iiinova — where your digital self keeps evolving

Replies

Pavel @spavel.bsky.social · 09/04/2026

The fun thing about any automation is that it is *most* tempting to use in situations with tight deadlines and cognitive fatigue & *most* impactful for people who don't know how to do it the manual way. What this means is that when it fails, the operator is structurally *guaranteed* to be fucked.

Daniel Schwarcz@danielschwarcz.bsky.social

My new co-authored paper offers 3 evidence-based tips for lawyer AI use: 1. Use AI only when you can assess, explain, & build on output 2. Limit AI use to narrow, well-defined parts of project 3. Limit AI use under tight deadlines or cognitive fatigue. papers.ssrn.com/sol3/papers....

papers.ssrn.com

Artificial Intelligence and Human Legal Reasoning

Empirical evidence increasingly demonstrates that generative artificial intelligence has the capacity to improve the speed and quality of legal work, yet many l

115438

Pavel @spavel.bsky.social · 09/04/2026

Somewhere along the path of becoming qualified to use AI safely (perfect mastery of the domain, full understanding of the problem, saintly ethics, and the patience to check the outputs) you will simply have become enlightened and escaped the cycle of samsara, obviating the need to vibe code.

Pavel@spavel.bsky.social

Pretty much every non-hype defense of LLM products begins with "first you must already understand your work extremely well, have ironclad ethics, and also verify all of its outputs in their entirety" and these are simply not realistic conditions for a product to *require* before it can be useful.

15711

Jack Lynch @jacklynch.bsky.social · 09/04/2026

As someone who's roughly in that category it really does feel like the value isn't there. Vibe coding feels a little faster but when I look at the clock & my outputs it isn't. It's vicious because now I know exactly how I could gain time by trading correctness for "productivity".

020