Content
@
0 reply
0 recast
0 reaction
Stephen Caudill
@mrmemes.eth
“Here’s the crucial insight: as models improve, the value of engineering effort diminishes…” - if that’s true (and I suspect it is), how should it inform our approach to building AI products right now? https://lukaspetersson.com/blog/2025/bitter-vertical/
1 reply
0 recast
1 reaction
Stephen Caudill
@mrmemes.eth
The easy answer is that we should be building model-agnostic applications that don't over-invest in engineering that will be devalued - but that's virtually useless to say. I'm honestly not sure where the line lays between what's a useful level of augmentation of current capabilities and rapid depreciation of value as new models become available. What's useful today could be completely superseded by e.g. O3.
1 reply
0 recast
0 reaction
Stephen Caudill
@mrmemes.eth
I'm really just starting to think seriously about this, and am extremely curious what others think. How do we navigate these interims?
0 reply
0 recast
0 reaction
