serioussecurity

I'm very frustrated by this perspective.

I want to live in a world full of intellectually curious, creative people who build things. I think we can best do that by giving people access to education, health care, housing, food, and physical safety.

The positive externalities of an educated populace are massive.

I think the choice is between: (1) punish people because it's fair, and live an objectively worse life because everyone around us is miserable and dysfunctional (2) tune policy to support people, and live an objectively better life because everyone around us is healthy, productive and creative.

Immiserating debt destroys lives, and destroyed lives makes all of us worse off.

But there are competent folks in the field... ? Waymo's easily five years ahead of Tesla.

Previously: https://news.ycombinator.com/item?id=30182141

The parent post is joking that your human brain doesn't reset between games, unlike the other neural network being trained in this game.

This definitely violates the terms of service, if that's what you're asking.

Tesla is 5 years behind and has no plans to begin work on a number of the key challenges they're facing.

Microsoft didn't run automatic code analysis tools as part of their code review process until 2017? What on earth was wrong?

It's been a big deal for hundreds of years, most notably in economics and the military. Any reasonable company worth their salt was doing "data science" in the 1950s but calling it statistics or logistics or business intelligence to grab a phrase from yester decades.

What boat do you think you're catching besides a reliable middle class job?

Search handles probably an order of magnitude more changes per second than GMail+Docs combined

I've seen individual engineers make a mistake that caused 100 engineer's salary worth of damage.

That's quite often illegal, unfortunately.

You don't get to use retaliatory behavior to discourage lawsuits, in this form, especially as a service provider.

If they wanted to sue Epic back over frivolous patent violations, they're allowed, but if you're going to engage in retaliatory behavior you can't do it in a way that violates your own ToS.

Yes and no. They run on the same hardware and share a lot of the internal tools. Which is a very big lift.

this abjectly fails to deal with heteroscedasticity

Uber / Lyft

No one stopped Uber, no one has stopped Tesla (Tesla hasn't been able deal with regression testing on a lane finding system...)

If they know there's a high probability they'll be hit they'll adapt their behavior. The uncertainty of whether or not you're painted is significant.

Most fire is suppressing fire. If you could have the equivalent of suppressing fire without needing to waste bullets that'd be a huge change.

Most of my interesting papers have resulted from letting go of my "logical" brain and letting my emotions play with the problem until a line of attack opens, then I turn on the logic brain to get it done.

As I get older I understand the systems better but I get lazier. Give me 20 people and I will blow your mind, but on my own I get very little done. Part of why I manage people is I know what to do but can't actually do it myself (emotionally). Junior engineers are hungry and willing to burn themselves out working to learn what I know (like I did). I love love love junior engineers.

We don't even know what fraction of what we term consciousness or intelligence comes from the brain.

Antonio Damasio is a good read.

Mehhhh. BERT is mind blowing, Waymo has started fully driverless service.

Sweep a physical occluder across the lens, measure abatement?

Back testing however is possible. You could collect all recorded data from all driverless orgs, and require any new deploy be run over it. Plus a simulation.

For example, creating a standards body that collected this data and certified and release could be done by industry.

Tell me more.

Does that hold when they're actively working to harm you and yours?

Ipcc report

NLP systems don't "just fail".

https://www.npr.org/sections/thetwo-way/2017/01/10/508363607...

Epstein would use his connection to these institutions as cover with peers, and as a way to find clients. There's evidence this includes Marvin Minsky.

Do Not Use This With Your Own Money.

Literally millions of people have already trod this path, the low-hanging fruit in finance is already picked, and in many other contexts as well. The big mistake people new to the field make is to reason about machine learning as if it's a replacement for human intelligence, rather than just applied large scale statistics.

Do you know some other area extremely well, where you could apply these methods? Given how vague your question is, it sounds like you don't have much background or sense of direction. You will get taken to the cleaners if you try to compete with an established group; if you have some specialized context in which you can use machine learning, by all means, but the short answer is that you can't make money with machine learning. You may be able to find a problem or an application you can improve with machine learning, and make money that way. But it's very easy to lie to yourself in your evaluations with methodological errors, and then ship a product that does nothing remotely like what you intended / evaluated it for.