"Oh no we're not going to release GPT-2 because its so advanced that it's a threat to humankind" meanwhile it was dumb as rocks.
Scaremongering purely for the sake of it.
The only remotely possible "safety" part I would acknowledge is that it should be balanced against biases if used in systems like loans, grants, etc.
"""This decision, as well as our discussion of it, is an experiment: while we are not sure that it is the right decision today, we believe that the AI community will eventually need to tackle the issue of publication norms in a thoughtful way in certain research areas. Other disciplines such as biotechnology and cybersecurity have long had active debates about responsible publication in cases with clear misuse potential, and we hope that our experiment will serve as a case study for more nuanced discussions of model and code release decisions in the AI community.
We are aware that some researchers have the technical capacity to reproduce and open source our results. We believe our release strategy limits the initial set of organizations who may choose to do this, and gives the AI community more time to have a discussion about the implications of such systems."""
- https://openai.com/index/better-language-models/
> The only remotely possible "safety" part I would acknowledge is that it should be balanced against biases if used in systems like loans, grants, etc.
That's a very mid-1990s view of algorithmic risk, given models like this are already being used for scams and propaganda.
> Due to our concerns about malicious applications of the technology, we are not releasing the trained model. As an experiment in responsible disclosure, we are instead releasing a much smaller model(opens in a new window) for researchers to experiment with, as well as a technical paper(opens in a new window).
If you note, that's pretty much verbatim to what I said. So no, people don't have defective memories, some people just selectively quote stuff :P
You should actually read the paper associated with it. It's largely a journey in "why would you think that" reading.
Because:
> some people just selectively quote stuff
And that's what I'm demonstrating with the bit I did quote, which substantially changes the frame of what you're saying.
Our written language doesn't allow us to put all the caveats and justifications into the same space, and therefore it is an error to ignore a later section of the same document that makes the what and why clear, along with caveating this as "an experiment" and "we know others can do this" and "we're not sure if we're right".
However, given the capacity for some of the more capable downloadable models to enable automation of fraud, I am not convinced OpenAI is incorrect here.
If OpenAI and Facebook both get sued out of existence due to their models being used for fraud and them being deemed liable for that fraud, the OpenAI models become unavailable, the Facebook models remain circulating forever
(Failing to stop and think in advance what might go wrong, seems to be why Facebook didn't invest much in moderating the content in Burma until much too late; they'd have also looked to many people like they were 'worrying about nothing' if they'd done that investing before the genocide gave everyone a concrete example of why that matters)
All of the stuff I saw OpenAI being concerned about in the link, has in fact happened with the bigger models (that said, this is a skim-reading, not an exhaustive analysis):
--
"""We chose a staged release process, releasing the smallest model in February, but withholding larger models due to concerns about the potential for misuse, such as generating fake news content, impersonating others in email, or au- tomating abusive social media content production [56]."""
(Basically all of that)
"""One key variable affecting the social impact of language models is the extent to which humans and machines can detect outputs. We found reasons for optimism as well as reasons to continue being vigilant about the misuse of language models going forward. Our thoughts on detection at this time are:
• Humans can be deceived by text generated by GPT-2 and other successful language models, and human detectability will likely become increasingly more difficult."""
(I've been accused, here, of being an LLM. Have you, yet?)
"""Security: There is a tradeoff between the number of partners and the likelihood of a model being prematurely released, accounting for hacks and leaks.
Fairness: The high cost of compute used in powerful models like GPT-2 raises concerns about accessibility and equity in future AI research [13]. Private model sharing should not excessively harm researchers with limited computing resources, and conflicts of interest related to model sharing should be avoided in commercial contexts."""
(One of the GPT-2 models got leaked, so did one of the LLaMA models; at the same time, there's loads of people who think it's big and clever to call OpenAI "ClosedAI" and even to throw hissy fits about the other models failing to provide the training set and thus revealing that they have no idea how expensive the models are to train).
An experiment to find something out isn't hyperbolic even when the result is "hahah no". A requirement for the concept of a test is more than one possible answer.
Paying attention to potential risks before you have had a chance to evaluate them, is exactly what people demand whenever a group fails to do so and finds out there was a risk by causing harm.
Or have you never noticed that? "Why didn't the government prevent this attack!" and "Why didn't Facebook realise their software was enabling a genocide!" etc.
Perhaps I was being overly generous by blaming this on memory rather than on reading worse than the very LLM being laughed at.
People trot this out every time this comes up, but this actually makes it even worse. This was only part of the reason, the other part was that they seemed to legitimately think there could be a real reason to withhold the model ("we are not sure"). In hindsight this looks silly, and I don't believe it improved the "discussion" in any way. If anything it seems to give ammunition to the people who say the concerns are overblown and self-serving, which I'm sure is not what OpenAI intended. So to me this is a failure on both counts, and this was foreseeable at the time.
"trying to be safe" means you sometimes don't do something, even if there's only a 10% chance something bad will happen
Why bother checking if there's a bullet in the chamber of a gun before handling it? It looks so foolish every time you check and don't find a bullet.
Bullets kill people when fired by firearms. I fail to see how LLMs do.
Is that really a danger? I can shut off a machine or VMs.
Should Google worry about Chinese state-backed attackers using attacking its systems to target dissidents or for corporate or military espionage? "Why, when they're using machines or VMs, and you can just shut those off?"
At a sophisticated-human level of capability, there are many established techniques to circumvent people trying to shut off your access to compute in general, or even to specific systems. It's certainly possible that AI will never reach a sophisticated-human level of capability at this task—it hasn't yet—but the fact that computers have off switches gives no information about the likelihood or proximity of reaching that threshold.