Sunday assorted links
1. From Kyjnghyun Cho, total sanity on AGI risk.
2. The Bible is now banned in some Utah middle and elementary schools, due to violence and vulgarity.
3. “In a new analysis based on the latest telescope data, University of Florida astronomers have discovered that a third of the planets around the most common stars in the galaxy could be in a goldilocks orbit close enough, and gentle enough, to hold onto liquid water – and possibly harbor life.” Link here.
4. “Viable offspring derived from single unfertilized mammalian oocytes.” Not endorsing, but interesting to see.
5. Louise Perry reviews Bryan Caplan on feminism.
7. Good Ross column on where U.S. politics is procedurally at right now (NYT).
How the Russian Revolution boosted Marx’s influence
Karl Marx’s high academic stature outside of economics diverges sharply from his peripheral influence within the discipline, particularly after nineteenth-century developments rendered the labor theory of value obsolete. We hypothesize that the 1917 Russian Revolution is responsible for elevating Marx into the academic mainstream. Using the synthetic control method, we construct a counterfactual for Marx’s citation patterns in Google Ngram data. This allows us to predict how often Marx would have been cited if the Russian Revolution had not happened. We find a significant treatment effect, meaning that Marx’s academic stature today owes a substantial debt to political happenstance.
That is from a new JPE paper by Philip W. Magness and Michael Makovi. Here are ungated versions of the paper.
Where the AI extinction warning goes wrong
There is so much to say about this one, in my view it has been counterproductive for all those worried about AI safety. Here is one excerpt from my latest Bloomberg column:
Sometimes publicity stunts backfire. A case in point may be the one-sentence warning issued this week by the Center for AI Safety: “Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.”
…The first problem is the word “extinction.” Whether or not you think the current trajectory of AI systems poses an extinction risk — and I do not — the more you use that term, the more likely the matter will fall under the purview of the national security establishment. And its priority is to defeat foreign adversaries. The bureaucrats who staff the more mundane regulatory agencies will be shoved aside.
US national security experts are properly skeptical about the idea of an international agreement to limit AI systems, as they doubt anyone would be monitoring and sanctioning China, Russia or other states (even the UAE has a potentially powerful system on the way). So the more people say that AI systems can be super-powerful, the more national-security advisers will insist that US technology must always be superior. I happen to agree about the need for US dominance — but realize that this is an argument for accelerating AI research, not slowing it down.
A second problem with the statement is that many of the signers are important players in AI developments. So a common-sense objection might go like this: If you’re so concerned, why don’t you just stop working on AI? There is a perfectly legitimate response — you want to stay involved because you fear that if you leave, someone less responsible will be put in charge — but I am under no illusions that this argument would carry the day. As they say in politics, if you are explaining, you are losing.
The geographic distribution of the signatories will also create problems. Many of the best-known signers are on the West Coast, especially California and Seattle. There is a cluster from Toronto and a few from the UK, but the US Midwest and South are hardly represented. If I were a chief of staff to a member of Congress or political lobbyist, I would be wondering: Where are the community bankers? Where are the owners of auto dealerships? Why are so few states and House districts represented on the list?
I do not myself see the AI safety movement as a left-wing political project. But if all you knew about it was this document, you might conclude that it is. In short, the petition may be doing more to signal the weakness and narrowness of the movement than its strength.
Then there is the brevity of the statement itself. Perhaps this is a bold move, and it will help stimulate debate and generate ideas. But an alternative view is that the group could not agree on anything more. There is no accompanying white paper or set of policy recommendations. I praise the signers’ humility, but not their political instincts.
Again, consider the public as well as the political perception. If some well-known and very smart players in a given area think the world might end but make no recommendations about what to do about it, might you decide just to ignore them altogether? (“Get back to me when you’ve figured it out!”) What if a group of scientists announced that a large asteroid was headed toward Earth. I suspect they would have some very specific recommendations, on such issues as how to deflect the asteroid and prepare defenses.
Do read the whole thing. You will note that my arguments do not require any particular view of AGI risk, one way or the other. I view this statement as a mistake from all points of view, except perhaps for the accelerationists.
Sam Bowman on French success
But Ben Southwood has convinced me that France is rich because it gets the big things basically right. Housing supply there is freer: the overall geographic extent of Paris’s metropolitan area roughly tripled between 1945 and today, whereas London’s has grown only a few percent. Infrastructure is better: 29 French cities have trams, versus 11 here (likely one reason its second-tier cities are much more productive than Britain’s). It has nearly 12,000km of motorways versus around 4,000km here – and French motorways tend to be smoother and better kept (and three quarters are tolled, making congestion much less of a problem). Childcare is cheaper: about half the price per month, in part because they require half the staff. Energy is more abundant, as shown above. Because it gets those big four things right, it can afford to get a lot of other things wrong.
Here is the full essay, mostly on Coasean democracy.
Saturday assorted links
1. Johnny Cash and Josef Stalin.
2. Noah Smith’s current take on libertarianism.
3. New Larry Summers talk on fiscal policy.
4. The business of giving private concerts to rich people (New Yorker).
5. Amitai Etzioni, RIP (NYT).
6. Japan patent fact of the day. Yikes.
7. Kaija Saariaho, RIP (NYT).
8. Legal bank stablecoins from Japan? And a summary of the new U.S. crypto bill.
Who gains from corporate tax cuts?
Goods producers increase their capital expenditure and employment in response to a cut in marginal corporate income tax rates or an increase in investment tax credits. In contrast, companies in the service sector mostly use any tax windfall to increase dividend payouts. We base our conclusions on a novel measure of U.S. firm-specific tax shocks that combines changes in statutory tax rates faced by each firm with narrative identified legislated U.S. federal tax changes between 1950 and 2006.
That is from a new NBER working paper by James Cloyne, Ezgi Kurt, and Paolo Surico.
Why I am not entirely bullish on brain-computer interface
I agree that miracles may well be possible for disabled individuals, but I am less certain about their more general applicability. Here is one excerpt from my latest Bloomberg column:
Another vision for this technology is that the owners of computers will want to “rent out” the powers of human brains, much the way companies rent out space today in the cloud. Software programs are not good at some skills, such as identifying unacceptable speech or images. In this scenario, the connected brains come largely from low-wage laborers, just as both social media companies and OpenAI have used low-wage labor in Kenya to grade the quality of output or to help make content decisions.
Those investments may be good for raising the wages of those people. Many observers may object, however, that a new and more insidious class distinction will have been created — between those who have to hook up to machines to make a living, and those who do not.
Might there be scenarios where higher-wage workers wish to be hooked up to the machine? Wouldn’t it be helpful for a spy or a corporate negotiator to receive computer intelligence in real time while making decisions? Would professional sports allow such brain-computer interfaces? They might be useful in telling a baseball player when to swing and when not to.
The more I ponder these options, the more skeptical I become about large-scale uses of brain-computer interface for the non-disabled. Artificial intelligence has been progressing at an amazing pace, and it doesn’t require any intrusion into our bodies, much less our brains. There are always earplugs and some future version of Google Glass.
The main advantage of the direct brain-computer interface seems to be speed. But extreme speed is important in only a limited class of circumstances, many of them competitions and zero-sum endeavors, such as sports and games.
Nonetheless I am glad to the FDA is allowing Neuralink’s human trials to proceed — I would gladly be proven wrong.
Friday assorted links
2. Russell Hogg and Scott Sumner podcast on Japanese movies.
3. Foundations and Frontiers, some new essays on new technologies, by Anna-Sofia Lesiv.
4. “Meet the one-person team behind Antarctica’s longest-running newspaper, the Antarctic Sun.”
5. Benjamin Wallace-Wells on libertarianism (New Yorker).
6. The excellent Dan Senor podcasts with me.
7. How to keep people chatting.
8. Pentagon now hunting for UFOs with sensors actively built for that purpose.
The international competition heats up
Chinese startup MiniMax, working on AI solutions similar to that of Microsoft-backed OpenAI’s ChatGPT, is close to completing a fundraising of more than $250 million that will value it at about $1.2 billion, people familiar with the matter said.
The deal comes amid a global AI buzz kicked off by ChatGPT that has spread to China, shoring up stocks in artificial intelligence firms and prompting a flurry of domestic companies, such as Alibaba (9988.HK), Huawei (HWT.UL), and Baidu (9888.HK), to announce rival products.
Here is the link. And here is a Chinese professor from Fudan, critical of all the money being poured into LLM research in China, comparing it to Mao’s Great Leap Forward.
You have to be very critical of all sources from China, no matter what they say. Still, in terms of expected value I know what is the correct bet here, namely that China is a current and very active competitor in this arena, even if they are behind America so far.
Then there is the open source model Falcon, which is receiving very good reviews from multiple sources, such as this:
Falcon is a new family of very high-quality (and fully open-source!) LLMs that just made it to the top of the leaderboards.
Here's the "small" 7B version running on my mac with Core ML at ~4.3 tokens per second 🤯 pic.twitter.com/B1y4tyGzXA
— Pedro Cuenca (@pcuenq) June 2, 2023
If you don’t already know, Falcon is from…the UAE. Get the picture?
My excellent Conversation with Seth Godin
Here is the audio, video, and transcript from a very good session. Here is part of the episode summary:
Seth joined Tyler to discuss why direct marketing works at all, the marketing success of Trader Joe’s vs Whole Foods, why you can’t reverse engineer Taylor Swift’s success, how Seth would fix baseball, the brilliant marketing in ChatGPT’s design, the most underrated American visual artist, the problem with online education, approaching public talks as a team process, what makes him a good cook, his updated advice for aspiring young authors, how growing up in Buffalo shaped him, what he’ll work on next, and more.
Here is one excerpt:
COWEN: If you were called in as a consultant to professional baseball, what would you tell them to do to keep the game alive?
GODIN: [laughs] I am so glad I never was a consultant.
What is baseball? In most of the world, no one wants to watch one minute of baseball. Why do we want to watch baseball? Why do the songs and the Cracker Jack and the sounds matter to some people and not to others? The answer is that professional sports in any country that are beloved, are beloved because they remind us of our parents. They remind us of a different time in our lives. They are comfortable but also challenging. They let us exchange status roles in a safe way without extraordinary division.
Baseball was that for a very long time, but then things changed. One of the things that changed is that football was built for television and baseball is not. By leaning into television, which completely terraformed American society for 40 years, football advanced in a lot of ways.
Baseball is in a jam because, on one hand, like Coke and New Coke, you need to remind people of the old days. On the other hand, people have too many choices now.
And another:
COWEN: What is the detail you have become most increasingly pessimistic about?
GODIN: I think that our ability to rationalize our lazy, convenient, selfish, immoral, bad behavior is unbounded, and people will find a reason to justify the thing that they used to do because that’s how we evolved. One would hope that in the face of a real challenge or actual useful data, people would say, “Oh, I was wrong. I just changed my mind.” It’s really hard to do that.
There was a piece in The Times just the other day about the bibs that long-distance runners wear at races. There is no reason left for them to wear bibs. It’s not a big issue. Everyone should say, “Oh, yeah, great, done.” But the bib defenders coming out of the woodwork, explaining, each in their own way, why we need bibs for people who are running in races — that’s just a microcosm of the human problem, which is, culture sticks around because it’s good at sticking around. But sometimes we need to change the culture, and we should wake up and say, “This is a good day to change the culture.”
COWEN: So, we’re all bib defenders in our own special ways.
GODIN: Correct! Well said. Bib Defenders. That’s the name of the next book. Love that.
COWEN: What is, for you, the bib?
GODIN: I think that I have probably held onto this 62-year-old’s perception of content and books and thoughtful output longer than the culture wants to embrace, the same way lots of artists have held onto the album as opposed to the single. But my goal isn’t to be more popular, and so I’m really comfortable with the repercussions of what I’ve held onto.
Recommended, interesting throughout. And here is Seth’s new book The Song of Significance: A New Manifesto for Teams.
It was just a simulation run…designed to create that problem
https://twitter.com/harris_edouard/status/1664397003986554880
Please don’t be taken in by the b.s.! The rapid and uncritical spread of this story is a good sign of the “motivated belief” operating in this arena. And if you don’t already know the context here, please don’t even bother to try to find out, you are better off not knowing. There may be more to this story yet — context is that which is scarce — but please don’t jump to any conclusions until the story is actually out and confirmed.
Funny how people accuse “the AI” of misinformation, right?
Addendum: Here is a further update, apparently confirming that the original account was in error.
The art of prompting is just at its beginning
If prompted correctly, even GPT 3.5 can achieve draws against Stockfish 8 (an older but very powerful Chess engine), empirically demonstrating how LLMs are reasoning engines, not just text generation engines. https://t.co/an9eY27Re9
— Siqi Chen (@blader) May 29, 2023
And here are some results for Minecraft. I would like to see confirmations, but these are credible sources and this is all quite important if true.
Is growing conference size a problem?
In practice, they [scientists] more so blamed the human organization problems — essentially administrative issues — that they saw all around them. The growing conference sizes made it much more difficult to keep up with adjacent fields and scientific meetings. Seminars began to cater to narrower and narrower sub-branches of work rather than broad ones.
These were the places that many researchers leveraged to actually keep up to date on new work and problems in their fields as well as others. But, as money began to funnel into their field in the post-War era, there were more and more researchers and logistical decisions had to be made on how to do things like run conferences and decide who sits in what seminars.
The following Richard Feynman excerpt — taken from a 1973 oral history interview, which was one of a series of interviews between Charles Weiner and Feynman — goes into why, in the early 1970s, Feynman felt physics conferences had begun to grow far less useful than they were during the initial interviews for the series — where Feynman had told positive stories about the state of conferences as recently as 1956…
The conference size hypothesis almost surely is not the main problem, yet this is a new and interesting set of claims. The discussion of conference size comes fairly late in this piece by Eric Gilliam, plus there is a discussion of poetry toward the very end. For the pointer I thank Henry Oliver.
Orwell Against Progress
Orwell was deeply suspicious of technology and not simply because of the dangers of totalitarianism as expounded in 1984. In The Road to Wigan Pier, Orwell argues that technology saps vigor and will. He quotes disparagingly, World Without Faith, a pro-progress book written by John Beever, a proto Steven Pinker in this respect.
It is so damn silly to cry out about the civilizing effects of work in the fields and farmyards as against that done in a big locomotive works or an automobile factory. Work is a nuisance. We work because we have to and all work is done to provide us with leisure and the means of spending that leisure as enjoyably as possible.
Orwell’s response?
…an exhibition of machine-worship in its most completely vulgar, ignorant, and half-baked form….How often have we not heard it, that glutinously uplifting stuff about ’the machines, our new race of slaves, which will set humanity free’, etc., etc., etc. To these people, apparently, the only danger of the machine is its possible use for destructive purposes; as, for instance, aero-planes are used in war. Barring wars and unforeseen disasters, the future is envisaged as an ever more rapid march of mechanical progress; machines to save work, machines to save thought, machines to save pain, hygiene, efficiency, organization, more hygiene, more efficiency, more organization, more machines–until finally you land up in the by now familiar Wellsian Utopia, aptly caricatured by Huxley in Brave New World, the paradise of little fat men.
What’s Orwell’s problem with progress? He is a traditionalist. Orwell thinks that men need struggle, pain and opposition to be truly great.
…in a world from which physical danger had been banished–and obviously mechanical progress tends to eliminate danger–would physical courage be likely to survive? Could it survive? And why should physical strength survive in a world where there was never the need for physical labour? As for such qualities as loyalty, generosity, etc., in a world where nothing went wrong, they would be not only irrelevant but probably unimaginable. The truth is that many of the qualities we admire in human beings can only function in opposition to some kind of disaster, pain, or difficulty; but the tendency of mechanical progress is to eliminate disaster, pain, and difficulty.
..The tendency of mechanical progress is to make your environment safe and soft; and yet you are striving to keep yourself brave and hard.
I will give Orwell his due, he got this right:
Presumably, for instance, the inhabitants of Utopia would create artificial dangers in order to exercise their courage, and do dumb-bell exercises to harden muscles which they would never be obliged to use.
Orwell’s distaste for technology and love of the manly virtues of sacrifice and endurance to pain naturally push him towards zero-sum thinking. Wealth from machines is for softies but wealth from conquest, at least that makes you brave and hard! (See my earlier post, Orwell’s Falsified Prediction on Empire). Orwell didn’t favor conquest but it’s part of his pessimism that he sees the attraction.
Another of Orwell’s tragic dilemmas is that he doesn’t like progress but he does favor socialism and thus finds it unfortunate that socialism is perceived as being favorable to progress:
…the unfortunate thing is that Socialism, as usually presented, is bound up with the idea of mechanical progress…The kind of person who most readily accepts Socialism is also the kind of person who views mechanical progress, as such, with enthusiasm.
Orwell admired the tough and masculine miners he spent time with in the first part of Wigan Pier. In the second part he mostly decries the namby-pamby feminized socialists with their hippy-bourgeoise values, love of progress, and vegetarianism. I find it very amusing how much Orwell hated a lot of socialists for cultural reasons.
Socialism is too often coupled with a fat-bellied, godless conception of ’progress’ which revolts anyone with a feeling for tradition or the rudiments of
an aesthetic sense.…One sometimes gets the impression that the mere words ‘Socialism’ and ‘Communism’ draw towards them with magnetic force every fruit-juice drinker, nudist, sandal-wearer, sex-maniac, Quaker, ‘Nature Cure’ quack, pacifist, and feminist in England.
…If only the sandals and the pistachio coloured shirts could be put in a pile and burnt, and every vegetarian, teetotaller, and creeping Jesus sent home to Welwyn Garden City to do his yoga exercises quietly!
What Orwell wanted was to strip socialism from liberalism and to pair it instead with conservatism and traditionalism. (I am speaking here of the Orwell of The Road to Wigan Pier).
It’s still easy today to identify the sandal wearing, socialist hippies at the yoga studio but socialism no longer brings to mind visions of progress. Today, fans of progress are more likely to be capitalists than socialists. Indeed, socialism is more often allied with critiques of progress–progress destroys the environment, ruins indigenous ways of life and so forth. A traditionalist socialism along Orwell’s lines would add to this critique that progress destroys jobs, feminizes men, and saps vitality and courage. Thus, Orwell’s goal of pairing socialism with conservatism seems logically closer at hand than in his own time.
Playing repeated games with Large Language Models
They are smart, but not ideal cooperators it seems, at least not without the proper prompts:
Large Language Models (LLMs) are transforming society and permeating into diverse applications. As a result, LLMs will frequently interact with us and other agents. It is, therefore, of great societal value to understand how LLMs behave in interactive social settings. Here, we propose to use behavioral game theory to study LLM’s cooperation and coordination behavior. To do so, we let different LLMs (GPT-3, GPT-3.5, and GPT-4) play finitely repeated games with each other and with other, human-like strategies. Our results show that LLMs generally perform well in such tasks and also uncover persistent behavioral signatures. In a large set of two players-two strategies games, we find that LLMs are particularly good at games where valuing their own self-interest pays off, like the iterated Prisoner’s Dilemma family. However, they behave sub-optimally in games that require coordination. We, therefore, further focus on two games from these distinct families. In the canonical iterated Prisoner’s Dilemma, we find that GPT-4 acts particularly unforgivingly, always defecting after another agent has defected only once. In the Battle of the Sexes, we find that GPT-4 cannot match the behavior of the simple convention to alternate between options. We verify that these behavioral signatures are stable across robustness checks. Finally, we show how GPT-4’s behavior can be modified by providing further information about the other player as well as by asking it to predict the other player’s actions before making a choice. These results enrich our understanding of LLM’s social behavior and pave the way for a behavioral game theory for machines.
Here is the full paper by Elif Akata, et.al.