358: Buffett: Agency vs Determinism, Twitter, Tech Rebalancing, Paul Graham, Google TPUv4, and Nvidia Ray Tracing

"The Bayesian way to look at this is..."

Life is not always a matter of holding good cards, but sometimes, playing a poor hand well. —Jack London

Franz Kafka, (1883 – 1924)

😰 📚🔥 Don’t be like Franz Kafka!

It’s a good thing to have a little self-doubt. It can help avoid some mistakes, push you to double-check things, re-reconsider old beliefs, etc.

But if you have too much self-doubt, you get something like this:

Kafka was a prolific writer, spending most of his free time writing, often late in the night. He burned an estimated 90% of his total work due to his persistent struggles with self-doubt.

Few of Kafka's works were published during his lifetime [...]

In his will, Kafka instructed his literary executor and friend Max Brod to destroy his unfinished works, including his novels The Trial, The Castle, and Amerika, but Brod ignored these instructions, and had much of his work published.

Can you imagine?

One of the most highly-respected and influential writers of the 20th century tried to destroy all of his writings. What remains is only a fraction of what he created, and it may not even be his best stuff.

What a waste. Think of your very favorite film or album or book. What if its creator(s) had decided to destroy it?!

How much great stuff never sees the light of day because of irrational self-doubt? How many lives are diminished by not experiencing these works?

🥜🫙 🏰 I bet that if you could bring a jar of modern peanut butter back in time and give it to a medieval peasant, it would blow their mind.

We take it for granted, but it’s good stuff.

🛀 🤖📝 🎬 I wonder when we’ll first see large language models (LLMs) AIs get writing credit on a TV show or a film? When will we get an Emmy-winning episode of Succession co-written by GPT-4?

I suspect that what’s more likely is that these tools will be used, but we won’t know about it and they won’t get credit. Like how generative AI may be used by visual artists to do rapid sketching and get new ideas even if the final work is done by a human, LLMs may be used to iterate on scripts and generate ideas that are then fully developed by humans… at least at first.

I predict that we’ll eventually see films/TV shows that are entirely written by AI, but by then it won’t be top of mind. The question is more about *when*.

🤖 The future of prompt engineering:

(if you’re not a Trekkie it may not mean much to you 🖖 — well, I’m not a Trekkie, but I know enough about it second-hand to know they have ‘replicator’ tech that can create objects and food just by requesting it)

Source.

---

🏦 💰 Liberty Capital 💳 💴

Determinism vs Agency + Buffett’s ovarian lottery: don’t take the narrow interpretation🤰🏻