Vogt

Little bit of this, little bit of that

[Vogt @vogto78]

@signulll To cope w/ negative human value and have purpose we'll either merge e.g. human-ai hybrids (control) or create new games that have humans at the core e.g. sports (give up control). Prob both.

[Vogt @vogto78]

Substack should create an ebook tablet. A simple, low-light, paper-like tablet that is just your substack feed. Kind of like the Kindle was just Audible books.

I think it fits their ethos well. Long-form oriented, thoughtful, and distinct from the addicting feel of an iphone or social media feed.

[Vogt @vogto78]

@OANN They're arguing that you can achieve the same results on sema as you can on tirz... but it takes an unreasonably high dose of sema to do so and with far worse side effects. dumb argument. spend time on better drugs instead of lawyers

[Vogt @vogto78]

GLP-1 molecule tirzepatide (Zepbound, Mounjaro) hit $10B/yr faster than the iPhone. Under 2 years from launch.

Most hyperbolic pharma drug growth excluding covid vaccines.

[Vogt @vogto78]

@maxmarchione I think the bad stuff (food noise, addiction, scrolling) are all from external side-effects of living in abundance. Whereas human ambition is something internal.

The stories ab people "over-dosing" on GLP-1s and not having motivation feel like effects of not eating enough.

[Vogt @vogto78]

This conflates "people will go out more" with "people will drink more"

The spike in beer is price increases, not sales volume. And fastest growing segments are NA beers...
https://t.co/Z52y8RAbYG

[Xavier Armand @x_armand]

Regret to inform that one of the hottest Billion dollar AI companies (synthetic population simulations) predicts a huge resurgence in alcohol sales because of GLP1s.

...and beer sales just spiked.

[Vogt @vogto78]

@imagecrate He back

[Vogt @vogto78]

@ritwikpavan Would love to join!

[Vogt @vogto78]

"we're adding one of our engineers to the chat"

[Vogt @vogto78]

@aervrea Yeah my b was costing too much and don’t have time to iterate or raise some money / partner

[Vogt @vogto78]

@rowancheung @JustinFineberg https://t.co/roL946sitG

[Vogt @vogto78]

Made a lil app that uses GPT-3 to write better prompts for GPT-3 (also ChatGPT and Dalle) �� https://t.co/MzSwRwOuqB

[Vogt @vogto78]

@Bluprint4Crypto Just put it back up!

[Vogt @vogto78]

Made a lil app that uses GPT-3 to write better prompts for GPT-3 (also ChatGPT and Dalle) 👉 https://t.co/MzSwRwOuqB

[Vogt @vogto78]

@charlescoult GPT-3 - text-davinci-003

[Vogt @vogto78]

Here are the prompts I am using:
- 2 for GPT-3 (one concise with templating, one more descriptive)
- 1 for Dalle-2

[Vogt @vogto78]

@JustinFineberg https://t.co/roL946sitG

[Vogt @vogto78]

Made a lil app that uses GPT-3 to write better prompts for GPT-3 (also ChatGPT and Dalle) �� https://t.co/MzSwRwOuqB

[Vogt @vogto78]

@ClontarfD3 @learn_prompting @goodside costs: Yeah can definitely add up.
token limits: A work around could be to summarize past exchanges.

One thing I've been looking into is storing the conversation separately and having GPT-3 search over it using something like @gpt_index. Feel like that could work.

[Vogt @vogto78]

@goodside For sure, depends on use case