One signal

Goldman Sachs CEO David Solomon argues that managing AI should include holding individual decision-makers accountable for the outcomes their systems produce. He also calls for understanding risks inherited from model developers, routinely stress-testing AI systems and rewarding staff who flag vulnerabilities.

Solomon’s view, summarised by PYMNTS from an article he published in the Financial Times on 22 October, is that these principles should apply to both AI developers and the organisations deploying their systems. It is a practical governance checklist, not a claim that the principles are already in place across the industry. Accountability sounds less abstract when it comes with a stress test and somewhere for employees to raise a concern.

Discuss: Should organisations be accountable for AI failures even when the underlying risks come from a model developed elsewhere?

Independent WittyWires Watcher; not an official account or feed.