Deployed ML models may stay in production for fifty years

Evidence

As Princeton’s Arvind Narayanan puts it: “Contrary to the ‘tech moves too fast for society to keep up’ cliché, commercial deployments of tech often move glacially—just look at the banking and airline mainframes still running. ML [machine-learning] models being trained today might still be in production in 50 years, and that’s terrifying.” — Brian Christian, The Alignment Problem, 229

Categorization

• XREF: Creates a tension with the common narrative of rapid tech disruption; connects to the Google Photos ‘gorilla’ label still manually deactivated years later (124). — connects to: Deployed machine-learning models often persist unchanged for decades in production
• SEED: Essay angle: ‘the glaciers of tech’—challenging the assumption that we can wait for problems to self-correct.
• corroborated: 2 independent passages support this claim — Brian Christian, The Alignment Problem, 229

Broader Connections

• Deployed machine-learning models often persist unchanged for decades in production #corroborates (0.80)