Computer-played chess or go games are already more interesting than the human-played ones. If you get computer-written novels indistinguishable from the best human-written novels, or even better, would you read them? There will be 1 million of them generated every day, all better than Harry Potter, or by taste, War and Peace.
The better question is, would you read anything else? What happens when we can just dial up the story we want? What happens when these models combine the skillfulness of a human writer with the command of facts, theories, behaviors, and lore of computer records?
Imagine. You could feed it all of Tolkien's writings and tell it to give you an epic saga of novels set in the First Age of Middle Earth. You could tweak a parameter and instantly get a version of your favorite novel that proceeds with a different main character or a different choice and have it be indistinguishable from a version written by the original author and one hundred percent consistent with events and previous plots.
I'm not too ashamed to say that if memory-erasure technology existed I'd be sorely tempted to experience my favorite media for the first time over and other. Creative AI is the looming spectre of that danger, as it becomes more and more skilled at giving us the best, we will grow intolerant of anything less.
>> Imagine. You could feed it all of Tolkien's writings and tell it to give you an epic saga of novels set in the First Age of Middle Earth. You could tweak a parameter and instantly get a version of your favorite novel that proceeds with a different main character or a different choice and have it be indistinguishable from a version written by the original author and one hundred percent consistent with events and previous plots.
Arguably, you are describing about 80% of the Fantasy genre, only it's all written by humans.
Sorry if that sounded like a "shallow dismissal" as per the HN guidelines. I used to love F/SF and then I found I couldn't read it anymore because it all felt like the same few story elements were reused over and Over and OVER again. I get the same feeling of dread boredom when I browse the Fantasy isle at bookshops today. <evil power> is rising in the <cardinal direction>. <Hero> must <undertake heroic quest> to defeat <evil power>. It all got so formulaic you could write a Cluedo variant based on it: The Orc Captain with the Great Axe in the Dark Tower.
This, btw, was all before I (very belatedly, because I'm not a native English speark and so I read only what was translated to Greek) discovered Terry Pratchett, Jeff Noon and Iain M. Banks, who were a breath of fresh air (unfortunately rarely refreshed since then) and yeah, as a kid, I read a lot of F/SF. Explains my school grades, hah.
Achieving Strong AI makes no sense because we don't understand the human brain and how it works. Even if a non-biological AGI passes the hardest imaginable Turing test, there will be arguments against it because "it is still not like us".
This post may be immoral if it hopes for the next big thing. I wonder if it does more harm than good for the world. If someone is to indeed invent the next big thing after the Mobile Phone, for sure they don't need this kind of motivational post. For "smaller" ideas, yes, read ahead. But think it is immoral to encourage a wide audience to shoot for the stars, and try big ideas that have a 0% chance of succeeding and 100% chance of causing lost time and effort, and even financial, social and personal hardship.
Good points. The problem with asking the same coding question makes my brain lazy until I get an interesting response that wakes me up (~10% of the time).
I am toying with the idea of asking coding problems for which I do not know the answer (I did not try to solve it yet) so I can get engaged. This has obvious risks of course...
YouTube is really silly to ban PragerU. Although I disagree with maybe 70% of PragerU's content (I listen to their podcast), I find it very informative of the conservative POV. It also exposes me to ideas that makes me do my own further research. I listen to them regularly; there is much worse content on other places in YouTube.
They haven't banned it; their channel is the first search result on YouTube for the keyword 'PragerU'.
They've flagged it as restricted content so that advertising isn't attached to it, organizations who specifically ask Google "Don't show me restricted content" (such as schools) can't reach it from the networks they control, and auto-recommendation engines are down-sampled on it. Essentially, PragerU is trying to argue that they should have the right to force YouTube to classify their content into the same bucket as "baby shark" videos.
Presumably schools don't want their students ogling T&A during study hall. Does that mean students shouldn't see any politics? These are two entirely different things that should be placed in different categories.
I believe it falls under "Incendiary or Demeaning Content" (https://support.google.com/youtube/answer/7354993?hl=en). And I wouldn't be surprised if a lot of schools consider that inappropriate for e.g. middle or high school (though I certainly haven't taken a poll).
I haven't watched anything from Prager in at least a decade, but I'd be surprised if it was anything more "incendiary or demeaning" than "Trump is a great president!" The point stands, that regardless of how many categories they list, the switch that turns off porn shouldn't also turn off political speech. That is a flaw in Youtube, that probably wouldn't exist if they didn't have a monopoly.
"Shouldn't" in what sense? Are they leaving money or users on the table by failing to increase the granularity of the filter?
The core market for the feature is schools, and that group is generally okay with the filter being conservative and low-granularity (read: little thought required for configuration)
Money: These videos are playing less than they would have without being restricted, therefore less associated advertising is playing. It's probably not much money on Alphabet's scale.
Users: Somewhere in this nation, there is a high school administrator who consents to students learning about politics. There might even be more than one.
As a private organization, if that's money they want to leave on the table and users they don't choose to serve, that's their right. As you've noted, it's not much at Alphabet's scale.
Another company could pick up the slack. Hell, PragerU could do it. Nothing stopping them from brokering their own ad deals to run prerolls on video they host with their own infrastructure. A few schools might allow access, and many will just black-hole prageru.com in their firewalls.
This feels like ground we've already covered ITT. Sure, the current interpretation of current laws allows Alphabet to do basically anything they want. We can still talk about what they should do, and better serving users and making more money are both criteria that can inform that discussion. I thought you had suggested these criteria above.
Indeed; I was speaking broadly, not in absolute categoricals, and my question should have read "Are they leaving large amounts of money or users on the table." My apologies for being unclear. Yes, every decision they make in both directions leaves nonzero money and users on the table.
At Google scale, one has to weigh the risk of false-positive and the risk of false-negative on a low-granularity feature like this, and the risk of error due to misconfiguration being blamed upon YouTube if it's turned into a high-granularity feature. Because the failure mode for YouTube if they false-negative something that is visible to students who should not have seen it (and generate a negative press cycle for themselves) is that schools choose to black-hole youtube.com instead of bothering with Restricted Mode at all. That's what was happening before they added the feature, and it's the reason Restricted Mode exists.
That's a pretty clever way to add immediate framing for anyone viewing the content: conservatism is - by definition, apparently - incendiary and demeaning. Convenient.
Mature subjects: Videos that cover specific details about events related to terrorism, war, crime and political conflicts that resulted in death or serious injury, even if no graphic imagery is shown.
. . . .
You can’t post videos on Terrorism, War, Crime, and Politics.
It's interesting that all USA politics is acknowledged to be "political conflicts that resulted in death or serious injury", but somehow I doubt that MSNBC will be held to this standard...
They appear to be. Only a handful of PragerU's videos are marked restricted. A cursory check on https://www.youtube.com/user/msnbcleanforward/videos with restricted mode on and off shows a handful of their videos are also marked restricted, for example:
I have no idea by what criteria, precisely, these videos or the PragerU videos were marked restricted. Neither does PragerU, for that matter; Google doesn't surface that level of detail on the Restricted algorithm to end-users. I've been making educated guesses at the detailed cause, but it's somewhat irrelevant; the answer is, really, "Google's algorithm deems this content should be restricted."
By that criteria, MSNBC and PragerU appear to be treated the same way.
(Edit: FWIW, I went ahead and clicked through one video and that micro-sample should probably be flagged for "politics;" pundit accusing the President of, at best, being ignorant of science, and at worst, misleading the American people intentionally. That's clearly inflammatory political content).
That's key to the current frustration with YouTube, outside of constitutional grounds: as a platform, it's become wildly unstable to depend on, as you simply are not given the ability to understand where it will or will not draw the line, and their application of their standards leaves such an ambiguous and inconsistent trail of evidence that it's harmful to everyone who tries to rely on it.
I know you keep linking this blog post as if it proves a point, but its argument is tenuous at best.
The crux of the author's case is that PragerU's videos fall into the category of "videos on Terrorism, War, Crime, and Politics."
This is a remarkably wide net to cast. Surely you can think of countless YouTube channels which were not subject to the same scrutiny that Prager was, yet obviously have videos on politics and war.
At the end of the day, Google disagrees with Prager's politics. I don't think it's much more complicated than that.
Under the same Restricted Content scheme, Google also flagged LGBTQ+ videos (https://www.theverge.com/2018/6/4/17424472/youtube-lgbt-demo...). I don't think anyone argues seriously that the company's politics (in general) are at odds with supporting LGBTQ+ causes and people.
It may be worthwhile to ask not what YouTube wants, but what the audience for these features wants. Are PragerU videos restricted because most schools think they're worthy of restriction?
Depression is common among highly successful and/or popular figures. I read that depression was common among astronauts after they completed their missions. Virtually nothing that they (can) do later in life compares with the scope of their past missions, and that leads to depression or worse.
These days, part of the astronaut training includes planning for their post-mission goals.
Also, this used to cause trouble with phone operators connecting phone calls between USA and England: "are you through?" could mean either "are you connected?" or "are you done with the conversation?"
Vultures can fly over the border but their nests and young offsprings can't. Maybe this small restriction got some influence over their flight patterns over time.
