Content
@
https://warpcast.com/~/channel/all-in
0 reply
0 recast
0 reaction
nir.eth
@nir
Shocked to hear the AI czar say AI has just been a “good search engine and helped students cheat so far”
1 reply
0 recast
3 reactions
Zach
@zd
He's right though? Based on a quick Perplexity search, only ~3.5% of the American workforce works in tech, and only ~0.9% are SWEs So 3.5% of the workforce are power users, and 0.9% are super users The other ~96% of the workforce thinks ChatGPT = AI, and the questions they ask are basically Google searches or prompts to write essays (Anecdotally, my smartest "non-tech" friends who keep up decently well with tech news have zero idea what Claude or Perplexity are...)
1 reply
0 recast
2 reactions
nir.eth
@nir
Here’s one counter: Waymo’s ridesharing market share in SF grew to 22% in 2024, owning more of the market than Lyft and clearly a massive threat to Uber. 3.5% of the workforce in the US are drivers (trucks and rideshare). Feels kindaaaaa important to note
1 reply
0 recast
2 reactions
nir.eth
@nir
Second example: cursor doing $1B ARR in like a year of existing while SWE jobs all time low. Not homework, not search
1 reply
0 recast
2 reactions
nir.eth
@nir
Fastest co ever to $1B ARR and they’re 12 people like whaaaaa
1 reply
0 recast
1 reaction
Zach
@zd
These are all good points, and I didn't realize the % of drivers = the % of engineers - that's making me think a bit more about this... One idea: I think the difference is that autonomous driving has a clear workflow to automate. Yes, it's very hard to automate that workflow, but we have years and years of data to support "how to drive." The thing that's unclear about LLMs for the non-average user is just how many workflows they can tackle with them, which leads to a form of analysis paralysis. When I ask my non-tech friends "Have you tried asking GPT about x?", their response is typically "No I haven't - I didn't know it could do that."
1 reply
0 recast
1 reaction
