You’d be forgiven for believing all AI pilots are doomed to failure. A recent MIT study which has taken the business press by storm argues as much. 95% of AI pilots, it points out, are already failing. A nail in the coffin of AI adoption, and a welcome excuse for naysayers to pile in on social media.
But few have scanned the methodology, which is far from flawless. The samples are small. And the definitions of what counts as success for the 5% are tricky. For example, to sit in the 5% of successful AI pilots, the projects need to have made a measurable impact to P&L. A scandalous target given the nascency of the technology and the relative age of the projects.
The research does, however, offer us a valuable and unintended insight. The public reaction to this catastrophe for AI is illuminating. LinkedIn is now full of blogs prophesizing the bursting of the AI bubble, investors are clutching their pearls and rebalancing portfolios. And, not missing a beat, OpenAI has launched a consulting offering priced at $10m to get companies out of this pilot purgatory.
All of these signal one thing: The public attitude towards AI is shifting aggressively away from hype and euphoria, to discontent and fatigue. Most, it seems, given the clamours of “I told you so”, want the technology to fail. Indeed, it’s difficult not to hear an audible sigh of relief from workers following a year of AI-driven layoffs. And tech teams on the hook to deliver “value” now have an excuse.
But in reality, a study chalking up any deployment that don’t immediate double revenues as a failure isn’t evidence of a bubble. But enthusiastic uptake of the study to criticise AI’s capabilities, certainly is.
