whyte_mackay
Joined 8 years ago
Karma
11
Submissions
11
Member since
Nov 16
Recent submissions
Comment · 7 years ago
Comment · 8 years ago
I think game-theoretically it is best that the self-driving car always prioritizes its passengers. If it has to quickly hack a competitor's car to avoid a dangerous head-on collision, so be it: That market will sort itself out. Let self-driving cars play a tit-for-tat Prisoner's Dilemma strategy. Classic capitalism at work, because right now only rich people can afford a Humvee, and will come out on top when they crash with a second-hand Fiat Panda (even when it is their fault).
A bus with 40 passengers crashed because the driver wanted to avoid a raccoon crossing the road. I don't want my family on that bus, and will pay gladly for something better if I have to. The market will offer me that better option. If it has a customizable setting of "how many strangers to kill, before we kill your family first?" I will set it to "all of them." as I expect others to do so too.
Comment · 8 years ago
Adding fairness to your models usually incurs a cost which can be measured. You have to choose between max profit or equal opportunity, you can't have both (do you run over the investors or over the minorities?).
See here a visualization of trade-off's between global accuracy and fairness:
http://research.google.com/bigpicture/attacking-discriminati...
Comment · 8 years ago
I hear this often (that trolley problem is not relevant), but then I discovered that a lot of realistic ML fairness problems can be restated as a trolley problem.
You have a classifier for credit assignment (giving a loan, etc.). The classifier is 99% accurate on the entire population. The classifier is 55% accurate on a small minority. You can improve the minority accuracy to 90% at the cost of 0.3% decrease of general accuracy. What do you do?
For self-driving: Your accident rate is 0.0001% for the entire population. Your accident rate is 0.0003% for black pedestrians at night. You can allocate more compute/research/resources to equalize the accident rate of black pedestrians at the cost of increasing accident rate for the entire population to 0.00011% (or keeping it constant where you may have seen an improvement if you focused on the general population). What do you do?
Comment · 8 years ago
Yes. Cryptocurrency will eventually have to bite the bullet and become legit: KYC/AML, deanonymizing bad actors, defunding terrorists, hitting ICO's with laws, exchanges who cooperate with the IRS. It is getting there whether its users want to or not.
Comment · 8 years ago
Yup, that is a hairy decision to make if you advocate for privacy and anonymity, and against censorship: What to do with the bad actors who will make use of it?
From the Tor FAQ:
> Some advocates of anonymity explain that it's just a tradeoff — accepting the bad uses for the good ones — but there's more to it than that. Criminals and other bad people have the motivation to learn how to get good anonymity, and many have the motivation to pay well to achieve it. Being able to steal and reuse the identities of innocent victims (identity theft) makes it even easier. Normal people, on the other hand, don't have the time or money to spend figuring out how to get privacy online. This is the worst of all possible worlds.
> So yes, criminals can use Tor, but they already have better options, and it seems unlikely that taking Tor away from the world will stop them from doing their bad things. At the same time, Tor and other privacy measures can fight identity theft, physical crimes like stalking, and so on.
BTW: Those examples I gave are for Milo Yiannopoulos and WikiLeaks. I am not sure they fit into any category you mentioned.
> If when you say whiskey you mean the devil's brew, the poison scourge, the bloody monster, that defiles innocence, dethrones reason, destroys the home, creates misery and poverty, yea, literally takes the bread from the mouths of little children; if you mean the evil drink that topples the Christian man and woman from the pinnacle of righteous, gracious living into the bottomless pit of degradation, and despair, and shame and helplessness, and hopelessness, then certainly I am against it.
Comment · 8 years ago
> Its advocates don't understand finance, business, government, politics, computer science, math, sociology, monetary policy, macro or micro economics, or human psychology.
Weird, I must work with a bunch of dunces who turned millionaire by sheer luck.
Comment · 8 years ago
At least crypto is way more censorship resistant than oldschool money. That's why it is called "censorship resistant" not "censorship proof". Claim stands.
Make a Jewish joke and Paypal will not be your friend. Piss of governments and Visa and Mastercard will make it impossible to donate to your cause.
Comment · 8 years ago
Early adopters are silent, because they have nothing to prove and are too busy enjoying their wealth.
Holders are positive, partly because they have financial motive to do so.
Nocoiners are negative, partly because they are kicking themselves for not cashing in on one of the biggest opportunities of the last decade. They like to see crypto fail so they can say: See, I told you so! and feel better about missing out.
Then there is the perversion of the cryptopunk community, where it became all about greed and hype, and much less about the technology and vision. I wouldn't mind if the greedy folk dropped crypto and it became more about sticking it to the man (because I already largely cashed out, and kinda feel bad for the environment and taking last year's mom-pops Christmas money).
It pays to revisit the very first post on Bitcoin on HN and maybe realize that what HN thinks about crypto is not so relevant as it may seem:
https://news.ycombinator.com/item?id=599852
> Well this is an exceptionally cute idea, but there is absolutely no way that anyone is going to have any faith in this currency. - May 2009
Comment · 8 years ago
The one difference is that Microsoft has always been transparent and open in their support for the US military. They even advertise it: https://azure.microsoft.com/en-us/global-infrastructure/gove...
On the other hand, Google has been on the record saying they won't develop military AI, that an autonomous weapons race is dangerous and unwanted, and that they want to organize the world's information to keep it accessible to anyone.
Microsoft employees knew what they signed up for and have moral authority to make that decision. Google employees thought they signed up to "do no evil" and their efforts were used by a secret project that violates all but one of their safe AI guidelines (the one about delivering technical excellence).
I myself am a bit more agnostic/apathetic about creating technology that could be used for bad: I just want to create and focus on the best possible (gun turret, data mining, missile tech), that will deliver on its promise of what it said out to do with utmost accuracy and robustness. That's why I find it really important to work for companies and governments that I can trust to not abuse my technology once it is out of my hands, and are transparent about its usage. But I don't find any damage to my soul if the decision makers make an evil usage decision: That is fully on them. I am not going to handicap or refuse to work on something interesting, because their morals are out of whack.
Comment · 8 years ago
Interesting that I always used Andreessen's: Why software is eating the world? [1] as a temper to AI hype. You can replace "blockchain" with "distributed database" or "AI" with "software" and look at the result for if the use of AI or blockchain was just buzzword lingo or actually semantically relevant.
But now they have a chapter: "Giving Your Software AI Superpowers", which breaks this technique hard.
AI's history, to me, starts with Operational Research:
> Employing techniques from other mathematical sciences, such as mathematical modeling, statistical analysis, and mathematical optimization, operations research arrives at optimal or near-optimal solutions to complex decision-making problems. Because of its emphasis on human-technology interaction and because of its focus on practical applications, operations research has overlap with other disciplines, notably industrial engineering and operations management, and draws on psychology and organization science. Operations research is often concerned with determining the maximum (of profit, performance, or yield) or minimum (of loss, risk, or cost) of some real-world objective. Originating in military efforts before World War II, its techniques have grown to concern problems in a variety of industries.
Later authorities were just (in part) rebranding OR for Darpa/Iarpa grant money.
I like this executive summary though: It is good reading for managers and CEO's who may not be familiar with AI and its possibilities. For practitioners it is mostly fluff though and any intro course will give a better overview. A real playbook has yet to be written.
[1] https://www.wsj.com/articles/SB10001424053111903480904576512...
Maybe Eric should not have a private Instagram account which follows young models. Maybe Eric should not ask Google to delete results about his political donations.