Wow, thanks, I have not seen this comment, yet I hinted about this in some of my other replies that I’ve done before.
Yes, I think ML is fair use, but there it would also be fair to force something into the public domain/open source if, in order to be accrued, it has to make use of fair use at unseen amounts of scale.
This would be a difficult to make law, though. Current ML is very inefficient in the amount of data it requires, but it could (and should) be made better.
Now I sail the high seas myself, but I don’t think Paramount Studios would buy anyone’s defence they were only pirating their movies so they can learn the general content so they can produce their own knockoff.
We don’t know exactly how they source their data (and that is definitely shady), but if I can gain access to a movie in a legal way, I don’t see why I would not be able to gather statistics from said movie, including running a speech to text model to caption it, then make statistics of how many times a few words were used, and followed by which ones. This is an oversimplified explanation of what a LLM does, but it’s the fairest I can come up, and it would be legal to do so. The models are always orders of magnitude smaller than the data they are trained on.
That said, I don’t imply that I’m happy with the state of high tech companies, the AI hype, the energy consumption, or the impact on the humble people. But I’ve put a lot of thought into this (and learning about machine learning for real), and I think this is not a ML problem, but a problem in the economic, legal and political system. AI hype is just a symptom.
It’s not AI
It’s not AGI, it’s not general intelligence, and it’s not comparable to a human (well, you can compare anything, but human and ML are just very different things in tons of ways).
But it is AI. The ghosts that chase Pacman are AI. A search algorithm is also AI, dammit. Of course an LLM is AI. Any agent that maximizes a function is AI. You are just embarrassing yourself.
But then it does go on to quote materials verbatim, which shows it’s not “just” ‘extracting patterns’.
Is is just extracting patterns. Is making statistical samples of which token (“word”, informally speaking) is likely followed given the previous stream.
It can only reproduce passages of things it has seen many, many times. I cannot reproduce the whole work. Those two quotes can be seen elsewhere on the internet plenty of times. And it’s fair use there, so it would be fair use with a chat bot as well.
There have been papers published where researchers were able to regenerate an image that was present in the training set of Stable Diffusion. But they were only able to find that image (and others) in particular, because they were present in the training set multiple times, and the caption was the same (it was the portrait picture of some executive at a company).
when given the book and pages — quote copyrighted works
Yeah, you are not gonna be able to do that with an LLM. They will be able to quote only some passages, and only of popular books that have been quoted often enough.
Even if they started to use my service to literally copy entire books?
You cannot do that with an LLM.
Why are you defending massive corporations who could just pay up? Isn’t the whole “corporations putting profits over anything” thing a bit… seen already?
I hate that some corporations are burning money, resources and energy on this, and the solution is not to restrict fair use even further. Machine Learning is complex, but if I had to summarize in some way is “just” gathering statistics of which word comes next (in the case of a text model). This is no different than getting a large corpus of text, and sample it for word frequency, letter frequency, N-gram frequency, etc. It is well known that this is fair use. You only store the copyrighted works to run the software and produce a very transformative work that is a summary many orders of magnitude smaller than the copyrighted work. This is fair use, and it should still be. Changing that is gonna harm the public, small companies and independent researchers way more than big tech companies.
As I said in another comment, I would very much welcome a way to force big corpos to release their models. Make a model bigger than N parameters? You needed too much fair use in one gulp: your model has to be public, and in the public domain. I would fucking welcome that! But going in the opposite direction is just risky.
I don’t understand why small individuals think that copyright is their friend, and will protect them from big tech companies. Copyright will always harm the weak and protect the powerful as a net result. It’s already a miracle that we can enjoy free software and culture by licenses that leverage copyright in our favor.
“Theft” is never a technically accurate word when dealing with the so called “intellectual property”, because the digital content being copied without authorization is legal in tons of cases, and because, come on, property is very explicitly exclusive. I cannot copy my house or my car, but I can make copies of my works for virtually 0 cost.
Using data for training ML models is even explicitly allowed in some jurisdictions (e.g. Japan), and is likely to be fair use everywhere else. LLMs are very transformative, and while they often can produce verbatim copies of fragments of copyrighted works, they don’t store the whole works or significant pieces of them.
Don’t get me wrong, I don’t like big companies making big money. I would not mind a law that would force models to be open sourced. But restricting them to train their models on public data by restricting fair use, it would harm them very little (they could pay something if they are making some profit), while small researchers or companies would never be able to compete, because they would not have the upfront costs, nor the economic engineering to disguise profits and pay less.
Related: There is an article on LWN called Lua and Python, which is mostly about the approach of the two languages WRT being “batteries included” or not.
I think Lua being a bit barebones is 100% fine… if you just pair it with a good helper library, or set of libraries with a coherent API, that allows it to thrive. Then you can either use the framework library or not, depending on whether your project requires the extras, or can do without.
As a parallel, I’ve been doing C++ development for almost two decades, and I cannot imagine doing anything non-trivial without Qt. For example, Qt has a debug framework that pretty prints automatically most containers, and adds the newline also automatically. Also, QString is an actual string type, whereas std::string is more like QByteArray. It’s functionality that it’s essential for me (and it’s just the minimal examples… then Qt has all the GUI functionality, of course, but I use Qt even in console-only programs!).
This is surely opinionated on my side, and most C++ devs don’t see it this way, but my point is that a language with a “core experience” that it’s lackluster to you should not be a bad thing if the language is capable enough to provide an ecosystem with a good 3rd party library that adds exactly what you want. In the Lua ecosystem that maybe it’s Penlight.
But I totally get your point. Penlight doesn’t even seem to have a math library, so I found no round implementation there. This can be not a problem for some, but deal breaking for others.
I’d have to dig it, but I think it said that it added the PID and the uninitialized memory to add a bit more data to the entropy pool in a cheap way. I honestly don’t get how that additional data can be helpful. To me it’s the very opposite. The PID and the undefined memory are not as good quality as good randomness. So, even without Debian’s intervention, it was a bad idea. The undefined memory triggered valgrind, and after Debian’s patch, if it weren’t because of the PID, all keys would have been reduced to 0 randomness, which would have probably raised the alarm much sooner.
no more patching fuzzers to allow that one program to compile. Fix the program
Agreed.
Remember Debian’s OpenSSL fiasco? The one that affected all the other derivatives as well, including Ubuntu.
It all started because OpenSSL did add to the entropy pool a bunch uninitialized memory and the PID. Who the hell relies on uninitialized memory ever? The Debian maintainer wanted to fix Valgrind errors, and submitted a patch. It wasn’t properly reviewed, nor accepted in OpenSSL. The maintainer added it to the Debian package patch, and then everything after that is history.
Everyone blamed Debian “because it only happened there”, and definitely mistakes were done on that side, but I surely blame much more the OpenSSL developers.
Is it, really? If the whole point of the library is dealing with binary files, how are you even going to have automated tests of the library?
The scary thing is that there is people still using autotools, or any other hyper-complicated build system in which this is easy to hide because who the hell cares about learning about Makefiles, autoconf, automake, M4 and shell scripting at once to compile a few C files. I think hiding this in any other build system would have been definitely harder. Check this mess:
dnl Define somedir_c_make.
[$1]_c_make=`printf '%s\n' "$[$1]_c" | sed -e "$gl_sed_escape_for_make_1" -e "$gl_sed_escape_for_make_2" | tr -d "$gl_tr_cr"`
dnl Use the substituted somedir variable, when possible, so that the user
dnl may adjust somedir a posteriori when there are no special characters.
if test "$[$1]_c_make" = '\"'"${gl_final_[$1]}"'\"'; then
[$1]_c_make='\"$([$1])\"'
fi
if test "x$gl_am_configmake" != "x"; then
gl_[$1]_config='sed \"r\n\" $gl_am_configmake | eval $gl_path_map | $gl_[$1]_prefix -d 2>/dev/null'
else
gl_[$1]_config=''
fi
this has to do with writing ‘better’ code, which has proved impossible over and over again
I can’t speak for C, as I don’t follow it that much, but for C++, this is just not fair. It has been proven repeatedly that it can be done better, and much better. Each iteration has made so many things simpler, more productive, and also safer. Now, there are two problems with what I just said:
If that last part is what you mean, fine. But the way that you phrased (and that I quoted) is just not right.
At this point it’s literally easier to slowly port to a better language than it is to try and ‘fix’ C/C++.
Surely not for everything. Of course I see great value if I can stop depending on OpenSSL, and move to a better library written in a better language. Seriously looking forward for the day when I see dynamic libraries written in Rust in my package manager. But I’d like to see what’s the plan for moving a large stack of C and C++ code, like a Linux distribution, to some “better language”. I work everyday on such a stack (e.g. KDE Neon in my case, but applicable to any other typical distro with KDE or GNOME), and deploy to customers on such a stack (on Linux embedded like Yocto). Will the D-Bus daemon be written in Rust? Perhaps. Systemd? Maybe. NetworkManager, Udisks, etc.? Who knows. All the plethora of C and C++ applications that we use everyday? Doubtful.
I’m not fully sure what the intent of the joke is, but note that yes, it’s true that a header typically just has the prototype. However, tons of more advanced libraries are “header-only”. Everything is in a single header originally, in development, or it’s a collection of headers (that optionally gets “amalgamated” as a single header). This is sometimes done intentionally to simplify integration of the library (“just copy this files to your repo, or add it as a submodule”), but sometimes it’s entirely necessary because the code is just template code that needs to be in a header.
C++ 20 adds modules, and the situation is a bit more involved, but I’m not confident enough of elaborating on this. :) Compile times are much better, but it’s something that the build system and the compilers needs to support.
Precisely, Gary Bernhardt has given a talk on ideology. I don’t think he’s precisely someone who thinks in absolutes. It’s just preaching that some stuff is (probably) used more than it should. I’ve seen way, way, way worse projects that over engineered things and made things slow and unmanageable, than the opposite. Of course, everyone has seen different things, and our perceptions are amplified and biased by that.
I’ve wanted to start a project in Rust, but for the ideas that I have (and the time that I have for a hobby project, as for work it’s rarely starting a new one, but continuing and existing one), Rust seemed a viable, but not ideal alternative to just doing it all in C++, for which I already have enough knowledge and very well proven libraries. I will look again soon, and I will keep looking because eventually something will surely click, it’s just that so far, the time has not been right.
Note that my point is not that it’s unusable for everyone. Just that it’s false that “some people just can’t seem to let [C or C++] go”, as the previous comment said. I can’t let go something that works well for something that doesn’t, given the projects that I have to work on.
It’s just time to move on from C/C++, but some people just can’t seem to let go.
The Rust community has 2 websites that I keep periodically checking: Are we game yet? and Are we GUI yet?. The answers on those sites are respectively (as of February 2024, when this comment is written) “Almost. We have the blocks, bring your own glue” and “The roots aren’t deep but the seeds are planted”. I’ve seen the progress in Bevy and Slint, but it’s still the same, those websites don’t change, and my situation WRT to making a Rust project for fun or work it’s the same.
I’ll be happy to start doing Rust projects whenever I get the chance (which will be when it’s a sufficient tool for my use cases). But I’m tired of smoke sellers.
The github project page is for developers, and Github already gives you tons of ways to make a user website. Don’t ask your users to visit github.com/group/project, make them visit group.github.io/project, like any sane person.
Same with Gitlab, BTW.
And if you don’t like the full static site, use the wiki, or guide your users in the first paragraphs of the README so they find the user information if they must.
The very first moment that I had to use JSON as a configuration format, and I was desperate to find a way to make a long string into a JSON field. JSON is great for many things, but it’s not good at all for a configuration format where you need users to make it pretty, and need features like comments or multi-line strings (because you don’t want to fix a merge conflict in a 400 character-wide line).
Doesn’t YAML have a (seldom used) feature of a start and end of document marker? The “YAML frontmatter” that a few markdown documents have, uses this.
You entirely ignored this part.
You basically proved my point in doing so, BTW. You cannot do what you claimed with an LLM. And I’m not saying, and I never said before “ChatGPT” or “OpenAI”. I don’t understand why you think that I might be “defending these hypocritical companies”, when I literally said the opposite at the end.
You are entirely fooled by the output of ChatGPT and you are not arguing in good faith (or you are entirely unable to understand what I said).
Edit/addendum: And to stress out my point, given that the person to whom I’ve replied to showed the output of ChatGPT as if it were any kind of proof, this is what other LLMs say. This is 4o mini:
And this is Llama 3.1 70B:
Feel free to give them a shot in: https://duck.ai