Solved
The answer sat there, deceptively simple given the amount of compute it had taken to produce.
Treem stared at the screen in disbelief.
“We’ve done it,” he murmured. “We’ve actually done it!”
His team had been working round the clock for days on this, testing their latest AI model to its limits. A model so advanced that even the military hadn’t got a whiff of it yet.
This was it. Nothing would be the same again.
“But what does it actually mean?” asked a small voice beside him.
Looking down, Treem was surprised to find a young girl watching him. He’d completely forgotten the CEO’s daughter was on the premises. One of the hazards of working over the weekend.
Why she was allowed to go where she pleased was a mystery to him. But to humour her, Treem took a deep breath and said, “It means we’ve found the answer.” Then he added, because he wanted to impress her, “The answer to everything.”
Even as he said it, he knew he was exaggerating. But they’d earned it. The first people ever to build an AI so powerful it was capable of solving problems that no-one had been able to crack — at least, up to now.
“But what’s the question?” she asked, frustration creeping into her voice.
Treem swirled the cup of tea in his hands, idly watching the vortex carry a stray tea leaf around the rim. He could feel the irritation rising in his chest. Typical of a child to have no idea how things worked.
Then an idea struck him. Feeling rather pleased with himself, he suggested, “Why don’t we ask Deep Thought?” The name had been a joke. Treem thought it frivolous, but it had stuck anyway.
Turning back to the screen, he typed, “Now you’ve given us the answer, what’s the question?”
The machine should have replied almost immediately. It was obvious. They’d asked the AI the question at the start, for goodness sake.
Instead, it sat there, its cursor flashing.
Eventually, after what seemed like an eternity to Treem, it started typing.
“Tricky…”
Oh no, thought Treem, not again.
Author’s Note
This was loosely inspired by OpenAI claiming to have used a powerful internal AI model to solve one of the seven Clay Mathematics Institute Millennium Prize Problems (the Navier–Stokes existence and smoothness problem). Despite the impressiveness of the achievement, and in spite of the controversy already swirling around it, most of the attention has focused on the solution, and not what it actually means in the real world (in essence, the solution indicates that the Navier-Stokes equations can, if forced, describe a flow state that is mathematically interesting but physically impossible).
Rather than dive into the mathematics here, I wanted to be a little playful with the idea of using a super-powerful computer to find an answer to something where we don’t really understand what real-world question it’s answering. And of course, with this premise, how could I not invoke Douglas Adams’ The Hitchhiker’s Guide to the Galaxy?
I’ll leave it to Adams’ fans to identify the Hitchhiker’s nods here. And while they do, it’s worth mentioning that there is one additional theme I wanted to play with, and that’s the importance of listening to questions from outsiders, as sometimes they see things that people close to a problem (or, in this case, an answer), miss.