Just sayin': We wrote a whole paper in late 2020 (Stochastic Parrots, 2021) pointing out that this head-long rush to ever larger language models without considering risks was a bad thing. But the risks and harms have never been about "too powerful AI".
Instead: They're about concentration of power in the hands of people, about reproducing systems of oppression, about damage to the information ecosystem, and about damage to the natural ecosystem (through profligate use of energy resources).
>>
This is the best piece I’ve read yet on the dangers of AI.
“Language is the operating system of human culture. From language emerges myth and law, gods and money, art and science, friendships and nations and computer code. A.I.’s new mastery of language means it can now hack and manipulate the operating system of civilization. By gaining mastery of language, A.I. is seizing the master key to civilization, from bank vaults to holy sepulchers.“
MAGA Logic:
-Abortion: Ban it
-Marijuana: Ban it
-Drag shows: Ban it
-Immigration: Ban it
-Dialogue on racism: Ban it
-Books by Black authors: Ban it
-Books by LGBTQ+ authors: Ban it
-Guns: BANS DON’T WORK YOU SNOWFLAKES
#CovenantSchool
One thing I haven't seen discussed as much are the knock-on effects of this.
Right now, Google can at best share some of the revenue from the third party sites generating the content. If link referral is replaced by chatbot summary, they'll be capturing virtually all of that revenue.
That changes the incentives for third-party content creation. Like it or not, there is a lot of good stuff on the internet that is there because creation and/or hosting costs are supported by ad sales.
I love the subtle pushback against racism and stereotyping that happens in Mark Rober videos. I wish more SF tech people were a little more like him. ♥️👍🏿
Plus, this Zipline drone delivery is pretty cool. I'm still sad that over the past 10 years, we put so much more time + effort + resources into trying to put monkey jpegs into people's 401ks, than into technology like this.
@johncarlosbaez @Lisanne Indeed, it isn’t obvious to me in this case. For many topologies & rectangle orientations, it’s intuitively clear that you could smoothly vary the ratio, and find the point where it a) forms a square, and b) all the pieces fit. Not here. In this case, my first impression was that the horizontal position of the split line going up from the pink subrectangle is a separate variable—one must make the two sides match, both top and bottom, and the pink block must have the correct ratio as well…
I’ll take a look at Lisanne’s proof. Thanks!
@johncarlosbaez Without checking the polynomial: it seems to me it would be a coincidence if it was real. There appears to be more than one constraint on the ratio.
Here is David Gerbet's new solution! He writes:
"After running the computation again, I was able to identify the possibly missing guy exactly: the 1055th ratio. The pictures of the partitions look all the same for smaller and larger ratios in the tables from Ian Henderson and the one I computed. You find it at page 106 in my table of partitions with 7 rectangles.
The ratio is the only real root of the polynomial
x⁷−3x⁶+9x⁵−10x⁴+12x³−7x²+2x−1
You can find a picture of a partition with this ratio in the files attached. Note that Ian Henderson has computed partitions with other ratios, but the same topology as this one, e.g. the 1022th ratio."
Here is the table that David is talking about:
This apparent new solution is made from rectangles with proportions of 0.7148589 to 1. But is it real? Should we welcome it to the elite club?
(2/2)
If you ask mathematicians who discovered a² +b² = c², they'll yell "Pythagoras!"
Then ask the mathematicians if Pythagoras ever traveled to another country to take a geometry class. They'll tell you yeah, he studied abroad in Egypt, and like, majored in triangles.
Egyptians had known about right triangles for thousands of years. Then ask the mathematicians if Babylonians knew about right triangles. They'll tell you, also yes.
On this one year anniversary of the war in Ukraine, let us call out and condemn the apologists and cowards who would give Putin free reign to invade and destroy other free nations and to slaughter their citizens.
They are on our some of our “news” networks and even in the halls of Congress. They are a threat to liberty, freedom and democracy for all of us.
This war must be won on the home front, too. We must continue military aid until the war is won.
Never back down! Slava Ukraini! 🇺🇦
OK, I've seen enough of these attempts at talking to language models that my opinion's gone from "spicy autocomplete" to "drunk old uncle internet mumbling repeatedly to himself again."
What bugs me is that when you recognize what the underlying idea is - "automated high-fidelity pattern recognition and repetition" - you can see that there are definitely hundreds of powerful applications of it that can elevate the state of humanity for the common good, and language models are likely the worst.
No notes.
RT @Brentus88@twitter.com
Hahaha. Democrat @RepKatiePorter@twitter.com in the chamber during #SpeakerVote
🐦🔗: https://twitter.com/Brentus88/status/1611579705899253761
Do you know how many times I've heard over the past few days how much the media is failing to follow the money in this story about the newly elected Rep. George Santos? This story makes it clear that's exactly what happened.
https://www.washingtonpost.com/media/2022/12/29/north-shore-leader-santos-scoop/
So let's follow the money, shall we? Starting with Mr. Santos' latest filing for The Devolder Organization, which is apparently responsible for his sudden unexplained wealth.
The organization lists its address as 336 N. Babcock St. Ste 104, and says another organization at the same address but Ste 101 is there called D&D International Investment Services Inc.
Both are affiliated with an individual named Devaughn Dames, who appears to be both a physician and a CFO somehow. He's also very passionate about spreading his knowledge about how to handle your money wisely. His linkedin page mentions both companies. He also seems to be an IT expert; previous domains include https://web.archive.org/web/20160518071721/http://www.manageditsolutions.net/about.html
One of Mr. Dames' email addresses (dev.dames at gmail) was used to register the vanity domain devaughndames[.]com. Firefox didn't like what that site tried to do when I visited it, but it loaded an empty page that had the heading "Financial Educational Services." But the pop-up takes you to a domain called myuwe.net, which is for an entity called United Wealth Education.
Now, UWE is full of stock photos and videos, and there isn't much there about who runs it. And after spending a few minutes on this site, you'll probably come away with the conclusion that the site is tied to some type of scam.
Well, if you look up the name of what Myuwe.com used to be called -- Financial Educational Services -- you'll see they were shut down JUST THIS YEAR by the US Federal Trade Commission for being a giant pyramid scheme that bilked people out of more than $213 million.
How's that for following the money? :) Wait, I'm not done yet.
WATCH: #GeorgeSantos has been exposed for the fraud that he is but still unanswered is the $705k question: Where did all this money come from?
BOOST if you agree Santos MUST be investigated and the #SANTOSAct should be passed.
@rob_fulwell @pluralistic I loved the article Doctorow references: a righteous rant by Cat Valente, about the pain of losing online communities, again and again, to corporations and despots and jackass billionaires:
"Stop Talking to Each Other and Start Buying Things: Three Decades of Survival in the Desert of Social Media”
https://catvalente.substack.com/p/stop-talking-to-each-other-and-start
📈 Doctorow's post deserves all the boosts it is getting and more! @pluralistic reminds me of the BBS/FIDONET days when I had so much hope for human connections across the network. I wished for a world where ideas are exchanged and evaluated based on their (then largely text-only) content.
Perhaps that hope can be rekindled with the rise of decentralized social media.🤞
#Mastodon #FediverseMigration
https://pluralistic.net/2022/12/23/semipermeable-membranes/#free-as-in-puppies
@climatenews This article has been on my mind for a few days now:
A group of ~370 scientists call for the abolition of research into solar geoengeneering to mitigate climate change. They have two main arguments: the moral hazard of mitigating carbon, and the unknown and unfairly-distributed side effects.
To me this is kind of nuts. You don’t worry about the moral hazard of the fire department when your house is burning. And the oil industry is not exactly open to moral persuasion, anyway.
As for unknown effects, that’s part of the point of the research. And if anything will have unfairly-distributed side effects, it’s climate change itself.
Engineer and dad. Interests include: ASIC design, network protocols, math, drums, California native plants, climate change (against), democracy (for).