Hacker News Reader

Comment by guywithahat | Hacker News Reader

Sometimes I wonder if these "ethical" articles are only put up to stop new companies from competing.

Like "ok now that OpenAI Inc has a basic AI, lets be sure to close the door behind us so we don't have to deal with competing startups"

View on HN

I believe this is a form of astroturfing and it is hard not to see this stuff. Remember all those articles about the AI winter a few years ago? In reality, there was no winter. Lot's of smart people took advantage of the opportunity and got incredibly wealthy during that period of time and now we see some of the fruits of that labour. It's just the beginning too.

It's important, in my opinion, to always ask 'who benefits from me believing this?'

I'm not sure how widespread the sentiment was that there was going to be a near-term AI winter.

What you did have was increasing skepticism that things like fully-realized door-to-door autonomous driving were going happen in the short-run--but a lot of the skeptics didn't think that would necessarily translate into an AI winter.

I mean -

> who benefits from me believing this?

In the case of this article, presumably the exploited* laborers whose labor conditions might improve if the population to whom this article is addressed both became aware and started to care about the labor conditions of the people behind the systems they use. Trying to 4D chess ourselves into not having to actually address the issue at hand isn’t clever or insightful, it’s just abdication with numbers attached.

* I’m not arguing whether or not the behavior described in the article actually constitutes exploitation or is just the noble hand of the market more perfectly forming working conditions - that’s not my point. The point is if you DO read the article and you DO feel that behavior’s exploitative, scratching your head over who’s got an agenda to want this thing you find abhorrent changed, as opposed to trying to change the behavior you find abhorrent, is the kind of thing people think smart people are supposed to do, and tends to contribute to more things you find abhorrent happening in the world.

It can be both true and used for astroturfing.

>Remember all those articles about the AI winter a few years ago?

FTA:

>"MMC Ventures surveyed 2,830 AI startups in the EU and found that 40% of them didn’t use AI in a meaningful way."

While not the same as an AI winter, it can still be indicative of "peak hype". If accurate, a large portion of companies are claiming to have some nebulous AI advantage aren't really delivering on that promise. It's doesn't have to be astroturfing to be pointing out a hype machine. Particularly if that hype is predicated on a shaky ethical foundation.

It doesn't have to be, no, but it might be. It's worth reconsidering the things we believe every now and then.

That's cause people today don't just say hey look what I built. Everyone acts as if they are doing something world changing. And it's bullshit, and the backlash is natural and well deserved.

By that logic any revelations about bad labor conditions underlying first world products or services only serve to prevent competition from also exploiting those conditions.

If your ethical conundrum is between continuing to allow the exploitation of the post-colonial third world, and to stifle competition in first world markets, I think you need to retake an ethics class because that is not a conundrum.

These articles are written to make people feel guilty about their lives. The ultimate goal is economic equalization.

The goal is awareness. There is no such thing as free lunch, somebody has to do the dirty work for too little money.

Are you sure it's too little money? articles like this often report wages that sound small to a person living in western Europe or US, but when compared to people in the same economy, these salaries are often competitive.

And yes, somebody has to do the dirty work.

This is why I can never understand these articles. How much is $1.46/hr? Is it a little? A lot? Average? So I head over to Google and look up the answer[0]. But everything is in yearly wages so $1.46*2080 ~= $3037. Looks like it would be a lot in Sudan or Myanmar, an average in Egypt, and not a lot in Sri Lanka or Ukraine. But are these numbers even trustworthy? We had a post about this just the other day.

But we don't even know the countries that are being used! So we're back to the original question: is it a little? A lot? Average? Honestly, when I see people spout out raw numbers alarm bells go off in my head. Doesn't matter if it is wage, number of car crashes, or number of murders. We have so many fucking people on this planet that a small percentage of a large number is still a large number. There are definitely people using this in deceitful manners. There are also dumb people, but idiocracy can still be accidentally malicious. The reason it is easy to lie with statistics and data is because it is actually really difficult to compare numbers and data across differing conditions. In this case, it really does not make sense to compare the wages of an American to the wages of a Sri Lankan. The price of an apple in both places isn't equal.

Still, I don't know the answer to my question.

[0] https://www.worlddata.info/average-income.php

The fact that it's a lot in Egypt or Myanmar just means wages are low in those countries. It's easy to say "well that's just the way it is" while you're sat near the top of the pyramid.

You're simplifying things too much though. There are important questions still. For example, how much does an apple cost? Rent? These are not equal in these countries. But also, let's say that they paid $3/hr in Egypt you could end up flooding the market. $15/hr and it could ironically be disastrous. We actually have seen stuff like this before when Westerners move in to other countries. The problem is that the local markets can't compete. Then we have a whole neocolonialism situation.

If your goal is to actually help people in a country get out of poverty this is actually quite a difficult goal to achieve without taking over that country (implicitly or explicitly). We don't want to physically invade. We don't want to economically invade. But we also don't want them to sit in poverty so we have to do something right? So how do you do that and let them maintain their independence? They might not even want to work with us to begin with! Don't pretend that these are trivial problems and us westerners sitting at a computer just know what's best. That's how colonialism started in the first place.

Paying extraordinarily higher than average wages in a poor country has downsides - most likely they'll be stolen (since most charity is stolen) but it also annoys the government when eg local doctors stop their practices to work for you.

Often competitive only means other jobs also pay too little money

The amount of dirty work increases as the price of labor goes down, because companies put less effort into inventing labor-saving techniques.

The circular reasoning is entertaining, as the low priced labor is theoretically generating the AI to be the ultimate labor-saving technique. But you don't need to save "too cheap to meter" labor. What was the last thing that was "too cheap to meter?" Ah yes nuclear power. Look how that turned out.

The two problems with trying to save the world via AI are the cheapest self driving car is a passenger train driven by "too cheap to automate" human, and an economy based on AI will be too poverty stricken from wealth inequality to permit the AI to generate a profit thus AI is not needed thus no need to destroy everything by applying AI.

The more AI is deployed, the poorer an economy will become, and the cheaper labor will become, making it quite a race to see if the AI gets smarter faster than economic activity implodes.

The most likely outcome of AI boosterism will be something like the environmentalist movement, but anti-AI. Who will be "The Lorax" of anti-AI?

Nobody can make you feel anything.

If true information induces feelings in you, perhaps it's more constructive to reflect on the information rather than the motives of the reporter?

I did journalism in high school and I know quite well that how you present true information (framing) or distort true information (for example, by comparing African salaries to the US) is done intentionally to make the reader feel something- guilt, anger, a desire to vote for somebody specific.

Does experience in high-school journalism support the claim that professional journalism is all a conspiracy to manipulate the reader? I doubt it, and I edited my high-school newspaper for two years lol

Did I say conspiracy? No. I'm making a generalization. If you have sources for objective journalism, I'd love to see them. The closest I've seen was the Economist about 5-10 years ago.

I once delivered papers so obviously I’m qualified to explain how the Star Chamber picks tomorrow’s headlines.

Like the movie says: there is no conspiracy. Big brother is not watching you. It’s a headless blunder operating under the illusion of a master plan.

Topical example to back up the claim that newspapers try to make you feel stuff, Prince Harry, UK tabloid headlines. Same fundamental information in all cases:

The Sun, positive: https://mobile.twitter.com/rainey_knight/status/161239752558...

The Mirror, positive: https://www.mirror.co.uk/news/uk-news/prince-harry-confirms-...

The Sun, negative: https://www.thesun.co.uk/news/21085974/piers-morgan-harry-ta...

The Mirror, negative: https://www.mirror.co.uk/news/royals/army-fury-prince-harry-...

> Nobody can make you feel anything.

A simple phrasing like this does not reflect the reality of humanities evolution and resulting biology. We're social creatures, requiring that others make us feel things: https://www.apa.org/monitor/oct05/mirror

Unless you're a psychopath, perhaps: https://www.medicalnewstoday.com/articles/321839

> Nobody can make you feel anything

The size of the advertising industry suggests otherwise.

> Nobody can make you feel anything.

This in 2023, when people feel offended by anything, sounds pretty off...

Nothing specific to 2023. In 399 BC, Socrates went on trial for corrupting the minds of the youth of Athens, and for two impious acts: "failing to acknowledge the gods that the city acknowledges" and "introducing new deities".

It is for sure specific to 21. century. Socrates was exposing fake specialists and decided to die instead of flea.. Specialists are debunked daily and they keep their posts, no consequences

If all you take away from this is that "someone is trying to make me feel guilty", then maybe you need to re-assess your own personal sense of guilt, and stop projecting it onto other people.

I don't see why somebody would feel guiltier about this than wearing Nike shoes?

It'll make a cottage industry of "data made in the USA"

It continues to answer the question: can the west run without maintaining a dedicated underclass -- no.

Should we be paying even less to make sure that we can keep that going? If those people get richer, these systems that benefit us will stop working

Would be less need for these articles if major tech companies would stop acting like villains in a sci-fi movie.

Wouldn't see them "acting like villains in a sci-fi movie" if there weren't systematic economic incentives for them to act this way.

And yet, every one of these organizations consists of individuals making sets of individual decisions on an ongoing basis, who indeed face incentives as well* but also persist in making the dubious ethical decisions that the company enacts, and we can in fact judge them for this, because companies are made of people who don’t actually abdicate moral responsibility for their actions by logging into a corporate email system.

*And here’s where I digress to say, again, that, indeed if you are the single mother of a child with a heart condition wholly reliant on the expensive company health care not lapsing for even a minute lest your child not just die but literally explode, taking your whole family and half the neighborhood with them, I’m not judging you. If, on the other hand, you’re a single dude between 20 and 35 with a college degree, significant assets, no debt, and a sellable skillset, yes, you’re the one I’m judging.

did you check in which conditions and countries your shoes, t-shirts, pants, etc are manufactured?

Yes. Wherever possible, I try to be aware of the consequences of my actions, and adjust how I do things when I find they’re leading to outcomes I feel are not in alignment with my ethics.

Nobody cares about some possible fake words, from the internet dude. My point was more to show you that this "if major tech companies would stop acting like villains in a sci-fi movie." is wrong. Companies are companies, and aim for profit at all costs, regardless of their branch.

Cyberpunk*

To me, the use of cheap labor for training suggests the level of importance AI companies place on avoiding social/cultural biases in the model in two ways.

1. Companies standards would be much higher if classification was done by engineers paid engineering salaries. And engineers would be directly accountable for model biases instead of bias being excused on the basis of how sausage gets made.

2. Outsourcing classification explicitly means classification is not valuable as a core competency; not an area in which expertise is important; and not an activity deserving of the benefits of direct employment such as financial rewards correlated to business success (e.g. bonuses, raises, options, etc.).

You are imagining that someone who otherwise would start a competing company would just choose not to after reading this article? Or what?

I noticed "Ethics of X" is generally a great field for those who can't cut it in field X. See the "Ethics of AI crowd as an example", that seems to be mostly made of "tech adjacent" folks.

Probably the best example here [0].

[0] https://syncedreview.com/2020/06/30/yann-lecun-quits-twitter...

That's not what this article is about.

I wonder if they are an "inside job": doing such a terrible job of making an ethical case against AI convinces people it must be fine...

Reply on news.ycombinator.com