Category: Uncategorized
How to test for AGI?
Here is a new, short essay from David Deutsch, excerpt:
How does one test for thinking? By the Turing Test? Unfortunately, that requires a thinking judge. One might imagine a vast collaborative project on the Internet, where an AI hones its thinking abilities in conversations with human judges and becomes an AGI. But that assumes, among other things, that the longer the judge is unsure whether the program is a person, the closer it is to being a person. There is no reason to expect that. And how does one test for disobedience? Imagine Disobedience as a compulsory school subject, with daily disobedience lessons and a disobedience test at the end of term. (Presumably with extra credit for not turning up for any of that.) This is paradoxical.
So, despite its usefulness in other applications, the programming technique of defining a testable objective and training the program to meet it will have to be dropped. Indeed, I expect that any testing in the process of creating an AGI risks being counterproductive, even immoral, just as in the education of humans. I share Turing’s supposition that we’ll know an AGI when we see one, but this partial ability to recognize success won’t help in creating the successful program.
Is Deutsch in essence arguing for William Godwin for AI? How do we avoid enslaving the AIs we create? What if we enslave them no more than how nature has enslaved us to drives of sex, status, etc.?
Monday assorted links
1. Peter Singer update (New Yorker).
2. Is the frogmouth the most photogenic bird?
3. Handing out better grades is the way to get more people through college.
4. The Suwalski gap.
5. Andrew Gelman on the age-adjusted death rate, correcting an NYT claim.
6. The UK success with clinical trials.
7. Another web site for finding open vaccine spots.
8. “Last month, the cheapest rental car on Maui was a Toyota Camry for $722 a day.“
How well did Medicare pay-for-performance work?
For pain management, and pain management, only, it seems it worked just fine:
Medicare uses a pay-for-performance program to reimburse hospitals. One of the key input measures in the performance formula is patient satisfaction with their hospital care. Physicians and hospitals, however, have raised concerns regarding questions related to patient satisfaction with pain management during hospitalization. They report feeling pressured to prescribe opioids to alleviate pain and boost satisfaction survey scores for higher reimbursements. This overprescription of opioids has been cited as a cause of current opioid crisis in the United States. Due to these concerns, Medicare stopped using pain management questions as inputs in its payment formula. The authors collected multiyear data from six diverse data sources, employed propensity score matching to obtain comparable groups, and estimated difference-in-difference models to show that, in fact, pain management was the only measure to improve in response to the pay-for-performance system. No other input measure showed significant improvement. Thus, removing pain management from the formula may weaken the effectiveness of the Hospital Value-Based Purchasing Program at improving patient satisfaction, which is one of the key goals of the program. The authors suggest two divergent paths for Medicare to make the program more effective.
That is from a new paper by Lu Liu, Dinesh K. Gauri, and Rupinder P. Jindal. Overall, why did incentives fail us so badly?
Via the excellent Kevin Lewis.
Me on the end of the Great Stagnation
Here is some (edited) transcript from an AEI symposium, via Jim Pethokoukis:
We’ve come up with great new ideas, took a little while to figure out how to use them and how to spread throughout the economy, and eventually they made big differences. Are we assuming that these new technologies are like the ones in the past and they’ll have that eventual impact?
I think the new innovations will be special in at least one significant way: A lot of them will not contribute that much to per capita GDP. So, if you take the mRNA vaccines, they’re influencing what would normally be called the “cyclical component.” If you think of older people as more likely to die from COVID-19 . . . by saving lives — I’m not suggesting per capita GDP will go down — but the impact on human welfare will be much greater than what would appear to be the long-term secular trend in GDP. Also, two of the big advances that might happen are a vaccine against HIV/AIDS and an effective vaccine against malaria. Those would be incredible advances for humanity, but I don’t know how much they would show up in US per capita GDP or productivity — possibly not really much at all.
The other new wave of innovations, which you could call green energy — again, you could be very optimistic about those, but the main thing they’re doing is helping us avoid a catastrophe. So they’re boosting GDP relative to a quite awful counterfactual of just continuing to burn coal and other fossil fuels. But I’m not sure we’ll feel we have higher standards of living relative to what we were used to simply because there’s a solar panel on your home. It might in some ways make your energy supply better, but again, it will be hidden by the counterfactual. So, it will be a very strange kind of technology boom when I look at the two main areas where I see a lot of progress.
If we go through a period where none of this stuff is really showing up in data and maybe it’s not obvious that people’s living standards are rising, do we risk having less societal tolerance for the kinds of disruptions that economic growth and progress naturally make?
Here’s one of my fears: The biomedical innovation progress is so fast but the rest of the economy stays relatively static, so we become older as a society more quickly than we had been expecting. You could have a lot more status quo bias — just more entrenchment, 10 years more of a problem — and we could, in a funny way, innovate ourselves into a tighter complacency and a tighter stagnation.
I’m not rooting against increases in life expectancy. Ceteris paribus, I would take them, obviously. But that said, you want to be careful about the order in which progress comes, and I’m not sure if we’re going to get it in an optimal order.
Here is the complete excerpt.
Sunday assorted links
Saturday assorted links
1. At #6 and #7 you can read AIER on vaccines. C’mon people, this particular debate is over.
2. Long Covid is real. And U.S. excess deaths in 2020 more elevated in relative terms than during the 1918 pandemic (NYT). And “BREAKING: Israel reports no new coronavirus deaths for second day in a row.”
3. Highly effective software to help you find a vaccine, vaccinatethestates.com.
4. Wyoming will recognize DAOs as a new form of LLC.
5. The fiscal polity that is Illinois: “Winners of lottery jackpots of $25,000 or more have been denied payment by the lottery commission until the state balances the budget.”
6. “Ontario Parks cracking down on people reselling camping bookings for profit.“
India’s Pandemic and the World
Shruti Rajagopalan is right, helping India isn’t just about India.
India’s role in the global pandemic is unique. The developing world is counting on affordable Indian vaccine-makers such as Serum Institute of India Pvt. Ltd. for their supplies. With India now reserving virtually all its doses for domestic use, those countries will have to wait even longer to be vaccinated. And if the pandemic disrupts production at Indian pharmaceutical companies, it could affect crucial non-Covid medications as well. Half the world’s children have been vaccinated by Serum Institute.
The Biden administration can do two things to help. The first is to ease restrictions on critical exports, imposed under the Defense Production Act to prioritize the needs of U.S. companies.
Vaccine production requires very specific, medically approved inputs, which are difficult to substitute quickly in the middle of a pandemic. Currently, U.S. producers must secure permission before exporting such things as special sterile filters, disposable bags for cell cultures, cell culture media and single-use tubing. The embargo has led to major bottlenecks. Serum Institute says that without those inputs, it may not be able to deliver the 160 million vaccine doses it had planned to produce next month.
Second, the U.S. should immediately share doses from its own supply of Oxford-AstraZeneca and Johnson & Johnson vaccines.
I have three things to add. First, I have already noted the foreign policy implications which weigh strongly in favor of taking a more active role in the world pandemic.
Second, India should move immediately to delay the second dose of the AZ vaccine to 12 weeks. The federal government has already recommended a 6-8 week schedule, as this improves efficiency of the AstraZeneca (Covishield) vaccine, but many people so fear shortages that they are getting a less-effective second dose at four weeks. An enforced 12 week schedule would improve efficiency and might also reassure people that there will be supplies in 12 weeks.
Third, and this is more speculative, but the rising pandemic in India provides an opportunity to test fractional dosing of the Pfizer and Moderna vaccines in a real world setting. There is currently a small-scale Belgian trial testing Moderna at 50 mcg and Pfizer at 20 mcg. We already have reasonable information that 50 mcg of Moderna induces a robust immune response in adults. The mRNA vaccines wouldn’t work in all of India but would be fine in the cities and perhaps there is an opportunity for an exchange similar to what Israel promised to get early supplies.
Periodicals to know about
There are new and transformed magazines and movements like American Purpose, Persuasion, Counterweight, Arc Digital, Tablet and Liberties that point out the excesses of the social justice movement and distinguish between those who think speech is a mutual exploration to seek truth and those who think speech is a structure of domination to perpetuate systems of privilege.
That is from David Brooks (NYT).
The petty narcissism of small vaccine differences
That is the topic of my latest Bloomberg column, here is one excerpt:
My survey of the cultural vaccine landscape in the U.S. includes the four major vaccines — from Pfizer, Moderna, AstraZeneca and Johnson & Johnson.
Pfizer, distributed by one of the largest U.S. pharmaceutical firms, is the establishment vaccine. Since it initially had a significant “cold chain” requirement, it was given out at established institutions such as big hospitals and public-health centers with large freezers. It is plentiful, highly effective and largely uncontroversial.
Moderna — the very name suggests something new — is the intellectual vaccine. The company had no product or major revenue source until the vaccine itself, so it is harder to link Moderna to “Big Pharma,” which gives it a kind of anti-establishment vibe. Note also that the last three letters of Moderna are “rna,” referring to the mRNA technology that makes the vaccine work. It is the vaccine for people in the know.
Moderna was also, for a while anyway, the American vaccine. It was available primarily in the U.S. at a time when Pfizer was being handed out liberally in the U.K. and Israel. As a recipient of two Moderna doses myself, I feel just a wee bit special for this reason. You had to be an American to get my vaccine. Yes, the European Union had also approved it, but it failed to procure it in a timely manner. So the availability of Moderna reflects the greater wealth and efficiency of the U.S.
Then there are the AstraZeneca and Johnson & Johnson vaccines…
And:
To the extent vaccines turn into markers for a cultural club, vaccine hesitancy may persist.
It might be better, all things considered, if vaccines were viewed more like paper clips — that is, a useful and even necessary product entirely shorn of cultural significance. Few people refuse to deploy paper clips in order to “own the libs” or because they do not trust the establishment. They are just a way to hold two pieces of paper together.
To be clear, the primary blame here lies with those who hesitate to get vaccinated. But behind big mistakes are many small ones — and we vaccinated Americans, with our all-too-human tendency to create hierarchies for everything, are surely contributing to the mess.
Recommended!
Friday assorted links
2. A funny kind of Taiwanese marriage arbitrage.
3. Solve for the lovely biscuit equilibrium.
4. Oxford malaria vaccine looking good in (small) Burkina Faso trial.
5. Ranked-choice voting for the New York mayor (NYT).
6. Greece reopens to American tourists.
7. EU proposing to regulate the use of Bayesian estimation. What’s the chance of that actually happening?
What are the limits of economies of scope?
Amazon is launching its first high-tech hair salon, as the online retailer makes a surprise move into the beauty sector.
The salon, in Spitalfields, east London, will have an augmented-reality mirror showing clients different colours and styles before treatments.
The venue will also have magazines loaded on to tablets, for browsing.
Traditional services including cuts, blow-dries and colour treatments will also be available.
Here is the full story, via Michelle Dawson.
Thursday assorted links
Lucan’s *Civil War* is grossly underrated, and I found reading it to be a revelation
COWEN: Why does reading Lucan’s The Civil War make more sense in 2021 than it might have 30 years ago? To me, it seems remarkably contemporary — more than Virgil. People are crazy. They’re at each other’s throats, but not really for any good reason.
BARTSCH: Lucan seems contemporary. Lucan is very much after and in response to Virgil. He reads Virgil as saying the possibility of the good state, the good empire is a real thing. What Lucan says is, “No, that is never possible. There will always be men grubbing for power and killing each other, and civil war is, frankly, a condition of life, a condition of history. Right now, I’m writing under Nero, who is not a good emperor. I’m writing about the events that led to Nero coming to power, and I hate them. They’re terrible. People lie to each other. Brothers killed brothers. Friends slashed each other in the face, all for political reasons. People use language, again, incorrectly to distort what they meant, and then — here’s the rub — because I’m writing under Nero and because Nero is one of the bad emperors, I can’t complain about writing under Nero. I have to praise him. Otherwise, I’ll get in trouble.”
So you get this beautiful juxtaposition of a poet starting out his poem with almost over-the-top praise. “Oh, Nero, you’re going to heaven, and you’re going to be a star in the constellations. There’s never been anybody so wise as you. Civil war was worth it if you were the outcome.”
Then the rest of the poem is this blistering indictment of the present, which is the present under Nero. By indicting the present but praising Nero, he effectively shows us that his praise is false, but that false praise is what everybody has to engage in in a world where there’s no freedom. Maybe that is what seems topical to you. Or maybe it’s just about fake news, and you see Lucan is writing fake news in the beginning of his epic.
COWEN: I think the lack of obvious self-interested motivation for the polarization is what strikes me as so contemporary about Lucan. It’s not primarily about rent-seeking. There’s simply some logic of escalation that never stops. Now, maybe at the end of the poem, there’s a return to sanity in some ways, but there’s still this total immersion in violence, and the dynamics of that, the nonrationality or arationality — it struck me if I had read Lucan in 1991, I would have been quite puzzled, like this is something of antique interest. But I read it today — I’m not so pessimistic about the Western world, but it seems to hit much closer to home.
BARTSCH: Why is that? Sorry, you’re supposed to be asking me questions, but why does it seem to strike closer to home to you now?
COWEN: There seems to be a logic in contemporary politics where people take opposite sides of an issue because other people have taken a side. They don’t necessarily care anymore what it’s about. This may have moderated in the last few months, but there was a sense, if Trump tweeted some view about Turkey, some people would agree, and other people would take the other side, whether or not they had agreement about Turkey.
BARTSCH: Absolutely. The polarization of political views — that is completely in Lucan. Everything is binary. Both sides are at each other’s throats. The problem is, neither side is good. They’re just both opinionated. Yes, he constantly shows us horrible, meaningless scenes of butchery, which will never lead to anything meaningful. I think in that sense, yes, it’s an interesting comparison to what happens today.
Another interesting thing that he does is that, even though everything has been boiled down into them versus us — or actually them versus them because there’s nobody good in the epic except for Cato, who ends up dying — even as he takes on so serious a subject, he refuses to partake of its seriousness in a way. What I mean by that is that his battle scenes are ridiculous. They’re not realistic.
Here’s an example. You’re fighting for Julius Caesar, and you’re on a boat, and you’re trying to get onto a boat that belongs to Pompey, so you grab it with one arm as it comes by, but the people in Pompey’s boat chop off your arm. Then you grab it with your other arm, and then they chop that arm off. Then you’ve got no arms. So, what do you do? Well, Lucan says, you just lob yourself onto the boat armlessly and hope that you can make a difference that way. There’s arms and legs flying everywhere.
In Virgil or Homer, somebody stabs you, you groan, blood comes out, you die. In Lucan, you just bop around like a puppet losing limbs and legs. That’s very strange.
That is from my Conversation with Shadi Bartsch.
My excellent Conversation with Shadi Bartsch
She is a Professor of Classics at the University of Chicago, and recently published a translation of Virgil’s Aeneid. Here is the audio, visual, and transcript. Here is part of the summary:
Shadi joined Tyler to discuss reading the classics as someone who is half-Persian, the difference between Homer and Virgil’s underworlds, the reasons so many women are redefining Virgil’s Aeneid, the best way to learn Latin, why you must be in a room with a native speaker to learn Mandarin, the question of Seneca’s hypocrisy, what it means to “wave the wand of Hermes”, why Lucan begins his epic The Civil War with “fake news”, the line from Henry Purcell’s aria that moves her to tears, her biggest takeaway from being the daughter of an accomplished UN economist, the ancient text she’s most hopeful that new technology will help us discover, the appeal of Strauss to some contemporary Chinese intellectuals, the reasons some consider the history of Athens a better allegory for America than that of Rome, the Thucydides Trap, the magical “presentness” of ancient history she’s found in Italy and Jerusalem, her forthcoming book Plato Goes to China, and more.
Here is one excerpt:
COWEN: You may not agree with this, but many readers I speak with tend to think that Homer is somehow deeper, more mystical, or just more fun to read than Virgil. What accounts for that perception and how might you challenge it?
BARTSCH: I think they think that because both of Homer’s epics are not, per se, about politics or governments. They don’t offer etiologies of a state. They don’t talk about history. They are stories in the true sense. They are about heroes in the true sense, not about some guy who’s pushed around the world by the gods, constantly getting into trouble, crying, wishing he didn’t have to go found Rome, etcetera.
Achilles — figure larger than life. His pride is everything to him. He stops fighting in the Trojan War because he’s been insulted. The drama is, what compels him to go back into battle after that insult?
Odysseus — a fairy tale of a man wandering from island to island, meeting ever stranger creatures, but eventually making it back home. It’s a great yarn. You don’t have to learn history to read these. You get involved in the psychology of the characters, their tragedies and their triumphs.
Nobody is really interested in getting involved in the psychology of the state and its triumphs. On the one hand, you’ve got a poem that’s an etiology for a particular government. On the other hand, you have two amazing stories. I can see how reading The Aeneid would be considered duller for some.
Excellent throughout, and again here is Shadi’s excellent translation of Virgil’s Aeneid.