Similar work done at Facebook: https://engineering.fb.com/developer-tools/predictive-test-s...
3,137 karma · joined November 12, 2012
It doesn't mean I believe everything that's done on our products is good for the world, but I do believe the good outweighs the bad by a wide margin. A big part of my job is to make sure that this margin increases.
Happy to answer more.
Still, unlike Tay, we purposely did not create a service for it and do not advise creating one. This is for research purpose only and more effort needs to be made on safety before it can me more broadly consumed.
Paper: https://arxiv.org/pdf/2004.13637.pdf
Open Source: https://parl.ai/projects/recipes/
Ask us anything, the Facebook team behind it is happy to answer questions here.
Our overestimation of the comprehensibility of the world may very well be some version of the Drunkard's search principle. We are much more likely to know about what's comprehensible than what's not.
Link: https://www.scientificamerican.com/article/no-one-can-explai...
"We often hear that AI systems must provide explanations and establish causal relationships, particularly for life-critical applications. Yes, that can be useful. Or at least reassuring.
But sometimes people have accurate models of a phenomenon without any intuitive explanation or causation that provides an accurate picture of the situation. In many cases of physical phenomena, "explanations" contain causal loops where A causes B and B causes A.
A good example is how a wing causes lift. The computational fluid dynamics model, based on Navier-Stokes equations, works just fine. But there is no completely-accurate intuitive "explanation" of why airplanes fly. Is it because of Bernoulli principle? Because a wing deflects the air downwards? Because the air above the wing want to keep going straight but by doing so creates a low-pressure region above the wing that forces the flow downwards sucks the wing upwards? All of the above, but none of the above by itself.
Now, if there ever was a life-critical physical phenomenon, it is lift production by an airliner wing. But we don't actually have a "causal" explanation for it, though we do have an accurate mathematical model and decades of experimental evidence.
You know what other life-critical phenomena we don't have good causal explanations for? The mechanism of action of many drugs (if not most of them). An example? How does lithium treat bipolar disorder? We do have considerable empirical evidence provided by extensive clinical studies.
This is not to say that causality is not an important area of research for AI. It is. But sometimes, requiring explanability is counterproductive."
Yes. Definitely.
> Do you think they have honored those policies in the past?
I believe they have though I have not been at Facebook very long so I don't have full visibility on this. But I don't think that's the most important question. The most important question is: were these policies strong enough? The answer is definitely no and that's why the company is pivoting: https://www.youtube.com/watch?v=UtxPdezclYw&feature=youtu.be
More info about the privacy settings: https://portal.facebook.com/privacy/
Disclosure: I work at FB, my team developed the underlying MaskR-CNN technology.
See https://news.ycombinator.com/item?id=19036507 or https://news.ycombinator.com/item?id=20768607
> Do you feel pressure from your peers on the outside about the moral compass of the company?
Pressure is too strong a word. I do get into discussions with friends and family on the topic.
> My impression is that a lot of engineers are opposed to how Facebook operates and that this reputation must make recruiting and retention difficult.
That's mostly not accurate. Remember that these engineers can freely express their opinion internally or question Zuck live every week.
I don't think we go back to old photos, I suspect it would not be practical. But I am not positive on this one.
In the case you describe, Facebook no longer creates a Face Template, that is some kind of ML model of your face that could be used to recognize it or identify that you are the same person appearing across new pictures.
But I can’t say Facebook no longer has a record of your face. It still has some pictures, uploaded by your friends, where you appear. It won’t know that it's you OR and if your friends upload two pictures where you appear, it won't know that the same person is in both unless they tag you manually (which they can only do if you are user and authorize them - but that's a different setting).