- do these people understand the principles of making good products?

- is anyone clearly working towards a microtransaction system that could replace advertising and subscription models?

After attending two conferences, hundreds of conversations and hours spent researching, my conclusion to both questions was no. The community felt more like an ouroboros. It was disappointing.

I don't want to pay NYT a subscription fee, I want to pay them some fraction of a cent per paragraph of article that I load in. Same goes for seconds of video on YouTube, etc.

Apparently I'm alone in this vision, or at least very rare...

For my own use case I was trying to test consistency or an evaluation process and found that injecting a UUID into the system prompt (busting cache) made a material difference.

Without it, resubmitting the same inputs in close time intervals (e.g. 1, 5, or 30 min) would produce very consistent evaluations. Adding the UUID would decrease consistency (showing true evaluation consistency not artificially improved by catching) and highlight ambiguous evaluation criteria that was causing problems.

So I wonder how much prompt caching is a factor here. I think these LLM providers (all of them) are caching several layers beyond just tokenization.

It makes sense to me that a culture that values collectivistic cohesion would shy away from paradigm shifting ideas (disruption). I also see the correlation between disruptive ideas driven by principled critical thinking over conventional thinking.

I guess on some level my assumption is that they are adjacent. Those embedded in a collectivistic culture can think critically but can run into walls within a sandbox of convention. This is how they can be great at iterative improvement and engineering but struggle with paradigm shifting ideas.

I think you have a point, but there's definitely some nuance here I'm still untangling.

It seems you are unnecessarily muddying the water.

I can't help but wonder whether the person who gave you advice "to think critically wasn't for [you]" didn't have YOUR best interests at heart, and/or wasn't a wise person.

I also worked jobs where I was actively discouraged to think critically. Those jobs made me itchy and I moved on. Every time I did it was one step back, three steps forward. My career has been a weird zigzag like that but trended up exponentially over 25 years.

We all have our anecdotes we can share. But ask yourself this: if you get better at making decisions and communicating with other people, who is that most likely to benefit?

The hardest part of startups is probably the making good decisions part. To be a good VC you need to be better at founders at judging startup decisions, AND you need to be good at LP deal flow AND you need to be good at startup deal flow. LP deal flow has to come first (otherwise there is no fund), and because of zirp a lot of VCs got funds up without good startup deal flow or the ability to judge startups well.

In other words it's hard to be good a VC too, but for a while it was artificially easy to be a bad VC.

Most VCs have no idea how to accuratly judge startups based on their core merit, or how to make good decision in startups (though they may think they do), so instead they focus on things like "will this founder be able to hype up this startup and sell the next round so I can mark it up on my books".

> Individual intelligence will mean nothing once we have superhuman AI, at which point the difference between an obscenely talented giga-nerd and an ordinary six-pack-drinking bozo will be about as meaningful as the difference between any two ants. If what you do involves anything related to the human capacity for reason, reflection, insight, creativity, or thought, you will be meat for the coltan mines.

Believing this feels incredibly unwise to me. I think it's going to do more damage than the AI itself will.

To any impressionable students reading this: the most valuable and important thing you can learn will be to think critically and communicate well. No AI can take it away from you, and the more powerful AI will get the more you will be able to harness it's potential. Don't let these people saying this ahit discourage you from building a good life.

Comments URL: https://news.ycombinator.com/item?id=47082452

Points: 1

# Comments: 0

For anyone seeing 404

They could do away with all these mazes of settings and configurations and just have a little chat thing. You pop open and then tell the AI hey I want to change the background and then it just does it. You could have a huge and complex array of settings that would be a headache to navigate in a typical form format, but a breeze with an llm that has an API into them.

As an aside, another one that I just find hilarious is the LLM implementation into Google sheets. I'll ask it. "Hey how do I do this?" and then it goes "I don't know" and I'm like WTF why is this here

Personally, I found this example really helpful to wrap my head around jobs to be done theory.

It switched their frame from trying to innovate on the milkshake itself to the whole experience. Instead of making the milkshake chunky or thick or sweet or whatever they move the milkshake machine up closer to the door so people can be in and out faster.

I'm curious why you didn't see it this way. Did you miss something, or am I missing something?