Then there are the gray areas like training statistical models that are not ML/ANNs, but are things like Hidden Markov Models and the Vertibi algorithm. While not using ANNs, they work in a similar way in that they are statistical models trained on a large corpus of data, and those models are then used to predict various things like the part of speech, end of a sentence, or the pitch of a phoneme.
The former has dubious quality but can be useful in certain contexts where human translation is unavailable.
The latter has anecdotally already failed spectacularly in at least two cases:
https://news.ycombinator.com/item?id=6156238
https://news.ycombinator.com/item?id=29739235
For example, what are positive applications of NLP you can think about? Most applications i hear about would be to replace human customer support with chatbots (very negative) or for intelligence gathering purposes (very negative). There's probably a lot i don't know about.
Also NLP can be used to protect vulnerable groups from online abuse. It protects message boards from porn. It helps people find related content (which they enjoy). It automates categorizing or grouping content/people. Some companies abuse these practices for money.
Don't blame technology for the bad actions of companies. Their intentions would have corrupted whatever technology is available.
Well yes that's a very reasonable expectation. In most fields, when you sell something that doesn't work, it's called fraud. But in fields where it actually matter for safety, regulations are most stringent, such as in mechanics or medicine. If a car had 1% chance of blowing up killing everyone inside, would it be legal to sell it? Why is it ok for OpenAI to sell a chatbot that has 1% chance to convince you of utter bullshit when you ask it a serious question?
I believe education and communication to be critical to society's wellbeing, so i would hold companies in this field to greater standards.
> Also NLP can be used to protect vulnerable groups from online abuse.
Automated filters do not understand sarcasm, quotation, criticism... Sure that avoids the scunthorpe problem, but for example, auto-moderation bots trying to prevent racist abuse often censor black people calling one another the n word, which is completely ok. Nothing can replace moderation.
> It helps people find related content (which they enjoy).
Maybe, but i doubt it. I've yet to see advertisement/recommendations for content i actually enjoy, or to meet people who have such experiences. I've only ever heard that from advertisers, but maybe you have better experience than i do?
> Don't blame technology for the bad actions of companies. Their intentions would have corrupted whatever technology is available.
Good point! But technology that makes it easier and cheaper to be evil should probably not be encouraged and widely available. At least here in France, i cannot go out and buy a gun or dynamite. Also, bad results of AI are not necessarily a result of malice, but rather a complete misapprehension of the technology which is just completely stupid math and not actual intelligence.
If you're curious I recommend watching James Mickens' USENIX talk on what he calls technological manifest destiny, where he expands on these topics: https://youtube.com/watch?v=ajGX7odA87k
As for GDPR, how would that be a regulatory overreach? If anything, that regulation is so loose and has so many loopholes that it's almost entirely useless to protect people's data, but it's still a win in a good direction. It's also based on privacy laws that have existed since the 70s in various forms in several member states, which have proved their worth so it's not exactly something entirely new their dropped out of their hat.
> we really redoubled our efforts to make sure that they were not inadvertently imposing expectations (...) onto open source developers who are often hobbyists, nonprofits or students
That's about as much info as we get.
Personal/controversial opinion: neural network approach to technology should be entirely banned as it has very bad results and very high computational cost. I don't care that the LLM model hallucinating answers to my serious questions is open source (whatever that means in the context of neural nets, that's another debate), the harm is done anyway. I still do appreciate that "open-source" neural nets could continue to exist as a research field, but that any regulator considers allowing Google/Microsoft/OpenAI to clearly lie to the public using their artificial stupidity is way beyond me.