jtraffic

Something I'll keep my eye on, for sure. In the meantime:

It feels like you've reinvented much by writing stuff from scratch. spaCy is fast, has tons of features, commonly updated, free, trained on the Common Crawl corpus. Why not just use that? I'm only curious, not critical.

I wonder how this article would be different had it been written after Medium introduced its new funding model. For one thing, a bunch of those publications, like The Awl are no longer there.

That's fair. But I guess I think of it as a somewhat continuous transition from human to machine manufacturing. Yes, currently manufacturing uses lots of humans, but that is changing, and the pace of change has not, AFAIK, reached any plateau.

What I think matters, is where we are going, not where we are. You give China as evidence of the human dominated state of manufacturing. I give it as evidence of the shift toward robot dominated manufacturing: https://www.technologyreview.com/s/601215/china-is-building-...

> Plus, modern factories rely more on automation than ever, so even if production comes back, it might be done by robots.

I feel this is the most important sentence in the article. The important question is not about individual products for end-users, but about intermediate means of manufacturing automation.

I really don't know which countries are most advanced in this area, but it makes the question of human labor less important. Or, at least it moves it up the skills ladder, from "making a glove" to "designing a robot".

Yeah. When I looked initially, I didn't go far enough into checkout.

The article says $230, but https://www.ridecabin.com/ says $115. Am I missing something? This makes quite a difference.

Philosophically, I wonder if it's better to just acknowledge the differences between AI and humans, and let them play to their strengths. It seems common to think that we need to constrain AI in certain ways to be more like humans, but the constraints are always artificial. APM is one constraint, but what about working memory? What about multithreading? We already allow the AI access to computational resources humans don't have. Why draw that line at APM, exactly?

Off topic: I posted this exact article four days ago: https://news.ycombinator.com/item?id=14813807

In the past, when I post exact duplicates, HN redirects me and automatically upvotes the original instead. I wonder why this doesn't always happen. (I'm not bothered, just curious.)

Double off topic: It's very interesting to see how much difference timing makes. My original had a single upvote, and this hit the front page.

I agree with others on the thread that "maintenance or innovation" is a false dichotomy. In particular, I think we need innovative maintenance. For e.g., this article about new ways to fix potholes: https://www.economist.com/news/science-and-technology/217003...

If I hadn't recently read Benoit Mandelbrot's book "The Misbehavior of Markets," I might dismiss this as empty cynicism. But I think there is some truth to it.

Macro and financial economists have traditionally been the worst offenders.

No, none. I suppose they want researchers to be motivated merely by the ability to contribute knowledge to the ML community?

Here is an Economist article that explains it much more detail: https://www.economist.com/news/briefing/21724368-recent-adva...

Or if you want the relevant portion quoted, see this article: (halfway down, below the graphene motorcycle helmet): https://hackernoon.com/an-innovation-taxonomy-d1ed0751b92b

A workable solution for kids would really be something. It's much harder to get anything done with bored kids than with a bored adult.

Maybe the bored adults can take care of the kids while the rest shop. ;)

I could ask the reverse: "why isn't China more successful?"

It's such a complicated issue that you could argue almost anything. I think any discussion must acknowledge Hong Kong, however, as an important data point.

My opinion is that the Communist party, even despite its "GDP obsession" [1], has gotten in the way more than anything.

As an (admittedly, very rough) estimate, if you take Hong Kong's GDP per person and multiply it by China's population, you get $58 trillion.

[1] https://www.economist.com/news/china/21689628-chinas-obsessi...

When a photographer takes or edits a picture, she doesn't need to predict or simulate her own reaction. There is no model or training necessary, because the real outcome is so easily accessible. However, she is only one person, and perhaps can't proxy well for a larger group.

The model has the reverse situation, of course: it cannot perfectly guess the emotional response for any one person, but it has access to a larger assortment of data.

In addition, in different contexts it may be easier/cheaper to place a machine vs. a human in a certain locale to get a picture.

If my theorizing makes any sense, it suggests that this technology would be useful in contexts where: the locale is hard to reach and the topic is likely to evoke a wide variety of emotional responses.

> they embody two alternative models of capitalism, and the one that wins out will shape the future of the economy.

Even in such a vague form, this is a touch sensationalist. The impact of 'the one that wins' may be negligible. There also may be no winner.

I see hindsight bias lurking. Now we see it as a "waste," but I dare you to make and publicly post predictions about a bunch of startups, and prove you are more accurate than VCs.

I agree. I got hung up on this: "Wolfram|Alpha uses natural language processing technology, part of the AI family, to provide students with an academic shortcut"

WA uses NLP for some things, but not for solving equations.

The New Yorker is well-known for this, and its tradition for using obscure words is long-running. It goes at least as far back as 1969, when "Business Adventures" was written by John Brooks, a long-time New Yorker author. I had to read that book with a dictionary, and used it every couple of pages. I still remember learning the word 'eleemosynary' from it.

Try Superforecasting by Philip Tetlock.

https://www.amazon.com/Superforecasting-Science-Prediction-P...

> This assumes the poor and middle class don't see a decrease in wages

I disagree. I think it only assumes that relative wages fall less than the relative incremental value of the new tech increases.

> Wages are a function of supply and demand

Yeah, but my point is that wages in isolation don't tell you much. The difficult-to-measure value accessible with those wages matters too.

> A good example is the effect globalization has had on the Midwest[1]

Irrelevant. Globalization need not move in lockstep with new technology.

I don't want to be reductionist, and perhaps my original comment came across as overconfident. My core assertion is that "it's not that simple," which decreases the value of conclusions based on oversimplified analysis. I think the article is oversimplified.

Abstract: "We analyze Airbnb's entry into the state of Texas, and quantify its impact on the Texas hotel industry over the subsequent decade. We estimate that in Austin, where Airbnb supply is highest, the causal impact on hotel revenue is in the 8-10% range; moreover, the impact is non-uniform, with lower-priced hotels and those hotels not catering to business travelers being the most affected. The impact manifests itself primarily through less aggressive hotel room pricing, an impact that benefits all consumers, not just participants in the sharing economy. The price response is especially pronounced during periods of peak demand, such as SXSW, and is due to a differentiating feature of peer-to-peer platforms  enabling instantaneous supply to scale to meet demand"

I have many problems with this analysis, but I'll name the biggest one: the author focuses on increases in value provided, but doesn't consider the cost (for consumers) to obtain that value.

If we receive the same value for lower cost, that's as good as an increase in productivity: I can earn the same money and afford a better living standard. Moreover, this kind of advance benefits the poor disproportionately (relative to the kind the author addresses).

Another case of the lab/field gap. So many things that work well in the lab are just terrible in the field.

In energy in particular, it seems like a good idea to increase the ratio of field to lab research.

I find that, ironically, professors are far too cavalier with posting other people's content online without permission.

The only real basis for my opinion is seeing .edu somewhere in the URL when these things are posted. So, I could be wrong, but I can remember two other recent examples: an HBS case, and a couple of chapters from Jared Dimond's Guns, Germs and Steel.

> have it printed there cheaper than it would cost to order from retropatents.com

Time is valuable, so is the design effort. If you could measure the value of those things, I'd say it's not really cheaper to do it your way (for many people).

I modified the title for accuracy. The original title misleads, slightly, IMO: "When Will AI Exceed Human Performance? Evidence from AI Experts." I swapped AI out and replaced it with ML.

The paper itself uses the acronym HLMI (high level machine intelligence). Quoting:

"High-level machine intelligence (HLMI) is achieved when unaided machines can accomplish every task better and more cheaply than human workers."

So a collection of machines could accomplish HLMI, without needing any single machine to do it alone.

>Microsoft's version has apparently been selling like hotcakes

I was going to ask you for more on this, but found an Ars Technica article with a bit more info: https://arstechnica.com/information-technology/2016/12/micro...

I don't believe missing the value of new technology can be relegated to "societies of the past."

Personal computers and the internet both had really slow starts, for two really big examples.

I'm sure there are technologies sitting around right now with vast potential that our society doesn't value.

I'd love to hear from anyone who is planning to buy one.

It feels as if collaboration tools are made for a type of collaboration that people don't really do. Consider this delightful gem of marketing copy:

"Jamboard breaks down barriers to interactive, visual collaboration across teams everywhere"

Whenever I've engaged in "interactive, visual collaboration," the visual content has been super context-specific. This meant there was usually a purpose-built tool for that kind of collaboration (e.g., wireframe mockups for websites, storyboards for videos).