AI is an irresistible idea – like flying cars nobody knows how to land or manage

3 weeks ago 1

Column Steve Jobs probably didn't remember how many times he skinned his knees learning to ride a bike before describing a personal computer as a "bicycle for the mind." Jobs' point was that both tools help us to go further, faster, with just a little extra effort.

Today, the remark serves as a reminder that it's not possible to immediately become productive using a technology you’ve never used before.

That's something I learned during my first encounter with the original Macintosh, the first computer billed as "intuitive." I managed to get myself in front of one at a local computer retailer about 48 hours after its January 1984 launch. Put my hand on the mouse. Dragged it around a bit. Um... So what happens now?

Like most of the population I was utterly naïve about the language of windows, pointers, folders and clicking, I had no idea what to do – and the Macintosh wasn't offering any assistance. After a few minutes of frustration, I gave up. A few days later, after a friend walked me through the "desktop metaphor," I became a rabid WIMP enthusiast. But I learned something from my encounter in perfect ignorance: Even the most intuitive piece of machinery requires users to have some level of prior knowledge about its operations and the metaphors that inform them.

If the personal computer is the bicycle of the mind, artificial intelligence must surely rate as the "flying car" of the mind: Not-quite-imaginary, something everybody wants, but nowhere near scalable because our existing air traffic control systems were not built to handle a few million flying cars.

And just like the bicycle, no one is going to climb into the cockpit of a flying car and immediately fly swiftly or safely.

Yet over the last three years vendors have been passing out "weapons-grade" AI tools like candy to all comers. Expecting someone with no training in artificial intelligence to make something of it, just because it offers an "intuitive" conversational interface, understates the profound nature of the innovation and overestimates users' prowess.

In the right and well-trained hands – and when used in the appropriate ways – artificial intelligence can relieve some everyday drudgery. But that's not something you can get from a generic tool, nor something you'll learn from a casual interaction with those tools. Careful consideration of workflow and process and data and outcomes – none of them very sexy, yet all essential – drive useful forms of artificial intelligence. Without all of that, a project inevitably ends up among the 95 percent of junked AI proofs of concept that never made it into production – because no one bothered to learn how to fly the car.

But still vendors insist on throwing naïve users the keys, expecting that they'll soon be soaring in the wild blue yonder, then make themselves scarce when – inevitably – a flaming wreck comes crashing back to Earth.

Anyone who reckons AI to be a tool that allows them to magic up some productivity without going through the hard yards of learning all the spells and incantations needed to make that wand do something worthwhile is heading for a hard landing.

Fasten your seat belts! ®

Read Entire Article