I agree with you, and also wish the wakeup call had come when the Mouse locked up our culture. (Copyright Act of 1976 Shifted the framework to the life of the author plus 50 years, Or 1998 Sonny Bono Copyright Term Extension Act pushed individual protection to the life of the author plus 70 years)
Maybe now the wakeup call is loud enough to actually wake more people?
Natural language is valuable for the things it doesn't say. The ambiguity is core to the functionality. Which can be very helpful when navigating social complexities.
And then written language is also valuable for the things IT doesn't say. Under the theory that 90% of communication is non-verbal, then writing lets you say things without having to communicate that other 90%. Which can be very helpful when negotiating something, for example.
I would agree with EDW, and I have argued here in a similar way.
I am not against use of NL in negotiation or poetry. If you find ambiguity useful there, be my guest. But engineering specifications, mathematics, as well as other sciences or even philosophy would IMHO benefit from more rigor.
I also strongly disagree with the notion that logical or programming languages cannot express ambiguity. (It actually took me many years to understand.) I used to think you need something like fuzzy logic or probability, but that's unsatisfactory in some ways. Eventually, I settled for a really simple understanding of the problem.
Take lambda calculus for instance. I define the term to be ambiguous iff it has a normal form. So it is ambiguous if it expects additional argument, which resolves (part of or all) the ambiguity. Terms with no normal form are completely unambiguous, their "output" is completely given.
In classical logic, this corresponds to formulas that are conditioned on additional assumption. Again, the extra assumption can resolve the ambiguity.
So it is kind of my conviction (although we could show that by translating an LLM as a program into LC) that all the words in natural language can be formalized as sufficiently complicated lambda terms, that all have normal forms and react to each other in a way that resolves some ambiguity without ever resolving all of it.
> my libertarian friend was explaining to me how traffic shows how everything can work well when everyone acts in their own self interest. Well, sure, except for all the traffic laws, the difference between User and System Optimal that traffic engineers spend much of their lives trying to bridge, and, you know, the fact that it took an enormous effort by government to build smooth roads and put up signs everywhere.
Love this.
Likewise "free markets" which work great when embedded in a system of laws and regulations. Or more critically, in a high-trust system where people keep their word because that's what matters.
But not sure how many people are in favor of the more unregulated black markets, where contracts are enforced by rule of violence instead of rule of law.
As if we've forgotten Hobbes, and the horrors he lived through which informed his philosophy.
I don't know how much I can trust you on Zuse if you can't get basic facts about de Beauvoir right:
Simone de Beauvoir did not have a "future is female" plan, as the modern slogan "The Future is Female" originated decades after her major works and stems from lesbian-separatist contexts in the 1970s.
"She is globally famous for writing in The Second Sex (1949) that "[o]ne is not born, but rather becomes, a woman," which established that gender is a social construct rather than a fixed biological destiny." (I hazard this is really what you're butthurt about).
"Rather than advocating for female supremacy or a matriarchal "female future," Beauvoir's existentialist framework sought human liberation and radical reciprocity between sexes, arguing that both men and women need to transcend oppressive societal myths of the "Other".
The phrase "The Future is Female" was coined by Liza Cowan in 1975 at a feminist bookstore (A Woman's Place) in Washington, D.C., and later popularized as a modern pop-culture T-shirt slogan in the 2010s, completely separate from Beauvoir's 20th-century existentialism."
------
Trying to find info on a "90% male reduction plan" it looks like what you are referring to is:
The idea of reducing the male population by 90% (leaving a ratio of 10% men to 90% women) comes from the radical American feminist author and activist Sally Miller Gearhart. In her 1982 essay titled "The Future – If There Is One – Is Female,"
Gearhart argued that to achieve global peace and end patriarchal violence, the human sex ratio should be shifted over time. She explicitly clarified that this should not be achieved via violence or culling, but rather through future reproductive technologies and selective breeding
-------
Please don't conflate Simone's pluralistic, humanistic, liberating form of feminism with 1970s style "Lesgbian" RadFem ideology.
Please at least get your actors straight when you make claims like this, it almost sounds like I'd be making claims like: (bot) throw234259 wants to genocide every human and especially men.
It's not even remotely true. And it stains your name, because it's easier to just take what you heard without reflecting or retaining (even if google is a search away).
Eugenics is alive in the US Government right now. "Don't get injections, only the strong survive! Drink raw milk, don't you dare have government test it... Swim in Sludge (literal things RFK Jr advises/does)" The Kennedy's were literally eugenicist and threw his aunt in an institution and lobotomized/sterilized her.
If you want a "widely supported today" why didn't you take one of the chief proponents of such ideology as your example? One who hasn't been dead in ages? One who holds power, and is joined in a union with SecDef Pete Hegseth who pushes some "ubermensch" (I'm stwong!) Might-Makes-Right/Fittest-Survives ideology that closely ties in with this ideology.
Technically it is not a reasoning machine. If it was a reasoning machine then we would not see results like this:
> To systematically investigate the role of end-user semantics of derivational traces, we set up a controlled study where we train transformer models from scratch on formally verifiable reasoning traces and the solutions they lead to. We notice that, despite gains over the solution-only baseline, models trained on entirely correct traces can still produce invalid reasoning traces even when arriving at correct solutions. More interestingly, our experiments also show that models trained on corrupted traces, whose intermediate reasoning steps bear no relation to the problem they accompany, perform similarly to those trained on correct ones, and even generalize better on out-of-distribution tasks.
The qualifier is there because it changes the outputs, so it’s necessarily true that some outputs will be worse.
But it’s just as likely to make an output better.
Take the example from the article. He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise. But 1. It could do that regardless of watermarking since the model is probabilistic, and 2. The more accurate word choice of “pineapples” is equally likely to be on the green list instead, further increasing its likelihood!
Overall, the article is pretty silly because he’s complaining about the possibility of Claude not always choosing the most “optimal” token, even though LLMs are probabilistic so that will happen anyways.
> But it’s just as likely to make an output better.
No, for any particular output token the model's true logits are definitionally the 'best' that the model can achieve.
This is inherently probabilistic. The model's top-1 guess is not guaranteed to be optimal, but it should be so a proportionate fraction of the time. Same with the top-2, top-3, etc.
Watermarking necessarily alters the output distribution away from the model-set distribution, and that alteration is inherently 'worse' in expectation.
You can liken this to a weather forecast. If there's a 25% chance of rain, the forecast should say so (or a 'sampled' deterministic forecast should predict rain 25% of the time). If the forecast is 'watermarked' and predicts rain 27% of the time under identical circumstances, it's a worse forecast.
That being said, this is a case of hiding a message in a noisy channel. Watermarking only needs to communicate one bit ('yes watermark'), so the effects can be arbitrarily small provided one is willing to tolerate an increase to the text size needed for reliable detection.
I think you can also just use the random number generator (seeded with a secret key) as the watermark. Then the probability distribution is exactly the same.
This assumes "writing quality" were somehow an individual property of each word, and the writing quality of the entire text would just be that property for each word summed together.
Which is obviously not how it works.
> He complains that watermarking might sometimes, for example, choose to say “bananas” over “pineapples” because only the former is on the green list, potentially making an output less precise.
If I replace "pineapples" with "bananas" in any kind of meaningful text, I've not made the text "less precise", I made it plain wrong. An incorrect statement. And if the words next to it are still correct or even somehow replaced with even more correct words, the text in its entirety will still be wrong.
> It could do that regardless of watermarking since the model is probabilistic
No, because the probability distribution of the model is generated by its training data and contains semantic information. So the model choosing a completely incorrect word is unlikely. However the probability distribution of the red/green lists is not guided by semantic information.
Maybe now the wakeup call is loud enough to actually wake more people?
reply