Exactly. Central Europe is one of the most densely populated regions on the planet outside of Asia. High population density makes fiber more economical, and low population density, the inverse. As other la have pointed out, India actually has very deep fiber penetration exactly for this reason. The Americas, by contrast, are largely devoid of people which makes the economics of any networking infrastructure harder
>they aren't obligated to open them up when they no longer feel it's worth supporting them.
Creating a legal obligation to release the weights of discontinued models doesn’t seem absurd. These models are built on existing publicly available information; a requirement that it be returned to the commons once it is no longer in commercial use hardly seems like a substantial regulatory burden.
Presumably, a “Stop Killing AI” movement, mirroring the Stop Killing Games movement, would require a provider that revokes access a previously available model to make it open weights at the time of death.
This is strong “criticism of the man in the arena” energy on my part, but I’m kind of disappointed the text just wipes across, rather than the ink sort of “emerging” from the page like in the movie, with the heavier parts of the font appearing first and the thinner lines appearing last.
Incidentally, have any of the major AI provider's solved this problem for voice chats yet? It feels like even something like a simple keyword like “stop” would make having a conversation with an LLM a much better experience than a chat interface on a phone.
> The B2B SaaS Consultant
patron saint: Matt Yglesias
AI will quietly boost total factor productivity by some single-digit percentage and you think that's actually a huge deal, wonkily speaking. You're bullish in a spreadsheet way, not a singularity way. You've explained the Jevons paradox to someone who didn't ask.
I wouldn’t put MattY as the patron saint for this category, as I think he expects it to be substantially more transformative than this. Maybe Derek Thompson would be a better pick?
If I had to guess, it’s an ownership issue. There’s an enterprise team, a consumer team, a cloud team, etc. so there’s no single vision and plan. Which could be okay, but then each of those teams likely has not been given the authority to develop their own branding or naming and just defaults to using whatever is available, pre-canned.
Yeah, the most obvious recent example of this is RealPage’s YieldStar product. It advised property managers on what they should set their rental rates to, and allegedly established a cartel in which RealPage’s customers coordinated in pricing their units.
YieldStar was technically an “AI” product, but I don’t really think the computational abilities were what enabled the collusion. RealPage’s employees (according to the DoJ[0]) would actively monitor whether companies were following their pricing recommendations and call up companies that defected. And the software itself used dark patterns to make it easier to simply follow the YieldStar pricing suggestions, rather than set a lower rental rate and be more competitive. The algorithmic pricing I think did allow people to launder their own judgement and simple “trust the process” in a way that in the past would have required knowing complicity with the cartel, but I don’t think it required substantial compute capacity.
(This isn’t a comment on the paper by the way, which I glanced at but did not have the background knowledge to fully comprehend)
I did wind up choosing GCP to be our primary enterprise AI provider and it has definitely been a challenge to try and explain the difference between Vertex AI Studio, Gemini Enterprise, and Gemini Enterprise Code Assist, to say nothing of trying to explain that these three products are not the same thing as the consumer version of Gemini that they can find when they google “Gemini AI”. The branding is godawful.
You didn’t answer the actually interesting question I posed, who should be the gatekeeper instead?
Yes, yes, journalists and scientists have bad incentives and the general public is dumb. You’re not exactly setting the world on fire with that observation. The problem is that there is no better alternative. Any conceivable gatekeeper to scientific knowledge will be no smarter than the research scientists producing the results and will certainly have problematic incentives of their own. And a gatekeeper will also lack the local knowledge that might determine whether the information might helpful or harmful to the potential reader.
>The legacy of bad science being picked up is why this is a bad idea, even you personally don’t think it’s an issue the risk reward isn’t about just you
Who do you believe should be the gatekeeper here? Why can’t the scientist and the news outlets be trusted to make the decision about whether to publish or not themselves? Why can’t the general public be trusted to evaluate the quality of the news outlets they read?
Understandable reaction. That being said, thousands of people already pay for the privilege of inviting an actual human into their home every week to clean. For those people, that doesn’t seem likely to be a hurdle.
Personally, I’d probably be willing to stomach a teleoperator but what I would not be comfortable with is the company retaining images, video, and other telemetry from my condo on their servers for who knows how long.
While healthcare spending isn’t included in some economic measures like wages (which has contributed to the distorted productivity-pay gap discourse), labor share as discussed in this article is actually calculated using total compensation, the “total of payments to labor to produce output, including wages, benefits, and other monetary or nonmonetary payments,” which includes employer contributions to medical care not just wages and salaries.[0] They do discuss payroll share later on though, which doesn’t include non-wage compensation.