4 ms·

We need to focus on the AI harms that already exist

13 points by benwerd 3y ago

cs702 3y ago
...and also other things that are far more urgent, such as : * ending current armed conflicts by finding realistic peaceful solutions to them, * reducing hunger caused by lack of food in many parts of the world, * reducing the rate of preventable diseases due to absence of basic healthcare, * reducing ignorance due to lack of access to basic decent education, * increasing human rights in areas that utterly lack them, * developing stable political institutions in places that lack them, * engineering solutions to prevent or delay catastrophic climate change, * etc. There are so many unsolved problems that are so much more urgent right now. But none of these problems are as sexy nor as important to vested interests as AI.
- DangitBobby 3y ago
  The common thread is "who is affected by these problems". The people affected by the problems you've listed aren't writing opinion pieces about them.
- TaylorAlexander 3y ago
  Just wanted to add * dramatically reducing prison populations and focusing on more humane solutions to crime such as affordable housing, training, and jobs programs for the poor
- add-sub-mul-div 3y ago
  Waiting for AI to become an urgent problem before dealing with it sounds like similar thinking to what led those other problems becoming urgent. --- Kif: Sir, can I speak with you? Zapp Brannigan: No! Kif: But, sir, it's an emergency. Zapp Brannigan: Come back when it's a catastrophe!
graemep 3y ago
> One problem with minimizing existing AI harms by saying hypothetical existential harms are more important is that it shifts the flow of valuable resources and legislative attention. From the point of view of the people promoting the legislation, that is feature, not a bug.
bwanab 3y ago
The author, Joy Buolamwini, had an article in this Sunday's Boston Globe Ideas section that explores the same topic. It was extremely well written and enjoyable to read, but unfortunately reached exactly the wrong conclusions. In pretty much every case she cited, the problem was with the humans involved and only tangentially about the AI. In one case, a man killed himself after spending many hours talking to an AI that evidently suggested that. If a man is spending hours talking to an AI about personal issues, there's already a really big problem. Admittedly, the issue that she raises about an AI not being able to identify her face because she has dark skin is disturbing, but she knows as well as I do that this is because the model hasn't been trained on a broad enough training set - not that there's something inherently wrong with the model necessarily. It's a sensationalist approach that unfortunately seems to be the norm in the AI dangers universe.

13 points by benwerd 3y ago

...and also other things that are far more urgent, such as : * ending current armed conflicts by finding realistic peaceful solutions to them, * reducing hunger caused by lack of food in many parts of the world, * reducing the rate of preventable diseases due to absence of basic healthcare, * reducing ignorance due to lack of access to basic decent education, * increasing human rights in areas that utterly lack them, * developing stable political institutions in places that lack them, * engineering solutions to prevent or delay catastrophic climate change, * etc. There are so many unsolved problems that are so much more urgent right now. But none of these problems are as sexy nor as important to vested interests as AI.

DangitBobby 3y ago
The common thread is "who is affected by these problems". The people affected by the problems you've listed aren't writing opinion pieces about them.
TaylorAlexander 3y ago
Just wanted to add * dramatically reducing prison populations and focusing on more humane solutions to crime such as affordable housing, training, and jobs programs for the poor
add-sub-mul-div 3y ago
Waiting for AI to become an urgent problem before dealing with it sounds like similar thinking to what led those other problems becoming urgent. --- Kif: Sir, can I speak with you? Zapp Brannigan: No! Kif: But, sir, it's an emergency. Zapp Brannigan: Come back when it's a catastrophe!

graemep 3y ago

> One problem with minimizing existing AI harms by saying hypothetical existential harms are more important is that it shifts the flow of valuable resources and legislative attention. From the point of view of the people promoting the legislation, that is feature, not a bug.

bwanab 3y ago

The author, Joy Buolamwini, had an article in this Sunday's Boston Globe Ideas section that explores the same topic. It was extremely well written and enjoyable to read, but unfortunately reached exactly the wrong conclusions. In pretty much every case she cited, the problem was with the humans involved and only tangentially about the AI. In one case, a man killed himself after spending many hours talking to an AI that evidently suggested that. If a man is spending hours talking to an AI about personal issues, there's already a really big problem. Admittedly, the issue that she raises about an AI not being able to identify her face because she has dark skin is disturbing, but she knows as well as I do that this is because the model hasn't been trained on a broad enough training set - not that there's something inherently wrong with the model necessarily. It's a sensationalist approach that unfortunately seems to be the norm in the AI dangers universe.