Artificial Intelligence and Corporate Social Responsibility | Hacker News Reader

Artificial Intelligence and Corporate Social Responsibility

medium.com

52 pointsbenbreen11 comments

I think this is going to be one of the biggest challenges in the field of AI going forward. How do we make sure that as an extremely powerful tool, it gets used responsibly. Also, acknowledging that like all tools, AI will be misused or applied in some truly heinous ways (examples across the internet if you're interested), how should the AI field respond and learn from this? This article is certainly evidence of people pushing in the right direction, even if some of the answers are a little woolly.

Ultimately, I think that the best answer is for AI Devs themselves to be extremely clued up on the potential social impacts of what they are working on and treat that with the same reverence as staying up to date on the latest literature. Only then do the kinds of "specific impacts" which the article brings up become clear enough to act upon and plan around. From there, one can work on safeguards, due diligence, or whatever else you want to call "making sure my tool doesn't hurt people".

You need to bridge the gap as much as possible between engineers sat in a plush office in SV, and the individual who is going to have their life affected by those engineers (and their AI imbued tools). Otherwise, it's just too easy to loose touch with what you & your tools do to the world and the people in it.

Some problems have no solution. Go a bit further: Should we even be relying on AI as a substitute for human interaction? Will AI ever be capable of replacing an empathetic human? Even if it can, should it? What if I could replace one of your family members with a perfect simulacrum that was indistinguishable from the real thing, except you knew it was a simulacrum. Would you accept it?

>What if I could replace one of your family members with a perfect simulacrum that was indistinguishable from the real thing, except you knew it was a simulacrum. Would you accept it?

You may not, but your kids generation that grew up with it always around would.

I think we are just fucked at this point. For a power of this scale, where potentially catastrophic results can arise from even a single source point, it really falls on the users/creators/management (ie us as a society). It can turn out great if we can all cooperate and play fair, but we have long ago blown that chance before any of us were even born. At this point the only working solution i can see is creating numerous AI simultaneously to balance each other out and regulate each other to the same effect that power is distributed amongst a group of people rather than be consolidated in a single individual. Creating a single AI or a handful of AI in the hands of a few is just too great a risk and power imbalance but in the beginning I fear it is what will happen as everyone is already in an arms race.

GE is currently unraveling because they hid their losses using creative accounting for 20+ years. Nobody will go to prison for this.

If you're expecting much in the way of social responsibility from a large corporation I think you're just being naive. People want to further their own individual careers and little else.

He makes a great point, which is that this is an issue that shouldn't just be on the minds of leadership, legal teams, and PR; increasingly decisions that impact people's lives in massive ways are made on the level of the engineers and data scientists who are actually making the implementations, and they need to be every bit as aware of the ethical implications of what they're doing as anyone else (if not moreso).

In this day and age a single line of code written by a junior engineer could drive something that millions of people interact with hundreds of times a day. This is going to have an impact on those people's lives.

So I'm way behind on developments in AI ethics. Are there actually any promising avenues for mitigating harmful impacts of AI? I'm possibly damaged by reading The Guardian, but it seems like the only options are a) making AI fairer and b) using humans instead. And option a doesn't really work because people can't agree on what's fair.

FWIW, humans aren't terribly fair either. You can create a shite review system, implemented by thousands of inconsistent humans, and end up with something crappy where it is far harder to point at any single decision and say 'this is crappy' because the system itself is masked by thousands of individual decisions. (Think 'customer service call centers' here.)

Or you can automate, and end up with a _transparently_ bad decision system backed by a tweak-able model, which - at least in this moment in history - is far easier to write outraged front-page guardian articles about.

The sorts of problems that we're deploying machine learning on often don't have single right answers. We've quickly gone from wonder that these things work at all to public outrage that they aren't providing perfect answers, but instead answers which often mirror our own societal shortcomings. The design choices made by the capitalists promote revenue over well-being, and the eigenvector of desire often points in the direction of the horrific.

The systems, as they stand today, are like any other tool, amplifying human ethical choices, for better or worse.

"... which - at least in this moment in history - is far easier to write outraged front-page guardian articles about."

which could be considered a positive feature!

Sort of reminds me of how you'll occasionally see articles highlighting the excessive permissions demanded by mobile apps, but they rarely mention that under the application models still used by PC operating systems (most notoriously on Windows pre-UWP, but also unsandboxed Mac OS applications, and on mainstream Linux distributions), most apps run with full user rights and therefore have more excessive permissions than mobile apps are even able to ask for.

The issue of AI and ML is only a subset of the many issues at hand, and until corporations are reeled in on the basic principles of ethics and social responsibility, there will be no progress on any subset of those issues such as AI.

a post on corporate responsibility that does not mention concepts like "income" "inequality" "wealth" or even "politics" requires a bit of rounding out before it's a complete discussion.

the author touches on a few points but tiptoes around the biggest issue: corporations need to pay in to society by design, and AI will be disastrous if they are not forced to uphold that standard. the same goes for AI's potential impact on privacy and free speech-- these are major major issues, but they are not the only major major issues.

economic inequality will continue to skyrocket if AI is remotely as efficiency-improving as everyone hopes; higher inequality tends to erode other democratic ideals like free speech, human rights, public political engagement/power, and the rejection of civil encoding of status.

artificial intelligence will make it far easier for corporations to avoid their fundamental economic responsibility to pay for the society in which they operate and directly enrich the lives of people within society in multiple dimensions. this is already happening. this is a problem that needs a solution.

ideally, AI is used to distribute resources for greater goods rather than entertaining individual gluttony for money.

we need to start agreeing on the axiom that the purpose of corporate activity is to increase the standard of living of society and increased profit margins should mean increased burden of charity rather than merely more zeroes on a billionaire's wealth.

that is the correct moral vision of corporate responsibility in the light of artificial intelligence.

getting into the mud with details on human impact assessments doesn't serve the fundamental truth that AI-- for all the benefits it will bring-- is an existential threat to the status quo that everyone knows will kill that status quo off over time and replace it with a new and unknown one.

Reply on news.ycombinator.com