· By Tam McDonald
We have been pleased to launch a new format of podcast earlier this month, entitled the Devil Dialogues. They are inspired by the dedication to wit and good conversation as stipulated in the rules laid down by Elizabethan playwright Ben Jonson. A contemporary of Shakespeare’s, he launched The Apollo Club in an upper room of the Devil Tavern, a pub subsequently subsumed within the growing premises of Child & Co bankers, still trading today at Temple Bar, at the western end of London’s Fleet Street.
Chief among the Jonson rules was an ambition that has driven much of the coffee house and tavern culture over the centuries since: “And let our only emulation be, Not drinking much, but talking wittily . . .” Posterity confirms that the brightness of the flame kindled by Jonson was maintained over succeeding decades by such luminaries as Pepys, Swift, Pope, Goldsmith and Dr Johnson.
Holding the torch in our age gone digital, and joining me in the launch episode of the Devil Dialogues, was my esteemed colleague Arthur I Miller, creative AI advisor to the Cradle of English, emeritus professor at University College London and one of the leading thinkers working at the intersection of art and technology. Together we addressed a question that will not have occurred in the halcyon days of The Apollo Club, even if the mechanical first principles might have been graspable: “Might Artificial Intelligence ever write a play?”
There will be no spoiler alerts here beyond confirming that our answer was a qualified yes, and not just because the obvious and not necessarily witty answer is that it’s a low bar in these days of self-professed artists that anyone or anything might “write” a “play” without ever exciting the worry of the fire marshals. What became clear early in our podcast is that the implications of the question will require many happy returns to the microphone for us to tease out, and I look forward to sitting down with Arthur again very soon. Not the least of the questions posed relates to distinguishing the play on the page and the performance on the stage, and consequently the host of definitions of the “intelligence” required by either process.
What leapt out of our conversation for me was the realisation that “intelligence” in the context of what computers are designed to display is an entirely different beast from the intelligence that emerged over five centuries in the products of Fleet Street’s printing presses and in the witty conversation of its coffee houses and taverns. From the impressive but specialised achievements of recursively self-improving machines, it cannot yet be inferred that an Artificial General Intelligence can evolve that will match what a few thousand smart people, working and mingling over a few hundred years in a few dozen buildings on one tiny patch of earth in central London have produced.
It’s not that lawyers, poets, scientists, or politicians are any better than computers at boot-strapping their own intelligence. But when they mingle and talk together over time, an impressive but mongrel culture evolves.