RT @lilianweng: Self-supervised learning opens up a huge opportunity for better utilizing unlabelled data while learning in a supervised l…

@gp_ifconfig What I meant is that one should study that kind of science from textbooks and books from world-class scientists, no matter how intimidating.

The science I needed in secondary school was neuroscience and cognitive psychology, not physics and chemistry. As important as the latter are, they don’t help you navigate and understand the world of people you live in and what is happening to you.

I genuinely believe that the best thing you can do for yourself in life is reading big, hard, complex science book about the brain and human behaviour. They will help you understand why you and others do certain things and give way less (or more) importance to those actions.

The number of ads about #cybersecurity in the #London tube continues to increase. I’m curious to know if there is a smiliar trend in #NewYork, and other financial capitals around the world. https://t.co/W2K2WoxjJp

Negligence or malice, I think it’s time to delete the app from my phone. I thought about it for a while, and this is a good enough reason > iOS Facebook app ‘secretly using camera’ while scrolling feed; likely a bug https://t.co/0mU4x0KAZt

If the results are not exaggerated, imagine the enormous potential in malicious applications > Lenovo pilots VR as an alternative to general anesthesia for kids https://t.co/ShHOeCVeXJ

True that it’s very limited, but some of PS technologies, like the new #AI-powered content-aware fill, are light-years ahead of competitors and well worth the wait. I tried: incomparable. > @Adobe deals with ‘painful’ early reviews of Photoshop for iPad https://t.co/rYUQDqUKXZ

I tried this new model and it’s important to clarify that while it’s exceptionally good (for an #AI) at articulating a position that has been previously defined by humans somewhere on the internet, result are poor/useless if you ask the AI to articulate an unpopular/new position https://t.co/aiUDyF7Sj0

“We’ve seen no strong evidence of misuse so far. While we’ve seen some discussion around GPT-2’s potential to augment high-volume/low-yield operations like spam and phishing, we haven’t seen evidence of writing code, documentation, or instances of misuse.” https://t.co/D8A47OJPLo