thanks Robin - I have no training in this - just show up and be myself. That has pluses and minuses ;-)

Expand full comment
Feb 26Liked by Nate Hagens

Thanks, Nate, for presenting this point of view. It's the first time I've ever heard you say outright, "I don't buy it" during an interview. Your curiosity and restraint are a great model for conversations with folks you disagree with. Next time I'm in such a conversation, I'll ask myself, What Would Nate Do?

Expand full comment

A system that decouples “wealth” from the material world has distorted our perception of value.

A system that decouples “intelligence” from the material world has distorted our perception of wisdom.

Why wouldn’t a system that decouples “reality” from the material world distort our perception of meaning?

None of this is smart or cool in the long run.

Expand full comment

Appreciate the podcast, it’s always good to hear different perspectives. I do feel there were shares of simplification with comments of sharping/maintaining skillsets to build and to grow. This episode has pushed me to realize that AI is not going away and will only get stronger so I feel it will be important to learn how to apply it properly, however, remaining grounded and engaged with the real and beautiful world around us.

Expand full comment

Oof this was a dystopian wet dream. Why have nature when you can simulate reality? Why have a dog when you can have an AI dog? Why have an Amazon rainforest when you can have a simulation of rainforest.

All simulations are approximations, they are lossy. Do you want the Monalisa or a print of one?

Perhaps pragmatic about the bleak future ahead but there was almost an excitement about the technological journey to get there. A blindness to another way or very identified with the way the world is and its “inevitable” trajectory. An excited doomsdayer.

This is perhaps symptomatic of the techno utopian dream. Progress is only up and to the right.

On AI. Synthetic data is generally used when you have a smallish dataset. In these situations it overfits to the data you have, meaning it will fail on datasets that vary from the training set. Synthetic data is used to add noise to the training set so that it will hopefully fit a more general case rather than a specific case.

Expand full comment

I liked his insight. I never heard of him before. That's what I like about TGS, I get introduced to new minds.

Expand full comment