pmichaud

37 ms·

331.
by pmichaud 11y ago
Characterizing it as a coin flip is unfair I think. If this system can promote individuals to attention out of millions of targets, telling us that this person out of all the millions has a 50/50 chance of being a terrorist, then that
332.
by pmichaud 11y ago
I'm deeply skeptical of this. It would be worth $100/hr to hand off highly complex task like planning a wedding (from their copy), but it's impossible--there is way too much interaction and iteration necessary on tasks like t
333.
by pmichaud 11y ago
I'm not modeling the future as car rentals, I'm modeling the future as driverless uber--on-demand transport from A-to-B. Fleets of cars in constant rotation. But you're right, rural areas are clearly harder to make that model
334.
by pmichaud 11y ago
You are not correct, there are people working on these problems who are experts in the relevant technical fields. Just not enough of them. But yes, I'm also concerned about the lack of safety-focused headliners at OpenAI, given the mes
335.
by pmichaud 11y ago
"AI" has a lot of attention, but AI Safety is underfunded, and understaffed.
336.
by pmichaud 11y ago
I would bet that owning your own car is going to be very marginal, and human-driven cars will eventually become illegal to drive on public roads.
337.
by pmichaud 11y ago
I think it's reasonable to be modeling the official recommendations as "the scientific and medical consensus," but if you looked a little into how the official guidelines are actually created--who creates them, and how the in
338.
by pmichaud 11y ago
Not putting the price on the site is an error that may break this idea. You use a system of some kind to calculate the price... just have the site automatically calculate it before the buyer has to do much. I'm in your target area, and
339.
by pmichaud 11y ago
I think the more nuanced version of the prevailing wisdom is "it seems like a bad idea to everyone else, because of a key set of facts that you know that very few other people know." Knowing that secret, whatever it is, probably r
340.
by pmichaud 11y ago
Not really. If you're moderately good at both then you'll get a similar score on both, and they will average out to about the same as both. So you'll still get the score you would've gotten before.
341.
by pmichaud 11y ago
This all makes a very compelling case for "using firearms in an optimal way is outside the limits of human ability, so police should not use firearms." That's assuming your terminal value is not "police life matters more
342.
by pmichaud 11y ago
Critical mass is like golden handcuffs here, I think.
343.
by pmichaud 11y ago
I can see this being big, It makes me wonder about the very long term. Saying that it's a tool meant to be used by lawyers is fine, but it's going to be fairly obvious to any lawyer who thinks about it for a moment that that is ba
344.
by pmichaud 11y ago
Ok, I think I'm starting to understand why we're disagreeing. I'm not sure what your background is, but I'm noticing that I probably have significantly more detailed models of both intelligence and cooperation than you d
345.
by pmichaud 11y ago
Ok, let me try to respond to both. 1. Can we invent AI? Ok, so to be clear, my belief is that not only can we do it, it is actually inevitable that we will do it. Fact 1: We definitely have people all over the world working on this tech, an
346.
by pmichaud 11y ago
I'm baffled and terrified by the idea that reasonably intelligent people think this. I think that you haven't grokked the scope of the issue at all, and I'm not sure how to convince you. I'm guessing that you can't
347.
by pmichaud 11y ago
I'm going to just quote my response from somewhere else in this thread: > ...when you're dealing with an actual superintelligence, all bets are off. It will lie, trick, threaten, manipulate, millions of steps ahead with a branc
348.
by pmichaud 11y ago
I think you're confused about the point of the test. The point is that an AI will be clever. Like, unimaginably clever and manipulative. Under the limited circumstances of interested people who know they are talking to Eliezer maybe yo
349.
by pmichaud 11y ago
I think you're being unfairly dismissive. I imagine you know as well as I do that what you wrote is a strawman. I have thought about what I would do to convince someone under these circumstances. My approach would be roughly: 1. We agr
350.
by pmichaud 11y ago
Yeah, modern web development is actually really fucking hard mainly because you have to know so many different things. To build an application from scratch, and do it well, you have to have deep knowledge of, let's see... * At least on
351.
by pmichaud 11y ago
I think we'll come to a point where we have strong AI, but we don't need strong AI to do pretty much every job humans currently do. It won't be one type of robot replacing all of us, it'll be specialized equipment doing
352.
by pmichaud 11y ago
For sure. I view vocational style degrees like "Web Development" as a red flag. My thinking is that the program is probably a bullshit money grab, and the person who went there is naive enough to go for it.
353.
by pmichaud 11y ago
There's an important distinction here though. This isn't a discussion about disclosing salary requirements, it's a discussion about disclosing previous salary, presumably so the company can make you an offer only marginally b
354.
by pmichaud 12y ago
$7 is not steep at all. DO is a commodity VPS that requires technical skill to even conceptualize, nevermind set up correctly. This is an easy, nontechnical thing selling convenience. For someone who wants a pretty, simple site, $7/mo
355.
by pmichaud 12y ago
The premise of human being replaced is all well and good, but it seems like you're assuming the superintelligence will do a better job. A paperclip maximizer may be able to deconstruct us on a molecular level, but that's not a thi
356.
by pmichaud 12y ago
I think the reason it's not discussed is that if there is a bound, it's far outside of human-scale intelligence. Even from my limited human vantage point I can see virtually limitless ways to improve my own intelligence. Perceptio
357.
by pmichaud 12y ago
I'm confused, do you mean to say that strong AI is impossible, even in principle?
358.
by pmichaud 12y ago
There's not a chance in hell that this is true. I'll be shocked if you can produce any evidence whatsoever that any version of this story happened.
359.
by pmichaud 12y ago
My guideline is that when something is failing for most people, it's not the peoples' fault, it's the thing's fault. The information in DARE is, in fact, not sound, which is a big part of why it's not only ineffecti
360.
by pmichaud 12y ago
I remember this idea from a while back, and I still think it's not going to work. It's too cumbersome to have to access your email every time you want to log into something. I love the creativity of the solution, but I just don&#x

More ›