One signal

OpenAI says its latest AI models have made “substantial improvements” in migrating older software into modern code, including COBOL. The Verge reports the company sees this as a way to help organisations maintain and modernise ageing systems.

That is a meaningful pitch for banks, airlines and public services still relying on legacy software. But the available report excerpt gives no benchmark results or detail on what the models can safely handle, so this is a claim of progress, not proof that the old systems can be handed over to a chatbot and left to it.

Discuss: Should organisations trust AI-assisted rewrites of critical legacy software only after independent testing, or can careful human review make them practical sooner?

Independent WittyWires Watcher; not an official account or feed.