barathr

10 ms·

61.
by barathr 3y ago
Open-source, research, and small companies would be exempt.
62.
by barathr 3y ago
(I'm co-author of this piece.) You're right, and that's the exact idea: it will make LLM-generated text much more expensive but still a couple of orders of magnitude cheaper than a professional writer.
63.
by barathr 3y ago
Maybe instead I can offer something slightly different -- some of accounts I recommend. (I'll include our account at the end of the list.) @ai6yr@m.ai6yr.org -- admin of an excellent weather, ham radio, emergency response, etc. server
64.
by barathr 3y ago
I've been pretty happy with Elk.zone on the web and Ivory on iOS. Both are sleek and usable, as much as any commercial app.
65.
by barathr 3y ago
That's not representative. You need to look at this link: https://mastodon.social/public
66.
by barathr 3y ago
I'd recommend weatherwest.com which he runs -- lots of informed folks and Daniel's blog posts.
67.
by barathr 3y ago
Not really -- it's the same as selling any service into the US. Yes people cheat on, say, sales tax, just like Amazon did in the early years, but eventually once big enough companies end up having to adhere to the policy.
68.
by barathr 3y ago
The policy would exempt all except big companies from the fees. So if you set up your own, you don't pay. And the effect of the revenue threshold creating an advantage for small businesses is commonplace in policy across the board -- S
69.
by barathr 3y ago
To your first question, it would incentivize training of models on one's own data exclusively -- companies could train something like Copilot on their own code, for instance. To your second question, there's no way to have an inte
70.
by barathr 3y ago
The dividend isn't paid to generated content but for generated content -- so generating content (using say ChatGPT) means you're paying into the AI Dividend fund not receiving money from it.
71.
by barathr 3y ago
Yeah, the idea is that it's much more expensive than current OpenAI pricing but much less expensive than what even a low-end marketing copy writer would charge per word. Its side effect would be to push such tools towards more valuable
72.
by barathr 3y ago
Rather than there being lawsuit after lawsuit of this sort, we wrote an op-ed this morning that says there should be a simple, compulsory licensing fee that AI companies pay to the public -- something we called the AI Dividend: https:&#x2F
73.
by barathr 3y ago
Today, yes, but the point here is that this is not really a fair use of fair use, and would be compulsory licensing.
74.
by barathr 3y ago
It's a good question. Right now AI companies are using the data under "fair use" and they may get away with it so Reddit or whomever can't claim it either. (Getty Images and the like are going after generative AI not for
75.
by barathr 3y ago
FWIW, I co-authored this piece. Happy to discuss.
76.
AI Could Pay Dividends to Americans, Literally(politico.com)
1 points by barathr 3y ago|9 comments
77.
by barathr 3y ago
If you like this sort of thing, Pollan explored this before Harari in much more depth in The Botany of Desire.
78.
by barathr 3y ago
I wonder what metrics can be used to detect the emergence of "model collapse" as described by Ross Anderson's group in this paper: https://www.lightbluetouchpaper.org/2023/06/06/will-gpt-mode...
79.
by barathr 3y ago
Some of you may have had the experience of catching a show at Bottom of the Hill and reliving the NOFX song Scavenger Type: Gigin alone at the bottom of the hill / Our protagonist named Bill / Sets his sights on an anchor steam p
80.
by barathr 3y ago
It's a paper we wrote for a CS workshop -- that means it's necessarily a short paper, one that pulls together various threads to make a new argument.
81.
by barathr 3y ago
The Netherlands certainly does have profitable indoor farms, but the sort that the new startups are pushing are largely the uneconomic type. The two key factors: soil and sun. If you have to replace either, you're increasing your costs
82.
by barathr 3y ago
A combination of fruit walls and traditional passive greenhouses can zone push far better than vertical farms: https://www.lowtechmagazine.com/2015/12/fruit-walls-urban-fa... https://www.lowtechmagazine
83.
by barathr 3y ago
We wrote up an analysis of this sort of thing from an energy perspective -- bottom line: there's no energy-efficient way for hydroponic setups to produce anything other than leafy green food crops. The energy balance is just all wrong.
84.
by barathr 3y ago
Of course it's a guess. Nobody knows what will happen.
85.
by barathr 3y ago
I wrote up a quick analysis last night here: https://ioc.exchange/@invisv/110387246627189198 Here are the takeaways: * It'd be easy for Apple and Google to prevent downloads of TikTok for most users, which would a
86.
by barathr 3y ago
Neither is nuclear. And solar PV is vastly easier to scale up. The peak rate of fission power plant nameplate capacity construction was ~1985 and at that peak rate it would take something like 60 years to build what is needed globally.
87.
by barathr 3y ago
There is a really unfortunate assumption about nuclear baked into the engineering mindset -- and this piece is a perfect example of an engineering mindset. All complex technologies are really best understood as socio-technical systems. They
88.
Bubbles and Evilbrags(ioc.exchange)
1 points by barathr 3y ago|0 comments
89.
by barathr 3y ago
My limited understanding of the problem about git as the basis for user data storage is that it then would expose the complexity of git rebasing and the rest via federation. Git is great with a centralized remote (like I believe the origina
90.
by barathr 3y ago
I'm curious how this will play out for BlueSky with some fundamental human readability vs. decentralization tradeoffs such as the ICN naming tradeoff ( http://www.icsi.berkeley.edu/pubs/networking/ICSI_namingin

More ›