throwawaywego

9 ms·

31.
by throwawaywego 7y ago
She can't keep getting away with this.
32.
by throwawaywego 7y ago
And for context: https://i.imgur.com/NpvySvG.jpg (Pewdiepie) (It's edgy humor/let kids play on their skateboard)
33.
by throwawaywego 7y ago
Not a credible threat, according to police. Also, only after she was targeted by the media (a common pattern for Youtube to take action). > “Susan, I’ve known your address since last summer. I’ve got a Luger and a mitochondrial disease.
34.
by throwawaywego 7y ago
Yup. And Google Analytics should (and I believe already has) be treated similarly to a tracking beacon.
35.
by throwawaywego 7y ago
Worse. You can make the ML model spit out the SSNs it was trained on. That's a problem when you can't manually curate billions of documents. If you didn't look, you wouldn't even know they were there.
36.
by throwawaywego 7y ago
Can you yourself reliably detect if something is a list of SSN's or some other type of identifier? I think you can. Now when we have neural nets capable of distinguishing 100s of different dog breeds, we still have to trip on the most
37.
by throwawaywego 7y ago
In most modern browsers you can't capture this when find-in-page box has focus. Only if you manually select text after searching can you capture it (or, like the other reply stated: capture the scroll distance). But you could easily di
38.
by throwawaywego 7y ago
That's a very D&D Rulebook type interpretation :). Not that rulings should be ambiguous, but usually some common sense can be applied (and is expected to be reasonably applied). The Facebook like button is a web tracker, disguised
39.
by throwawaywego 7y ago
What I think happened: Someone contacted Capital One by email to responsibly disclose to them that there were SSN's and other data on a Gist. That person found them with a simple crawler or search. Then Cap 1 thought: If some rando can
40.
by throwawaywego 7y ago
https://www.microsoft.com/en-us/research/publication/crypton... (2016) > We demonstrate CryptoNets on the MNIST optical character recognition tasks. CryptoNets achieve 99% accuracy and can make more than 5
41.
by throwawaywego 7y ago
Cloud Computing. Use the compute from a big company, without that company possibly knowing what they are computing on. Machine Learning. Apply neural networks to encrypted data and avoid privacy issues.
42.
by throwawaywego 7y ago
1943 First mathematical neural network model 1958 Learning neural network classifies objects in spy plane photos 1965 Deep learning with multi-layer perceptrons 2010 ImageNet error rate 28% 2011 ImageNet error rate 2
43.
by throwawaywego 7y ago
I think any AI researcher has a tale where an algorithm they wrote genuinely took them by surprise. Not due to wrong calculations, but by introducing randomness, heaps of data, and game bounderaries where the AI is free to fill in the blank
44.
by throwawaywego 7y ago
https://arxiv.org/abs/1502.04156 > Towards Biologically Plausible Deep Learning > Neuroscientists have long criticised deep learning algorithms as incompatible with current knowledge of neurobiology. We explore m
45.
by throwawaywego 7y ago
All sciences that collaborate with the field of AI: Cognitive Science, Neuroscience, Systems Theory, Decision Theory, Information Theory, Mathematics, Physics, Biology, ... Any AI curriculum worth its salt includes the many scientific and p
46.
by throwawaywego 7y ago
> The atomic bomb was based on science theory. Our study of (automated) intelligence is based on science too. > A computer ... will never be able to think by itself. Turing wrote an entire paper about this (Computing Machinery and Int
47.
by throwawaywego 7y ago
Please reread the article. The article itself admits that non-financial data is useful and used (where this is allowed). I agree 100%. You drew the wrong conclusion about something you don't know a lot about and doubled down. Good luck
48.
by throwawaywego 7y ago
The evidence is in the article. It tells of a US company using 10.000 data points in a jurisdiction where this is allowed. The article cites Turner, who a few years back wrote this: > The non-financial data was found predictive in all th
49.
by throwawaywego 7y ago
Other comments have pointed out the risk of redundant encodings /proxy variables. I like to point out that sources of unfairness are possible, even with 100% decorrelated non-protective variables. For instance, the data collection and
50.
by throwawaywego 7y ago
> The key here being that non-financial data isn't actually useful in predicting ability to repay loans. Sure it is. Even religion correlates with loan risk. (Could be a proxy for social status and people from your tribe helping you
51.
by throwawaywego 7y ago
There is multiple elements to your question. Two ways where these systems may give out more loans than is strictly profitable, and they are both investments: - Fairness. If you have a variable race and a zip code, you could account for disc
52.
by throwawaywego 7y ago
There is an ongoing lawsuit against Jones. The lawsuits contain unproven claims of harassment. There is no direct inciting, and the indirect inciting claims are half-baked. There is a class action lawsuit office looking for a pay day, so th
53.
by throwawaywego 7y ago
The article states positive impacts on science, but there are also negative impacts on science. For instance, the hype of AI has caused a brain-drain on related fields (such as cognitive science or applied mathematics). AI research itself s
54.
by throwawaywego 7y ago
Me too. But it is a mere suspicion. We know that: - GRU has been stepping up its attacks on the West (anti-doping, chemical weapons, MH17 investigation) - US, Britain, and Israel attacked Iran with Stuxnet - Iran retaliated with cyber attac
55.
by throwawaywego 7y ago
https://gizmodo.com/job-networking-site-linkedin-filled-with... > Want to know all the code names for America's massive intelligence gathering programs? Just browse through the "intelligence analysts" who
56.
by throwawaywego 7y ago
Got an invite from a fake account claiming to be a professor at my alma mater. The name only exists on LinkedIn and has currently a handful of connections. I reported the account, but it remains up as of today. I do not have a public accoun