Need Better Testing
Software Glitch has Global Impact
Today Crowdstrike, which supplies security for Microsoft Defender and used by Microsoft 365 rolled out an update which effectively crashed systems worldwide. Airlines, financial systems, banks and more were impacted. The company has identified the issue and is rolling out a resolution.
Software Glitch has National Impact
Some months ago, AT&T had a software glitch that impact millions nationally. Daily communications and emergency services are impacted.
Lessons Learned
Above are two recent events caused by issues with software. In the old days these types of things were contained within a relatively small population of systems. Today however, these software glitches have far reaching impact. National and even international systems are impacted. The ‘reach’ or impact of these issues are far larger then ever before. We have a limited amount of technology companies offering systems throughout the world. When there’s a bug in an update, millions (perhaps even billions) can be impacted. We must have better testing mechanisms put into place to reduce the possibility of these errors surfacing,
The Scary Part
The scary part is that the same software development methodology that is being used with AI. Software glitches and bugs will have a much larger impact than what we’ve seen from Crowdstrike and AT&T. That should be cause for concern. If we cannot implement stringent testing processes (and adhere to those) for software written by humans, what do we do to test software that is written by AI? To me there are two main scary parts.
- Insufficient testing processes for software, AI or not.
- Managements need to rush product out the door to meet their projected numbers. The ship it and we’ll patch it later methodology has to stop. Mediocrity has no place in software development.