Twelve paragraphs in, and you still haven’t made your argument?
Are the AI companies at odds with copyright laws when they train new models? I think, yes.
Ah, because it’s a bullshit opinion piece being presented as fact, that makes sense.
You overestimate the worth of your inconvenience.
MMORPGs are an easy example, where people form recognizable identities and communities in game. An extension of this would be Second Life, and somewhat more recently, VRChat.
The generalized approach in industry is to use API calls, and create classes to structure the data you receive as JSON or XML. At that point, it is entirely up to you how to format and display the data from your classes. Take a look at some of the Lemmy client code like Mlem, Memmy, or Voyager as examples. Though they have gotten more complicated, they all follow this client-server model for front end development.
However, due to recent shenanigans around API and RSS by companies, mostly those looking to prevent AI companies from using their data for free, the alternative, much worse method is to take the HTML output from a standard web request, and try to reverse engineer the page information into a class structure. This sucks, breaks frequently, and requires you to code around ads and other junk on pages in order to get at the content.
That ”sparkling water” is still carbonic acid.
Just drink water.
That’s a gauss gun, not a railgun. Still cool, though.
This seems like the best possible use case, presuming both the actor’s estate and the new voice actor will get paid. You have the benefit of a human actor driving the cadence and emotion, then overlay with the transformation so unknowing players aren’t taken out of immersion by a sudden switch in a character’s tone and demeanor.
This is what they are banking on.
What this guy doesn’t mention, though, is that his uncle works in the cafeteria at NASA.
This.
At some point, you need to be able to quantify the risk to your business before you can do this.
For instance, if your business earns $10 per transaction, and you perform 100 transactions per second, the difference between five and six nines (313 seconds vs 31 seconds) is $282,000; nowhere near enough to justify the added investment.
Edit: Important to note that for the first example, these are already enormously huge numbers. Such a business, assuming no holidays or weekends, would be grossing $31.5 billion per year, in the same ballpark as Oracle and Coca Cola.
So when we say the company is losing 282,000, this is a tiny, tiny fraction of revenue. Even 99.5%, which is almost two days of downtime, would “only” be a loss of 0.5% of all revenue for the year. Sure, this is $157M, but even that would probably not cover the cost of a six nines infrastructure (that said, they could save up to $120M per year by achieving 99.9%, which would be worth exploring).