Are they? Where?

For all the hype I haven't seen any obvious abuses of AI. I've seen better speech recognition and a few other useful things. I've seen stuff that's wrong but ultimately kind of trivial like auto-generated porn with celebrity faces, but that's the worst stuff so far.

I haven't seen clear, unambiguous cases of abuse beyond that. I've seen a lot of allegations that AI is being abused e.g. "Russian bots" but on investigation these stories usually evaporate.

If anything I've been kind of disappointed by AI so far. Amazing demo videos abound, but I'd guess 90% of the impact of AI in my life has been Google improving their already quite good services. Better translations, better search results etc. All very welcome but not really life changing.

Funny you say that. I was recently searching for a e-scooter lighter than 10Kg and all Google could find was the max allowed weight of the person riding the thing (around 100-130kg). Not to mention that it didn't understand how to make a conjunction and show me light e-scooters with suspension. It's just matching keywords without understanding anything about the relations between them.

I am disappointed at the current quality of Google search, especially in shopping related queries where money is to be made. Instead of stuffing my pages with irrelevant 'personally' targeted ads and tracking my every move they should make an effort in that particular moment when I actually want to buy something and give me a good suggestion.

That's the power of these things. Would you know you're reading a social media comment generated by something akin to this? No, at best it would be ambiguous.

There is no way for an everyday person to tell how much of their life is impacted by ML at this point.

I honestly wouldn't care much if I was reading stuff written by bots, as long as I didn't waste time talking back ;)

It doesn't matter how many times people say the market works. There are plenty of self-appointed policy experts but no external force (I.e. regulation) helping people to live up to what they're saying. I think AI policy is pretty well-catered for labour at this point, but enforcement is not.

Sure, use it to decrease your operating costs. But it doesn't get to be secret sauce.

It's not the model that makes the bias. It's the data. Garbage in, garbage out.

The tricky situation is that's the magic sauce to most ML algos (as really GPT2 demonstrates). I don't care about what model they use (CNN, transformer, whatever) or their architecture (layers and how things are connected). What I'm concerned about is the implicit bias in the data. This is the hardest things to figure out and so the issue is how do you build trust of that data? Handing out that data makes you lose your competitive edge in a marketplace. Not only that, in many cases that data contains sensitive information that people wouldn't want public. So it's hard to say what to do. 3rd party auditors? But how do you prevent them from becoming corrupt and complicit like the credit agencies.

I think it sounds like there's easy answers, but honestly this seems really difficult to me.

I can't see access to data ever being accepted, for the reasons you mentioned.

However, publishing the model in a way that third parties can fuzz it seems to allow for discovery of the worst bads.

For example let's say big company trains uber translation network. Releasing the trained model means anyone can now use that model.

Maybe we could have laws about this like patent laws but there's still difficulty because is fine tuning the model making it substantially different? How do you deal with countries like China that don't respect patents (especially when we're talking about technology worth hundreds of millions or billions of dollars)?

I'm not saying we should give up. I'm saying that there isn't a simple answer here. We shouldn't expect one either! But to get a good answer we need to discuss and figure out the nuance of the situation. Because on one side we can't trust these companies too much. It's too easy to make mistakes. On the other side we can't just bankrupt them because we'll lose a huge standing in the world economy. So where's the middle ground? That's what I'm after.

Furthermore, it's a huge field, of which NLP is a significant, but tiny, subset.

The vast majority of models being used in the real world don't generalize thusly, because the data sets and processes they're linked to are bespoke.

The middle ground is the black-box model (perhaps with technical safeguards against decompiling-equivalent, or hosted as a service). It provides the ability to statistically prove bias, while protecting the majority of privileged or private information.

What I'm saying is there's a huge downside to releasing the full model compared to holding it tight. Many companies don't patent things for similar reasons.