AI Isn't The Problem
Technology for technology's sake is
AI projects are failing. Go figure. đ
The numbers vary depending on the source, but itâs commonly reported that somewhere between 80% and 95% of AI projects fail to deliver any sort of value or return on investment.
Nearly every week thereâs news about another AI debacle.
Recently we heard about Starbucks walking away from their AI-powered inventory system after only 9 months. Before that it was the failed AI deals in Californiaâs LA and San Diego school districts.
Is anyone truly surprised by all of this?
AI is just another technology in a long string of advances weâve seen over the past 50+ years. If thereâs one thing Iâve learned over my career as a technology leader, itâs that we should never adopt technology for technologyâs sake.
But thatâs what weâre currently doing with AIâand to be fair, what we did with virtually every other technology that came before it.
AI is the shiny penny du jour.
When I was at Amazon, we were instructed to use AI and so use it we didâbadly at first, and without any clear intended outcome. AI wrote Forte feedback and OP docs, some so nonsensical they were functionally useless. AI chatbots served up (wrong) answers from long-since abandoned internal wiki pages (IYKYK).
I see the same AI charge in many of the colleges and universities I work with now and as a former CIO, it breaks my heart.
Everyone is scrambling to âdo AI.â
Every institution I talk to is in some stage of building their AI strategy, drafting an AI policy, establishing AI governance, or piloting AI tools.
We reacted and are still reacting: AI is here and we must do something with it and exert control over itâbecause who doesnât love a little governance?
We have an AI âhammerâ and weâre hammering away at institutional challenges like nobodyâs business.
Even if what they really need is a screwdriver.
AI isnât the problem.
The technology is good enough. Or it isnât, depending on who you ask or what youâre trying to do. Thatâs not really the point.
Weâve got to stop chasing shiny pennies and doing tech for techâs sake.
For many years I fought back against putting a separate technology priority into my then institutionâs strategic plan. Technology isnât a strategy unto itself. Neither is AI.
The strategy is and always should be an organizational oneâone that technology may enable or support.
I have no doubt that AI will be a transformative technologyâlike computers or the Internetâthat will fundamentally change our lives. To get there, though, we must:
Stop doing AI for AIâs sake, and start getting clearer on the specific business problem weâre trying to solve.
Stop applying AI to our existing jobs and processes, and start doing things that werenât previously possibleâuntil now. Itâs time to completely reimagine our organizations and our work with this new technology, not just do it incrementally better or faster.
Stop taking a one-size-fits-all approach to AI, and start building an AI toolbox with unique tools and capabilities that are fit for purpose.
Stop treating AI like a tech problem, and start recognizing that successful technology projectsâincluding and maybe especially AIârequire change. People, cultures, and organizations all must change. Thatâs a big lift. (See also: Change Is Messy.)
Itâs still early days.
AI has evolved faster than most new technologies, but is still in its relative infancy. Anyone remember the early days of the consumer Internet?
I do. I was in startups during the Internet boom of the late 1990s (and ironically am back in them once again for this next technological revolution). There was a lot of hype, but the websites were little more than digital brochureware and the business models were generally unsustainable. Turns out âeyeballsâ arenât such a good metric, after all.
Many of those experiments failedâlike AI is âfailingâ now.
We went from the dotcom boom in the late â90s to the dotcom bust of the early 2000s, and we learned a few things in the process.
Much like with the Internet, I have no doubt that weâll look back 20 years from now and not be able to believe that we once learned, worked, and lived the way we currently do.
But until then, weâre likely to see a few more failuresâbecause itâs not just technology that has to change, but all of us along with it.
Bonus read: Not to put too fine a point on it, but things are moving so quickly that this article I wrote six months ago is now basically obsoleteâalthough I did get a couple of things right in it.
Enjoying this blog? SGNR is 100% free and always written by a humanâme. Subscribe to get new posts delivered directly to your inbox. Thanks for supporting my work!