AI

AI is an amplifier: that is the thesis the DORA and delta team at Google Cloud placed at the centre of their April 2026 report on the ROI of AI-assisted software development. It magnifies the strengths of high-performing organizations and the dysfunctions of struggling ones at once. Injected into a system running manual testing, bureaucracy or fragmented data, it accelerates technical debt.

Other accounts describe subtraction. Dan Shipper, CEO of Every, builds his 21 May 2026 essay on a cycle in which AI devalues yesterday's human skill, that cheap skill is adopted at scale, abundance produces slop, and humans then demand difference again. Boris Cherny, reading the software landscape through Hamilton Helmer's 7 Powers, holds that switching costs and process power lose weight once the model hill-climbs a process, while network effects, scale economies and cornered resources hold. Hryhorii Tatsyi, CTO of Raiffeisen Bank Ukraine, found the compression landing on people: AI lifts underperformers to baseline, tightening the distribution from below rather than accelerating the top.

David Wallace-Wells supplies the political register. In the NYT Magazine of 8 May 2026 he places AI in the lineage of steam, electricity and the internet as a general-purpose technology, and as the object of a popular resentment, distinct from the movement that resentment feeds, a risk the five lab founders neglected while they debated existential ones. Pew Research recorded 50% of Americans more concerned in 2025 against 10% more enthusiastic.