With any of the "high power experiment" folks on YouTube, I'm always glad when they upload again... you start to worry after a few years!
I love how this video has a modern "podcast" setup, yet the lighting and camera quality are still consistent with a decade ago when I started watching the zany experiments.
I had similar question as you and conclusion as OC.
My answer maybe abstract in some ways, but:
- something that requires network effects, e.g social media, even twitter with megahitler stuff is going strong. Even though twitter's base app model could be done in 30min by LLM;
- something that maps to infrastructure in someway. Good luck copying Cloudflare, residential proxy provider, established email provider or VPS with LLM in a weekend. Though parts of it are definitely easier;
- something that is a platform (adjacent to first point). Google's search platform, ads, docs, youtube, music, email, business tools, etc. Similar to Meta (seemingly half of the world's small businesses control their business with whatsapp) and others. Though parts of those platforms can be disrupted with LLMs;
eh, compared to salaries even shiniest macs wouldn't register in USA, especially hubs like SV. Especially if it's issued once a couple or four years.
With AI spend - especially at Anthropic/OpenAI API prices - spending 0.5-1 macbook a month worth of tokens per employee seems normal (from what I've seen, heard, what happens in this forum and thread and what my employer does).
Tokenmaxxxing (followed by commit/PR-maxxxing) is celebrated. Don't think - just launch orchestrator that will create dozen child agent sessions to do useless work to impress your superiors. If buy the end of the day you will ask to increase your token allowance - you're a superhero.
I am probably looking waaaay too much into this, but I was told 5+ years ago that streaming platforms will develop AI to plugin "natural" product placement for each user.
For example, ad algorithm will draw product placement with specific products that it thinks you're the target. Protagonist will drink pepsi/coke/juice, will use linux/mac, will drive MB/BMW/Ford - depending on what ad algorithm thinks you'd like to buy.
Somehow I feel like this "A/B testing" feature is actually a cornerstone by YouTube for AI driven media future, for exactly this (dystopian) reason.
Early cars were called horseless carriages for a reason.
Car bodies were carriages for few first decades of car history. So if you can fix and maintain horse carriage - you can maintain car carriage.
Engines were simple too and designed to be maintained in a field environment with few tools. So it’s reasonable that this skill was learned, especially by mechanically inclined people.
reply