moxious

If the autonomous stuff works (HUGE if) it would completely change the core business. I can't see why they'd spin it off. Why sell your future unless you think it won't work out.

This article mainly addressed his politics. Crazy or not in the political world, he's a superb software engineer.

It's going to be a loooooooooooooooong goodbye. COBOL is still around for what it's worth. The dominant OS (Linux) is written in C. I'd assert that any language used to implement a widely deployed operating system has at least that operating system's lifetime + 20 years, assuming the OS isn't rewritten.

"Re-write Linux in Go" I think would be a massive but classic folly

I doubt that was the goal of series C as any investor in that round would have been throwing good money after bad. If you're the VC in that case you're probably better off investing in a growth operation that trying to salvage a previous bet

The severity of this persons claims are why arbitration was invented. If he's right, complaining on the internet isn't enough of a remedy. And if he's wrong, there ought to be some attempt to hear the other side

How is it that the business is pushing the costs off?

If you have a job, you're selling your labor. The commute is part of what you have to do to get your goods to market. Are you, as a purchaser of bananas, "pushing off the costs of transport" on to the seller? That seems silly considering the transporter prices that into the bananas, just like workers can price in longer commutes. Don't accept less money to travel further each day.

It is not true that whatever a human can do, a machine can be taught to do. The human must have insight into HOW they do it in order to teach it, or otherwise come up with some new algorithm. There are a large class of things humans do that they don't understand the mechanics behind, and for which there also aren't algorithms.

I'm not talking empathy or philosophy. How about just folding laundry. Not just one type, not in a controlled environment, but folding any laundry anywhere.

There's clear sampling bias. I read the article as a data irrelevant advertisement for the app

Would the CIA really do tha...oh.

The longer they hold it, the less likely you'll ever get it if things go bad.

If you live in a society where credulous people will accept stories like this, it's no wonder people wouldn't know or spend time with their neighbors. I can't help but see it as one grain of sand on a bigger beach.

Wanting to do something and doing it are just totally different. Ask any alcoholic.

So many layers are now built on top of the OS that it seems the most obvious way it will continue to grow is by gradually subsiding abstraction layers above that are so common, you'd never not want them.

This has already happened with things like sound drivers, TCP/IP stacks, firewalling, etc. but will continue upward

> while making one long argument from ignorance

Ignorance of what? You're saying his argument is lazy and ignorant, but you're not backing that up.

I think he is willfully refusing to "imagine potential problems" - because a core point of the article is that people are "imagining magic", which isn't really an argument, but more of a way of whipping people into an emotional frenzy (all jobs will be taken by robots, dystopian futures, etc)

Generalize does need to be more specific in order to make sense. In current ML work, generalizability means something else than what it typically used to mean in AI work.

One way to look at what generalizability is would be like this: problem domains typically come with sets of axioms. Generalizability is the ability for a solution approach to work across different domains that don't have 100% axiom overlap. The wider the difference between axiom sets for a domain, the more difficult / impressive the generalization.

Solving two problems within the same domain, necessarily sharing a 100% overlapping axiom set, is not generalization at all.

The reason the axioms supporting the domain matter is that they (in part) guide which heuristics work, and how they should be applied. And that's the sort of generalizability that is missing from current work: some solutions can make some pre-programmed choices about applying different heuristics depending on the problem set (driverless cars are doing this now). This is the "bag of tricks" approach. But they don't typically morph in how they are applied, or the end that they're trying to accomplish when they're applied.

#2 is definitely wrong. We aren't a collection of similar hacks; we are qualitatively different because we include the ability to gain new heuristics and to cross apply the ones we already have. No clever library of strung together hacks will have these 2 properties. In the same way, an ant hill is not just a bunch of ants, it has so many emergent properties that thinking of it in terms of its components is a mistake

We should however think more expansively about intelligence than just the sort humans have. The goal doesn't need to be straight mimicry of humans

As someone who has been into AI since the mid-90s, I continue to be deeply disappointed with the overall rate of progress, this includes with all of the machine-learning and stats based recasting of what's meant by AI.

The big fundamental problems are all the same, and we don't appear to have much progress on them. We have figured out very clever ways of teaching extremely useful new tricks to computers. I don't mean to denigrate the value of that work, it just resembles regular programming, but by different means.

Whatever your definition of intelligence, there's little to no generalizability of what's happening right now.

This is a good exercise too to point out all of the tiny little elements of car UX which you just expect to be right. Things like, placement of the sideview mirror in such a way as you can see the mirror, and not placing it behind a pillar that blocks the view.

When UX is right, sometimes it blends into the background and disappears, just lets you use the object. You don't really know what's right about car UX until you see it done badly, then all of those things you took for granted really pop out.

There's not going to be a next one, there's going to be 10.

20 years ago the web was a much smaller place, and a few billion people currently online weren't even aware that the place existed.

Now, the population has exploded and there will be lots of different sites, all good in different ways, to "nurture the next generation of nerds". In different languages, too.

Growth, evolution -- all good.

The Byzantine generals problem doesn't address malicious nodes, just failure prone ones. That's a pretty big difference

Blockchain seems like a brilliant solution when multiple anonymous untrustworthy parties need to get together and agree.

But to the point of this article, most situations aren't like that. You have identities, contracts, and trust at least as much as can be enforced with the legal system.

I have also seen a lot of over-application of blockchain to "standard" banking problems. I guess it's real value will depend on whether or not it's successful in "decentralizing all of the things". If that doesn't happen, then it may just be a very cumbersome way of doing something we already knew how to do.

I'm sure the paranoid prepped is there. But the nature of being a paranoid prepped is that he or she is holed up and not coming out to talk/share.

Now probably wouldn't be a good time to go wandering across other people's land

On the contrary it's a wonderful test because it establishes indistinguishability; namely if you pass it, the whole point is that a person can't tell the computer from the intelligent thing. Meaning that you can't really argue that the computer is different than the intelligent thing. Because how would you tell them apart?

Besides, the original claim was that the goalposts are moving. And even if you hate the Turing test, it's clear that the goalposts are not moving.

The unrestricted Turing test has been around since the 1950s as a test that hasn't changed.

No one is moving the goalposts, I think it's rather the opposite. Every ten years computers learn a new trick or two and people rush to claim that this time, it's intelligent.

What do you see as the basic design of modern AI?

The 30 year old discipline of symbolic AI doesn't have much to do with today's statistical AI.

At first I was going to agree with you, but I think this largely may be an artifact of where and how you started (and me too).

Text-based console editors I think are fundamentally unintuitive and make you learn a lot of different bizarre keystrokes that don't really map onto easy mnemonics or have much actual meaning, so I do kinda wonder if one set is better/worse than another.

NDAs aren't standard at all. They range from simple back of the napkin agreements to "we get your first born child".

Blindly signing subjects you to potentially nasty term, and obligated you not to talk about certain things. This in turn can be used as leverage by the other side if things go badly.

Really, it's just don't blindly sign any agreement. Unfortunately agreements are so ponderous (clickwrap) that most people are used to doing that

Why Is there a limit on the number of columns I can have, and if something bad happens from having too many, why isn't the limit much lower than 1600, which hopefully most of us can agree is already way beyond absurd?

Even memory is a tradeoff. Assuming excellent memory, even then people won't use 20 character passwords because they're too hard to type precisely. The ability to store/copy/paste passwords makes for stronger passwords because size and complexity just doesn't matter

Will it still be voluntary and consensual if the day comes where stores post signs: "Payment accepted in bitcoin only"?