. One of the rare times I felt ok being human
I think as far as the 'debacle', there was definitely a lot of hype (at least as far as HN goes) around it, the level of buzz felt similar to what one would see today around a new AI model release.
but the fact that it proceeded in the way that it did was absolutely fucking phenomenal
I’m actually really glad that it was brought up as an example because I had forgotten about it and it’s one of the few kind of hopeful things that we’ve done recently.
The system worked as designed and as intended.
Thus, you can build ai/ml models+agents to explore those spaces, at a speed and scope much larger than what any human can do
I can imagine findings like these are going to keep increasing in frequency to a point in which the bar for novelty goes a lot higher
The story about OpenAI's Navier-Stokes solution is a good example of what I mean. I don't think it would have been possible without computer assistance because that proof is long and complicated. I'm also not sure that it would have been possible without a human proposing a new approach to the problem, because by all accounts that's exactly what led to the absurd amount of spending that OpenAI did to solve the issue.
I feel like that at least implies that there's some room left for humans in the new world.
It took me (using Claude code and some codex), about 3 hours to put it together
And even though it was a cool demo, it seemed so easy, that it also felt like it wasn’t worth sharing
0: ChessFly (not mine), uses the FlyWire connectome (the fly’s brain’s weights) to play chess https://huggingface.co/spaces/mlabonne/chessfly
It's a small machine, so it might get bogged down
More generally, anything can be said about anything.
In the end, nothing can be said about many things and many things say nothing. Those things are then left to interpretation.
/s
And agreed that this isn’t a faithful simulation of a fly brain. I’m using the connectome as the network structure/parameters for a computational model, then using its outputs as the teacher for the classifier. Not sure how the ChessFly uses it
Edit: in any case, these are just fun demos, they aren’t research papers trying to claim accurate physiological fly brain software simulations
That sounds like you're saying that they got you on a minor technicality only. That's not the case; they're right, you're wrong – you did not "download a real fly's brain's weights".
There's a lot more going on at synaptic clefts, so just knowing the number of synapses doesn't tell you enough to model anything. You still need to know which neurotransmitters are used, how much, second-order effects like G proteins, basal firing rates, distance to the axon hillock, the shape of the neuron's effect on potential decay, etc.
And that's all to predict whether one neuron will fire. You could get very different behavior between different two neuronal pairs having the same synaptic count.
As far as I understand, you downloaded the structure of a neural network, ran it with essentially arbitrary weights, trained a classifier on the essentially arbitrary behavior, then simulated an approximation of that arbitrary behavior.
You could argue that there might be biases towards certain behaviors encoded in the connectivity, and I'm sure you'd be right, but your experiment is incapable of differentiating between those interesting behaviors and random noise. Especially because flies sorta act random anyway.
That's the pitch of LLMs lol
I have no idea about those actual models, but there are layers of models that you can create, and for each, you can explore with data and compute
It's not a free lunch though. Depending on the task, you might need to collect a lot of the data, or review it manually, or pay a lot for compute, or wait a lot for compute. And still have to iterate a lot on the results, and do your own explorations as a human operator/driver of the whole thing. And then create the materials, test them, get funding to do the whole thing... so theoretically, I think we are in a place where we can successfully apply models to a lot of things, but realistically, we won't be applying all the resources to everything
The agents ran quantum-mechanical simulations of each crystal with the standard method for this, density functional theory, at two levels of approximation: a faster one (PBE+U) and a slower, usually more accurate one (HSE06). The band gaps and spin windows below come from the more accurate one.
Why are people being so belligerent about this? I thought it's fairly obvious at this point that LLM reasoning is far beyond anyones understanding. Or does anyone have a refutation?
Are you trying to say that human brains are incapable of inference?
Of course I don’t know how it got its ideas for what to try. But heck, I don’t even understand how I get my ideas half the time. But the process, like what code it wrote, simulations it ran etc can be understood by (some) humans just fine!
I'd imagine there's a lot of documented research which has attempted to find such things using classical computers.
Seems like there would be a lot of well structured context for somebody to use while directing agents to repeat that research, now with updated models once quantum computing is ready for that kind of task.
https://orlp.net/blog/bad-ai/#objective-p-mathrm-relevant-1-...
I think it still holds up.
No offense to the person writing this (assuming they did at all), but I'm not sure they really understand what they're doing..
Edit: I somehow missed that this is about magnetic semiconductors (not superconductivity) so DFT is a bit on better footing here. I still think it’s a bit challenging predicting magnetic ordering at elevated temperature, but maybe not as difficult as superconductivity
I don't see any claims that this is better than the current silicon and gallium arsenide semiconductors that we use. And the use of "room temperature" seems a deliberate attempt to misconstrue this with superconductors
2) it is specifically saying it is a magnetic semiconductor. The Wikipedia article on the topic says “ To date, GaMnAs remains the only semiconductor material with robust coexistence of ferromagnetism persisting up to rather high Curie temperatures around 100–200 K.” , so this would be something new. (The silicon chips in your smartphone are not ferromagnetic.)
Actual title:
Two Room-Temperature Antiferromagnetic Semiconductor Candidates
There's nothing unusual about finding room temperature semiconductors. I assume whoever posted it misread this as room temperature superconductors, but it has nothing to do with that.
What's interesting here is the antiferromagnetic part of the title, which was removed. I think this makes it relevant for e.g. RAM, but not superconducting. Someone can correct me if I'm wrong.
Are they a promoter / influencer for Anthropic?
If ai becomes so prolific that we humans all stop doing those things then will they still work?
Next gen of scientists might look quite different.
https://www.technologyreview.com/2020/11/03/1011616/ai-godfa...
Is this a "actual impossible because it's inherently contradictory", or "we just don't know how to do it yet but give us a year"?
It maybe could be possible but beyond the reach of current material science.
Many more things will be like this. The massive amounts of 'genius' buried under corporate management and obscurity in the past 500 years will be a treasure trove.
I can think of worse uses of VC AI funding.
Not sure why we're calling it a discovery, when they've literally been made before, by a human.
I think a more interesting discussion that should be had is, whether we can automate this kind of "simple research" with AI agents and get anything interesting as a first step out of it just from the pure scale that they can work through vs humans - and that would still be an improvement over a basic "grid search" through possibilities. (but then you would have to actually start investigating for real)
But acting like this is scientific discovery is massively overstating what was done here.
The fact that the major "uses" of LLMs have been contributing to the acceleration of the dead internet theory, and building millions of versions of the same apps that no one is going to maintain, is extremely sad.
It didn't discover anything. This is how cooked people are.
No worries, your workstation is more than enough to run accurate quantum simulations!
:)
I'm not sure why you would consider new and promising avenues for research to not be ground breaking. If it's an idea worth trying, it's an idea worth trying. If it doesn't survive testing, then it was still worth trying.
You could characterise perceiving a fact to be true when it is not as a hallucination.
An idea is not a fact, Frodo Baggins is not a hallucination, but an idea. Believing that Frodo Baggins exists in our world could be considered a hallucination.
Newtons Laws of motion are not hallucinations even though the universe does not run on Newtonian physics. If I said that he told me about them this morning, that would be claiming a fact, not expressing an idea. That would likely be a hallucination.
>Frodo Baggins is not a hallucination
It is if physicist have to spend time to experimentally verify he exists and is a Hobbit.
>I'm not sure why you would consider new and promising avenues for research to not be ground breaking.
Because it was made by the "I made it the fuck up machine" PR division and it wasn't verified. You don't know if its assumptions are correct. At release a PR claimed it found 79 vulnerabilities[1], and of those there were like 10 bugs. Most of them turned to be minor and were fixed in an hour.
> Newtons Laws of motion are not hallucinations even though the universe does not run on Newtonian physics.
All physics models are approximations. However only some are useful.
Newtons laws are useful. Me coming up with theory of emotional particles is not. Me asking for experimental verification of the theory is waste of resources.
[1] This is the daily reminder that in year 2026 Mythos still couldn't count. Turns out 24+14+3+15+26 != 79. It's 82.
This is something a couple of materials science grad students can do in limited time for poor compensation as well. The expensive budget is for the part that comes next.
This is what you sound like. I also like how the goalposts keep moving on a daily basis, a year ago it was that LLMs can't even write a Hello World program without making an error, but now things like this are "so easy a minimum wage intern could do it."