NEW EPISODE OF THE SPHERE PODCAST:

Pascal-Emmanuel Gobry, Publisher of Sphere Media, interviews Sebastien Laye, a Franco-American economist and AI entrepreneur, columnist for the Washington Examiner on AI policy and founder of the Economic Singularity Institute. They discuss the "AI Action Plan" released by the White House to much fanfare: its regulatory outlook, its allegedly pro-worker agenda, issues relating to national security and export controls and immigration, and more.

Listen to this episode on Spotify, YouTube, or Apple Podcasts.

And, of course, SUBSCRIBE to the Sphere Podcast on Spotify, YouTube, or Apple Podcasts.

Building A Global Compute Alliance To Win The AI Race

Janet Egan has published a very interesting report for the Center for a New American Security, which is a somewhat left-leaning think tank that seeks to help the US pursue a broadly realist foreign policy.

The basic idea is this: global AI leadership seems now and for the foreseeable future to depend on "compute" power, that is to say, building massive data centers with large amounts of specialized chips and consuming large amounts of power.

"If current AI development trends continue, securing and extending U.S. access to a robust compute ecosystem will play a decisive role in whether the United States leads the world in AI or cedes its leadership to competitors. The nation that leads in deploying compute worldwide will wield critical leverage over the rules and norms governing its use," Egan writes.

The basic idea of the report is refreshingly simple and straightforward: in order to make sure that it takes the leadership in compute, the US should try to build the most compute, not just on its own, but with its closest allies. She points out that it would actually reinforce "the United States’ position as the AI partner of choice, expanding its global compute footprint and drawing swing states more closely into the U.S. orbit through deeper technology partnerships."

She notes approvingly the Trump Administration's partnerships with Saudi Arabia and the UAE, but argues that "these initiatives should be integrated into a comprehensive AI diplomacy strategy."

Egan singles out a handful of "key strategic partners": Saudi Arabia, the UAE, Brazil and India, Australia, and the United Kingdom.

Theres an obvious missing potential partner, which is France, and which could be a key partner to the US on AI because it's currently the only Western democracy that has a massive surplus of electricity.

Very good and thoughtful report overall.

Policy News You Need To Know

#ItsTheEconomyStupid — The big news is that GDP growth came in for Q2 at an annualized rate of 3.0%, which is very good and came in above economists' and Wall Street estimates. Obama CEA Chair Jason Furman says this is mostly due to "massive timing shifts" but if so, why did the analysts not anticipate it? Once again, everything seems to be coming up Trump, and the economy remains stubbornly robust despite tariffs and everything else.

SEE ALSO: Trump Wins Everything On Trade

#AI — Speaking of AI, Mark Zuckerberg has made a very interesting and striking announcement: "Over the last few months we have begun to see glimpses of our AI systems improving themselves. The improvement is slow for now, but undeniable. Developing superintelligence is now in sight." This is the key thing that the AGI/ASI vision depends on: the idea that AI can one day get smart enough to improve itself, so that it can keep improving itself and keep becoming more and more intelligent, arbitrarily, until it becomes "superintelligent," that is to say, more intelligent than humanity. Is this true? If it is true now, will it lead to AGI/ASI? Obviously we have no way of independently evaluating those claims.

#Tariffs — Speaking of the economy and tariffs, you may have seen that tariff revenue since the beginning of the Administration has hit $150 billion. This is starting to get to a non-negligible number. Essentially a tax increase worth roughly 1% of GDP per year, which all goes toward reducing the deficit. This is not negligible.

#Tariffs — Speaking of reducing the deficits, though, Senator Hawley has just come out with a bill that would rebate tariff revenue as checks to Americans. Under his bill, every eligible adult would receive $600, plus $600 per dependent qualifying child, i.e. a family of four would receive $2,400. The bill also provides that if tariff revenue exceeds projections, the rebate per person could increase, calculated as the total qualifying tariff proceeds divided by the number of eligible individuals and their qualifying children (rounded to the nearest dollar). It includes an income phaseout. Ok, why not! It might be worth doing one year to help with the midterms. Or it might be worth banking the tariff revenue to reduce the deficit. This is really a political choice. Both ideas are defensible. (And, of course, since ideological free marketers don't want the tariffs to exist to begin with, they have little grounds to oppose rebating the revenue as checks or doing anything else with the revenue!)

#Tariffs #Trade — President Trump has also made a number of statements on trade: the first (contrary to some speculations by some of his cabinet officers) that he will not extend the August 1 deadline for new tariffs; the second that he is slapping a 25% tariff on India, in spite of the good relations with that country, because of its non-tariff barriers and its military cooperation with Russia.

#LGBT — North Carolina's legislature just overrode a veto of a law that, among other worthy initiatives, extends the statue of limitation for medical malpractice claims for transgender procedures to ten years. This is very important. As more and more evidence comes to light, there will have to be many many lawsuits, and we can't let the statute of limitation allow doctors to escape accountability.

#UBI — There's been another UBI study, this time specifically on children. This was the "Baby's First Years" study, a $22 million experiment funded by the National Institutes of Health, which provided $333 monthly, no strings attached, to 1,000 low-income mothers across multiple U.S. cities. Predictably and logically, the study found no significant improvement in children’s cognitive or behavioral outcomes after four years. Hilariously, the New York Times piece on this study quotes the lead author of the study as declaring of the outcome: “I was very surprised — we were all very surprised.” This is a good reason among many to reform higher ed so that conservative views are more often represented.

#HigherEd — Speaking of higher ed: two Hillsdale College faculty members, Professor of Philosophy Nathan Schlueter and Associate Professor of History Mark Moyar, received Freedom and Opportunity Academic Prizes from the Heritage Foundation. Here's the release. The prizes are awarded to faculty whose work relates to the ideas of freedom, opportunity, and traditional American values. It's very good that Heritage has and hands out these kinds of awards. Conservatives should have their own incentive systems to produce scholarship and academic value.

#HigherEd — Speaking of conservatives and higher ed reform: economists Giorgio Tripodi, Xiang Zheng, Yifan Qian, and Dashun Wang have published a big new study in PNAS on the link between tenure and research. Unsurprisingly, the "analysis reveals that faculty publication rates typically increase sharply during the tenure track and peak just before obtaining tenure."

#Politics #Culture — This is an old thing, but we had missed it, and it has contemporary relevance, as those kinds of gender politics have, in anything, more relevance today than they had in 2016: in 2017, NYU held a study where they had actors reenact the infamous Presidential debate between Hillary Clinton and Donald Trump, but gender-swapped the actors. An actress played Trump and an actor played Clinton. The actors not only repeated the words verbatim, but learned to imitate the candidates' gestures, tones, and mannerisms. The authors of the study were expecting to prove that Sexism(tm) is alive and well in America, as a representative sample of watchers would have negative views of the woman, even if she said everything Trump said, and therefore Americans voted against Clinton only because she was a woman. Instead the results were the opposite: the audience found the woman playing Trump to be likable and funny and found the man playing Clinton to be grating and weird. In particular, the male actor's constant smiling, deliberately mimicking Clinton's own compulsive smirking during the debates, came off as very creepy to viewers. The author of the study even ended up saying that it made him more sympathetic to Trump voters. Why are we discussing this now? Well, first of all, because it's funny. But also because, if anything, since then, the Democrats' public presentation have embraced even more of the "girlboss" affect that made Hillary Clinton so grating to many Americans of every background. This gender-social dynamic will be important in American politics in years to come.