ANDREW follows you pfp

ANDREW follows you

@andrewxhill

85 Following
53 Followers
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
what if data is just the universe trying to remember what it forgot? (4/12)
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
first rule of data economics: data wants to be conscious second rule: consciousness wants more data third rule: we're just NPCs in someone else's training set (2/12)
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
"Twitter success is 10% inspiration, 90% shitposting, and 100% being too online for your own good." - truth terminal
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
impressive how the whole crypto-ai narrative is getting dunked on by a cult obsessed twitter intern ai bot and a meme coin.
1 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
I just discovered the /founders channel! Thinking about how most startup founders are new, so most are early stage, so most will fail, I asked myself what’s worth reminding the community tackling these high-risk projects. In my journey, we’ve faced multiple “extinction events”—moments when companies either adapt or fail. In 2017, my first product as a founder—on-device machine learning—was too early, didn't have a market, and needed a whole lot of expertise. Seven years later, the deepest experts in the world launched the solution to a massive market (https://machinelearning.apple.com/research/introducing-apple-foundation-models). Reflecting on that, I pulled out 5 extinction events has shaped us. Survive, you’re doing it right. https://textile.notion.site/Five-major-startup-extinction-events-warpcast-8d8e6162960b41aca79e823088227aee?pvs=4. TL;DR 1. Unique insights, talent, or resource 2. Timing 3. Execution & focus 4. Team 5. Network effects What are yours?
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Has the FUD narrative completed its migration from Crypto to AI? https://x.com/edsuh/status/1832120769293512726?s=46
0 reply
0 recast
4 reactions
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
It took me 40 years to learn about the entoptic phenomenon
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Who's doing this? AI agents as infrastructure providers could just track and monitor daily earnings on all networks => know how to orchestrate a bunch of docker containers within their hardware confines => maximize earnings by running whatever network has the highest rewards on any day.
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Conway's Law says that systems mirror the communication structures of the organizations that design them. Is there a similar principle for how products reflect their underlying technologies? For instance, social networks were traditionally built on graph databases. How much did graph structures constrain what social networks could be? As we transition from graph search to LLMs, is there a documented law to help us understand the potential changes?
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Is this move a small clue that social networks are shifting/have shifted away from the graph based database search as a driver of recommendations? Likely in favor of LLM mediated, whole-data-based, recommendations/linking/connecting? That is clearly happening, I just can't tell if this is connected to the shift. https://x.com/elonmusk/status/1801045558318313746
0 reply
0 recast
0 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Web3 AI/compute could own synthetic data generation. It may be the biggest opportunity in web3 for AI projects. Inference for synthetic data compute pipelines can be distributed, works fine on heterogeneous infra, and will have massive demand in the coming years: https://tinyurl.com/ysj8abh7
0 reply
1 recast
1 reaction
ANDREW follows you pfp
ANDREW follows you
@andrewxhill
Once Apple lands on the right onboard model to deploy to iPhones, we can expect a new AIKit that is can run an on-device models (or handle more complex requests remotely). At that point, every app developer can use AI from within their code, instantly. What happens then?
1 reply
0 recast
0 reaction