Failing fast is a lot harder than it sounds

"The tech industry has always been a big proponent of failing fast, but Corporate America is now catching on thanks to AI. The tech allows companies to quickly launch new tools to a wide group. And, perhaps more importantly, it's a way for executives to prove the big money they are spending on AI isn't going to waste. (Hence why they want you to fail fast.)"

"There are still some hurdles with the fail-fast approach. What does failing even look like? There isn't a big, red alarm that goes off every time an AI project fails. (Although that sure would be fun.) Executives will say they set clear guidelines for a prototype beforehand, but a tool's benefits can be nuanced. And maybe you just need a little bit more time to really make it sing. All that makes it a lot harder to decide when to pull the plug."

"Building airplanes in the sky. The rush to get things out can lead to the belief you'll just fix things on the fly. But that's a dangerous precedent. Just ask the video game industry. When games came in physical copies you had to blow on to get working, companies made sure their product was bulletproof. Nowadays, they can digitally ship a game knowing an update for a glitch wont be far behind. It's a dangerous game that can cause major headaches."

Samuel Beckett's 'fail better' ethos has become a corporate mantra as executives embrace rapid experimentation. Companies from Okta to Salesforce and Blackstone view getting things wrong as acceptable if organizations iterate quickly and extract lessons. AI enables fast launches and speeds ideation by allowing chatbots to research and develop concepts that previously took hours. Executives use fail-fast to demonstrate that large AI investments produce learnings rather than wasted spend. Challenges remain: defining failure, setting prototype success criteria, and deciding when to terminate projects. Rushing to ship can create persistent defects, while formal launches ensure alignment and stability.

Read at Business Insider