Sure, but isn't the response worse than the problem? When intellectual standards are abandoned (as opposed to correcting a specific bias), discourse degrades into mindless anger, conspiracies and leaders who defraud their own supporters.
Given the money available to conservatives, why haven't they been able to setup vibrant universities, films, art etc. The point here is not to blame conservatives, but push the analysis of the cause to something deeper than just saying that the opponents have all the power.
Restricting things like creation of a highly infectious virus is very different from restricting books or even guns. There is no 'monopoly' over such a technology, as a use of the technology will inevitably harm the creators themselves.
Restrictions on high end biology, chemistry would leave overwhelming number of use cases of LLMs unaffected - no need to ban open weight LLMs. Such restrictions can be even more effective, if it is coupled to researchers getting early access to see the possible problems and have an opportunity to prevent the outbreak or create new vaccines well in advance.
Restrictions are not enabling monopolies. The opposite is true, if a LLM engineered virus or other harmful technology is let loose, public opinion can very quickly swing towards draconian regulation. (see nuclear power after Chernobyl).
Speaking practically your hypothetical is a scenario that requires somebody that is proactively interested in, and theoretically capable of, making a e.g. dangerous virus, yet are unwilling/unable to do so without a chatbot. How many people might this possibly apply to? I think the number is literally zero.
I also don't entirely understand your comment, because your latter parts do not follow from your lead. You're 100% right that somebody who's not extremely capable messing with this stuff is overwhelmingly likely to just hurt themselves. And somebody relying on a chatbot to guide them in dealing with this sort of tech? Yeah, they're gonna win a Darwin Award.
---
I also think there's an entirely different, yet also compelling argument, against censorship. Local LLMs already exist and are advancing rapidly. There will come a time, probably in the relatively near future, when the state of the art big system and a decent uncensored local system will become practically indistinguishable in terms of capability. So not only will people be able to do this locally, but you lose something big in the process.
The reality is that our interactions with LLMs are 100% being actively surveilled, regardless of privacy promises of the companies involved. At the minimum, every chat is making it's way over to the NSA's Utah data center, one way or the other. Some guy trying to do something significantly malicious using an LLM is little more than a gift to the authorities, but this is only true with centralized/online uncensored services. Push people onto local models to do nefarious stuff, and law enforcement is blinding themselves.
>Speaking practically your hypothetical is a scenario that requires somebody that is proactively interested in, and theoretically capable of, making a e.g. dangerous virus, yet are unwilling/unable to do so without a chatbot. How many people might this possibly apply to? I think the number is literally zero.
I don't disagree with the rest of your post, but this doesn't seem correct.
I think I'd phrase it that there probably already exist, or will exist, people with the inclination to cause global mass death, but don't have the knowledge or ability to manufacture a virus to achieve this.
The important part is being theoretically capable of. Fortunately there are massive barriers to doing things like synthesizing deadly viruses, and it's not just a matter of knowledge but of skill. For instance there was a Japanese death cult [1] that at its peak included not only many graduates of top universities in Japan but tens of millions of dollars in funding. But their escapades read a lot like a satire of incompetence.
That's not to say they were harmless - they managed to kill numerous people, but they'd have killed vastly more if they just drove some trucks into crowds as is becoming a typical weapon of terrorists. And I think the main reason is because knowing how something is done, and actually doing that thing, are radically different.
For a goofy analog, think about assembling sofas or even certain desks/chairs from a kit. That can actually be fairly tricky, to the point that there's an industry built around doing it for you. But there it's literally following like a few dozen steps with a carefully manufactured set of goodies and all tools right in front of you. Imagine doing something many orders of magnitude more complex where you're improving everything, have guidance that may be simply wrong, requires not only extreme skill but also a wide variety of difficult to acquire equipment, and if you make any mistake - you stand a decent chance of killing yourself.
If it just a mundane chatbot, the discussion is moot. But, we already have AI making breakthroughs in research and approaching the abilities do science just like a scientist does. (The last two paragraphs of your comment also assume such a high capability scenario).
Imagine giving the access, to whoever wants it, to a scientist who may not have many fresh insights, but has the advantage of a huge memory containing all the scientific literature in their mind, the standard patterns of deductions, and the ability to work at a very fast pace 24/7. They could identify vulnerabilities in biological mechanisms, just like AI identifies security flaws in code today.
---
Regarding hurting themselves, I was not referring to someone who is too dumb to follow lab safety precautions, but someone who has a nihilistic mindset. State actors and militia use weapons to take over and enjoy the power they acquire - they dont want to get killed by a deadly virus(unless they engineer and selectively apply the vaccine before they release the weapon - but this is very hard to keep secret). Someone who is nihilistic wont have such reservations on using the weapon even if it destroys them eventually.
Regarding restrictions on API LLMs leading to use of local LLMs, it is the local LLMs which will be used anyway (once they have the capability). That we live in a mass surveillance envirnoment is common knowledge. The bottleneck, where restrictions can be applied, is not inference but training which requires hundreds of millions of dollars. Chinese scientists have themselves spoken about AI safety concerns and it is indeed a threat to China just like anyone else.
Also, restricting high end weapons ability does not interfere with 99.9% of LLM usage (open-weights or proprietary) - so it need not interfere with business strategy.
I'm amazed we didn't have the same moral panic when the web became popular. billions of people suddenly had access to knowledge about how to create dangerous viruses! sites like Wikipedia don't even check that you're a US citizen before letting you access pages on recombinant DNA and genetic engineering! the articles on sarin and VX nerve gas include syntheses!
Wikipedia is a presentation of partial selection of biology textbooks and research papers, not using them as a collective brain to generate new artifacts.
There is a big difference between having a large bookshelf of programming language/networking/OS manuals and the ability to generate a functional software product which previously required a hundred or more developers. Even a hundred developers may not be able to find a subtle exploit in code which requires a tedious scan of millions of lines. Computer security hacks can be much less of a problem in comparison to exploits in biology.
Also, even Wikipedia (and public resources in general) have restrictions - there is information dangerous enough to be not published. In the 1930's itself, Szilard (who discovered the chain reaction) and Bohr advocated for restrictions on openly publishing research on uranium fission.
There are plenty of weapons (see custom made virus) which no state actor (or even an informal militia) would want to release, as these weapons attack everyone. But, open access to details of its construction leaves everyone vulnerable to motivations of small groups of crazy individuals.
Particles accelerating in a cyclotron at sufficiently high energy reach relativistic speeds. You have to account for their relativistic mass increase to get the cyclotron to work. Figuring this out was a big issue in cyclotron design in the 1930s. The remedy is to strengthen the magnetic field near the outer edge of the cyclotron where particles move fastest, by adding coils there
to carry more current. I don't recall what the energy is where this becomes necessary - it is certainly needed at tens of MeV.
Plutonium was first synthesized in a cyclotron by Lawrence's group at Berkeley.
I don't know what energy they used so I don't know if they needed the extra coils, but they did know of the effect and must have considered it.
Also, U235 was separated at Oak Ridge using machines called Calutrons invented by Lawrence that might have encountered the same problem -- at least they must have considered it.
The atomic bomb certainly had a cursory dependence on special relativity. E = mc^2, you know.
Klystrons were used in WW2, and the beam current of a klystron scales as the beam voltage to the 3/2 power (due to space charge limits). Modern klystrons operate with relativistic electron beams, but I don't know if any of the WW2 ones did.
Is there any part of the theory or design or implementation of the atomic bomb that depends on E = m*c^2? Or is it: "if you could theoretically weigh the end-products, you would get a slightly smaller answer than the before-products". Seems like it would have been known that there was a lot of electrical potential energy stored in the nucleus (after the discovery of the proton).
Calculating the energy yield of a fission event is done by comparing the masses of the initial and final states. The difference is about 1/5th the mass of a nucleon.
Another situation that occurs in nuclear weapons and in reactors is the scattering of energetic photons off materials. This process is inherently relativistic when the photons have energies comparable to or greater than the rest energy of an electron. (Differential) cross section computations are necessarily relativistic.
At Los Alamos, various techniques using energetic photons and electrons were used to diagnose implosion systems in development of the plutonium bomb.
This doesn't seem to be investment focussed activity, but rather extending Claude credits for education and research. Which is a good thing, independent of other bad things that might be happening.
Anthropic profits from the PR, for one. And they likely hook these institutions on their products in the long term, for two – much like I was "stuck" on Azure until recently, thanks to their free startup credits pointing me to it a decade ago.
There are ways one can engage in financial engineering (is "accounting engineering" a term yet?) where despite not making a profit, you segregate a tax break, tax credit, charitable deduction, etc. into some other entity and then can sell that off as an asset that some other business that is making a profit buys and writes off against its own profits.
This is not about employer vs employee and job security. In fact, the post mentions that there could be good reasons for layoffs. What the post highlights is -
1. Trust - When an employer tells the employee something and then ignores it - then a truth based culture gives in to cynicism. Communications in the company become suspect. Even when there are win-win situations, where cooperation could lead to positive outcomes for both management and workers, a lack of trust means the company cant execute.
Also, this will affect communications with customers and shareholders.
2. Regardless of being right, the author is helping others in similar situations, who can adjust their expectations.
3. The post isn't so much about company vs employee, but competing factions within the company, who are invested in alternative tools/proposals. Promotion is used as a means of making one's faction stronger. This need not be for the benefit of the company or customers. Lobbying will also, of course, affect truth.
Factions might be inevitable (and there can even be good reasons - people genuinely have differences of opinion). But, if the company has good leaders, they will prevent this from erupting into a strong zero-sum conflicts which drown other goals - company's profits, promoting competent people, a culture of trust.
There is a purely geometric reason for why elliptic curves have group structure. A geometric shape which is also a group, such that the group operations are smooth maps, has to be homogeneous - it has to look the same from every point[1]. Not just that, if you have a vector at some point, there is a natural way to transport it to every other point on the shape. The only surface (curves over complex numbers are really 2d surfaces) which obeys this property is the torus[2].
[1] Why should the homogeneous property be true? Because in a group, multiplication by g, pushes the identity e to g. M_g(e)=g where. This is a continuous isomorphism of the shape. So the shape looks the same at g as it looks at a (a neigbhourhood of g looks the same as neighbourhood of e). So an 'X' or 'Y' shapes cant be groups, as there are points which are locally unique, but 'O' shape can be a group. Moreover, M_g can also push a fixed non-zero vector v at e to a vector v_g at g.
Don't want to get into low quality generalizations in your post except to note tahta casual Google search will show you that Tata group is one of the most philantropically oriented groups. Which of course, doesn't excuse this issue.
This is a great way to present the concepts. Something like this would have been useful some years back when I was trying to use the Haskell library for lens.
Given the money available to conservatives, why haven't they been able to setup vibrant universities, films, art etc. The point here is not to blame conservatives, but push the analysis of the cause to something deeper than just saying that the opponents have all the power.
reply