Advertisement
Advertisement

IBM Builds an AI ‘Undo Button’ to Fix Cloud Outages Faster

Story Highlights

IBM and the University of Illinois at Urbana-Champaign created a system called STRATUS.

IBM Builds an AI ‘Undo Button’ to Fix Cloud Outages Faster

Recent cloud outages are becoming more expensive, with some reports saying that they now cost about $14,000 per minute. This is nearly 10% more than in 2022. This has put pressure on site reliability engineers to fix issues quickly. Therefore, companies are turning to artificial intelligence for help. These systems help spot the cause of problems but aren’t yet trusted to fix things on their own, mainly because they lack an undo button and a clear way to reverse mistakes. However, tech giant IBM (IBM) and the University of Illinois at Urbana-Champaign created a system called STRATUS to help solve this issue.

Meet Your ETF AI Analyst

Like the undo shortcut Ctrl+Z, STRATUS can automatically roll back bad decisions made by AI agents during incident responses. For example, if a fix makes things worse, STRATUS resets the system to a safe checkpoint in order to try another solution. Interestingly, in tests using open-source cloud benchmarks from Microsoft (MSFT) and IBM, STRATUS outperformed other top tools by at least 150%.

Moreover, STRATUS relies on a method called transactional-no-regression (TNR), which means that it only allows changes that can be reversed and that won’t break anything. For example, if an agent tries to delete a database, the system stops it unless it’s safe and undoable. As a result, researchers believe that this model could be used for more than just cloud outages.

Is IBM a Buy, Sell, or Hold?

Turning to Wall Street, analysts have a Moderate Buy consensus rating on IBM stock based on seven Buys, five Holds, and one Sell assigned in the past three months, as indicated by the graphic below. Furthermore, the average IBM price target of $295.18 per share implies 3.4% downside risk.

See more IBM analyst ratings

Disclaimer & DisclosureReport an Issue

1