Weekend Reading #211
This is the two-hundred-and-eleventh weekly edition of our newsletter, Weekend Reading, sent out on Saturday 1st April 2023.
To receive a copy each week directly into your inbox, sign up here.
*****
What we’re thinking
Yes, we are hedge fund managers and it's been a great week on markets but to be honest this week we haven’t been able to think about too much else other than AI. After continuously researching and learning as much as we can without being specialists, this week saw a confluence of material for anyone eager to upskill. There is ZERO doubt in our minds that this is the biggest thing we have ever had to think about in our lives. It is not overhyped, and it does not get too much airtime. It is quite simply relevant to everything in our lives, personally and professionally from here on out so best we give it our best go to understand what is going on. Today we include links to interviews with Tyler Cowan, Geoff Hinton, Sam Altman and Eliezer Yudkowsky as well as a bit more. If you haven't done any research before, these chats are a good place to start. For me it’s not actually that clear yet which way it will go as we let the genie out of the bottle. The promise is immense, the risk literally existential.
What we’re doing
I recently went out to Ravenscourt Park in West London to go bouldering at the aptly named, RavensWall. I’ve gotten quite into climbing over the first few months, originally as a result of the new gym I decided to join post-covid. Despite having now left that gym in search of a more budget-friendly alternative, I’ve kept up the climbing and try to go a couple of times a month at least. It’s a great workout and feels very different to other sports I’ve tried as the sheer range of different muscles and parts of your body it trains is insane, resulting in soreness all over for the days after. Interestingly, it’s not just me that seems to be enjoying it; I’ve noticed that bouldering as a sport seems to be gaining quite some popularity recently with more of my friends eager to give it a try each time I go. HS
What we’re reading
This article, entitled ”The Age of Average” is a must read. Once your read it you the penny will drop immediately, and you will literally exclaim in agreement. Everywhere we go, everything we see is becoming more and more similar. Interiors, architecture, cars, brands, media and even the way people look! It’s pretty grim. Or is it? This piece goes deep into example after example and tries to shed some light on this. There is no lightbulb moment or satisfying explanation though. Only evidence. Maybe because it's so easy to find a guide for how to do anything these days, so few think from first principles anymore. I wonder where I’ve heard that before. DC
What does crypto have in common with payday lenders? For one, they’re very much a challenge to the incumbent banking system. More than that, according to a law firm called Cooper & Kirk in Washington DC, both have been (and in the case of crypto, still is) the subject of a subtle and coordinated plan of suppression. Cooper & Kirk published this white paper entitled “Operation Choke Point 2.0: The Federal Bank Regulators Come For Crypto” and it is quite the read. For context, the original Operation Choke Point, officially ended in 2017 under the Trump administration, was an Obama-era programme that sought to discourage banks from doing business with a range of companies deemed unsavoury. Cooper & Kirk were also the firm that successfully sued the FDIC, the OCC and the Fed in the lawsuit Advance America et al. v. Federal Deposit Insurance Corp. et al., exposing the extent of the regulatory overreach. With everything that’s happening to crypto, certainly moved up a notch since FTX’s downfall, it’s a compelling argument that this the second coming of Operation Choke Point. EL
What we’re watching
Geoff Hinton rarely does interviews. He is the grandfather of AI. Or the godfather as this interview with CBS is titled. He has been involved with machine learning since the 70s. The problem back then for neural networks was the computers weren't fast enough and the data sets not big enough. As such they were considered an inferior approach versus the established mainstream at the time. No such problems today. He chats about the moment his team proved neural networks were the superior approach and how it changed the movement at the time instantly. This was only in 2012 while working with Google. He speaks of his primary fascination with the human brain rather than machine learning and AI. He says that the machine learning practices today have diverged from the way the brain works, primarily because our brain is just so limited in terms of how it communicates versus the machines which can literally plug into each other create the powerful neural networks we have come to know. The good news is that in his view we are still far better at reasoning whereas todays AI definitively has more knowledge. The bad news is he is in the camp that its inevitable we will be overtaken. The small segue into a soliloquy on autonomous lethal weapons and AI was particularly fun. 50 years later, he has lots to tell and lots for us to learn.
When I listened to Eliezer Yudkowsky’s chat with the Bankless team a few weeks back, one of the frustrations was that I was listening to a guy who had a very strong perspective who has been around a long time but that the interviewer wasn’t able to get the best out of him. So, this week I got 3 hours of watching Lex Fridman with Yudkowsky and boy did they go deep. This was a fantastic conversation. Let me remind you that Yudkowsky is a highly unusual chap. Unfortunately, Yudkowsky’s case is even more clear now too. His point (in fairness as it was on Bankless) is that we only get one shot to get it right in terms of solving for alignment. His view is historically with all important technologies, none of which have as much risk as this, we could learn through iteration. In this case, one false step, one non-aligned machine and he believes we are goners. My favourite part is about 1 hour 30 mins in when Lex asks him about how exactly the machines will kill us and Yudkowsky proceeds to role play with Lex. Lex was rather perplexed but gave it a good go. Unfortunately for him the harder he tried, the more ridiculous he looked. Towards the end, with some softer questions, Yudkowsky gets quite emotional in response to questions about mortality. Most of us probably would too if we were that negative on the future of humanity. If the content of this disturbs you, you are not alone. But it's worth it just for the education and also for Yudkowsky’s extraordinary facial expressions.
My initial sensation when watching the Lex Fridman conversation with Open AI founder, Sam Altman, was one of boredom. I wouldn’t exactly give Altman a prize for charisma but for literally everyone on earth, this is a critical conversation. This man is in charge of the most (publicly known anyway) advanced piece of AI on the planet, a piece of technology that according to some (especially Mr Yudkowsky above) is pandora’s box. So, I pushed through to what is a very rewarding conversation. Altman is clearly an unusual chap, but he is under no illusions as to the risks involved in creating possible sentient machines. The question is whether he is overconfident in his team’s ability to keep any possible sentient machine constrained. What this conversation does open up are the positives that harnessing this technology can bring and though much time is spent on the possible risks of AI, the blue sky is deeply enticing. DC
What we’re listening to
Tyler Cowan appeared on the Honestly podcast with Bari Weiss this week and spoke about quite a bit including bank runs and AI. His view on AI is positive and he believes we are entering a golden age for creativity, productivity and solving some of humanity’s biggest challenges. He addresses the risks head on and is honest about them but the sheer possibility of what we can achieve gets him rather more excited. His focus is on how much the world will change and how quickly it can happen. This guy is not only super smart but simplifies things so beautifully.
David Einhorn appeared on Invest Like the Best this week with Patrick O’Shaughnessy. Einhorn is famous as one of the great value hedge fund managers and in this episode, he talks about what value actually is (spoiler: it's not what it was) and his views on its relevance as an approach in years ahead. He talks of the difficulties he faced as a manager over the past few years which caused him to completely re-evaluate his entire process. He also covers running a business and handling analysts who work for him. This was a brilliant listen. DC
This week, I'm heading to Printworks London for a sold-out music event, which is likely to be my last visit to the iconic venue before it's re-developed into luxury housing and office space in Late May. The event will feature Jax Jones and a recently announced special DJ gues, Hugel. It was unsurprisingly quite difficult to get tickets to the event, and in the days building up to it I saw the ticket price spike in the resell market. Whilst I’ve been tempted to sit it out and instead make a potential c. 50% ROI, I’ve decided that it’s worth a going to this time. HS