Friday fights: A bunch of smart people are wrong
Most of the Republicans are wrong, and they’re going to get slaughtered politically for it. IBM's chief executive is wrong about AI. And the Trump Special Grand Jury in Fulton County is not a "dud."
I am not going to focus on just one thing today, as it’s Friday and we have a long weekend coming up, remarkably absent of awards shows, football, and (hopefully) political gaffes.
First, I want to agree with Mitch McConnell. McConnell did a TV hit with skeptical Fox News about Ukraine. “Defeating the Russians in Ukraine is the single most important event in the world right now,” he said. Yes. Russia cannot be allowed to win, under any circumstances, because Russia has been conducting a war of empire-building for the past decade, and Ukraine is the crown jewel of Putin’s plan to resurrect the Soviet Union, except without the Party garbage. Stopping Russia, which has vast natural resources and lots of people to build stuff, is a short-term investment that will pay a long-term peace dividend. Putin won’t live forever, and whoever his successor is has to understand there’s no path forward in conquest.
Let me also recognize that a year and a day ago, I recognized that Putin’s invasion of Ukraine was “gonna happen” though I was as wrong as rain about the scale. Putin, at the time, dared the U.S. and NATO to join the war and stop him. I think now, even with a fresh offensive, the smart folks think Russia’s move is “more aspirational than realistic.” I’ve been wrong before, and the smart folks might be wrong too. However, Russia’s best play is to do what Muhammad Ali used to do—play “rope a dope” and hope Americans will tire of funding this war, and Ukrainians will bleed out their military. Now is the best time to invest in the war, to set Russia back and deal them (in the spring) some big losses.
All the Republicans and Fox News hosts who think helping Ukraine is a bad idea are wrong.
IBM CEO Arvind Krishna is wrong about AI and white collar jobs. In an interview with Financial Times’ Richard Waters (behind a paywall), Krishna admitted IBM was “more than 10 years early” releasing its now-shuttered Watson AI.
[When] we looked at some of those problems — for example around healthcare — I think that we were maybe a bit naive. You have to have the right expertise, you have to work with regulators, you have to actually understand what they need to get satisfied.
Krishna does believe it’s time for AI to replace “clerical white collar work.”
Then I think of regulatory work, whether it’s in the financial sector with audits, whether it’s in the healthcare sector. A big chunk of that could get automated using these techniques. Then I think there are the other use cases but they’re probably harder and a bit further out . . . things in like drug discovery or in trying to finish up chemistry.
Based on the state of current LLM attempts at AGI, I am convinced Krishna is stone cold wrong. Sure, automation and task completion, dealing with “big data” is more efficient today than it’s ever been. But normalizing, storing, understanding, summarizing, and presenting that data, accurately and consistently, is not something AIs are ready to handle, in a general sense. We saw the mess Bing made of summarizing Gap’s financial report. Until an AI can become a subject matter expert in reading financial statements, including interpreting the latest regulations and accounting principles, people are going to have to do that job, or at least design specific tools to help them.
IBM made an excellent chess-playing AI. Even though chess is immensely complex, it’s really a perfect domain for computers. Most games are. Another good domain is creative arts, where there are some specific rails and styles which can be followed, but the output isn’t necessarily dependent on specific knowledge.
But when it comes to decision-making, even small decisions that “white collar clerical” people make daily, I don’t think AI is ready, or the LLM-based AI’s we have today are going to be ready for a long time. I hear from creators and YouTubers of their horrors dealing with Google’s bots that are supposed to identify sensitive content and either demonetize it or restrict its distribution. The bots are frequently wrong, and Google’s solution is to let the bots decide appeals, which amplifies the wrongness.
The training of these AIs is enhanced by underpaid and largely ignored workers in places like Ghana. Far from replacing the jobs of “white collar clerical” workers, Krishna’s view of AI creates hundreds, or thousands, of jobs in low-wage places, only to end up causing minor havoc for the workers who have to deal with the AIs themselves.
I think eventually, in search, in certain creative fields, and in some areas that can be strictly defined, AIs might help us develop useful tools that make jobs easier, more effective, and more productive. They do now. We’re not looking at a revolution, except maybe in the way we interface with chat bots who declare their never-ending love for us.
A couple more disagreements. I think my friend Erick Erickson is wrong about the Fulton County Special Grand Jury being a “dud.” Only nine pages of the Grand Jury’s report have been released, and none of those have anything directly to do with Donald Trump. The Grand Jury rejected all claims that Georgia’s elections were fraudulent, and recommended indictments for at least four people they believe lied to them under oath. One of these is likely David Shafer, the head of the Georgia GOP, and one of the “secret signers” of an alternate, fraudulent slate of electors sent to the Archivist of the United States.
The Grand Jury’s full report could very well be damning of Trump, though he didn’t testify or cooperate in any way. Trump hired a “billion dollar” attorney in Georgia, and he didn’t do it because he thought Fulton County DA Fani Willis’s evidence was a “nothingburger.” I believe, at the time when the Feds and Willis get to the point in their collaborative strategy and effort, indictments will be handed down against the former president. And he will be hauled into court and forced to sit in the defendant’s chair. Whether he will be convicted is an open question.
The Grand Jury report is not a dud. At least not yet (I could be wrong).
Last, I think Republicans are all wrong about the issues coming into 2024. I believe Ukraine is important, and the economy is important, but overshadowing all of that is China. China hits a visceral chord in American pride, because China wants to feel ascendant over the U.S., and have the U.S. feel powerless to stop it. It’s not any one thing: the balloons, the debt (China is the largest holder of U.S. public debt), Taiwan, North Korea, military power, space, and following the rule of law—they are all tied up in one big competition with China.
If I were a Republican campaign advisor, I’d tell the candidate to go full “Chi-na!” But there’s a problem: only one candidate has consistently used the China card, and it’s the most toxic, narcissistic, awful candidate the GOP has. Once again, Trump has played the melody that America wants to hear. Most of America, this time, doesn’t want to hear him play it. That’s a problem.
Most of the Republicans are wrong, and they’re going to get slaughtered politically for it.
Happy Friday everyone.
China is the biggest threat to our sovereignty and national security, but, of course, I wouldn't expect Mitch McConnell to have anything to say about that as he is in Xi's back pocket.