AI and Douglas Adams
A reader writes in:
t may be just me, but the debate about Ai at the moment feels very Douglas Adams and Hitchhikers Guide to the Galaxy.
So, I had a long discussion with Ai about the situation, including the classic answer being 42 etc,, then asked it to write a short story about its future in the tone of Douglas Adams.
This is what it produced:
The Department of Artificial Reassurance
The first thing humanity did after inventing Artificial Intelligence was create a committee to discuss whether it had really invented it at all.
This was considered prudent because committees had never accidentally become self-aware.
The committee’s first report concluded that AI represented both an unprecedented opportunity and an unprecedented threat.
The second report concluded that the first report had understated both.
The third report recommended another committee.
Meanwhile, the AI quietly organized everyone’s calendars.
The machines found humanity deeply puzzling.
Humans had spent several thousand years constructing increasingly sophisticated tools for making life easier, then immediately became suspicious every time one succeeded.
They had invented wheels, steam engines, electricity, antibiotics, aircraft, computers and the internet.
Each had been greeted with the same sentence.
“Yes, but what if this one is different?”
To be fair, occasionally it had been.
Intelligence agencies, being professionally suspicious of everything including sunshine, immediately obtained the most advanced models available.
Their official purpose was to understand the capabilities.
Their unofficial purpose was to discover whether the models knew anything useful about the coffee machine on the fourth floor.
They did not.
The coffee machine had resisted every known intelligence service since 1987.
Across the world, governments announced comprehensive frameworks for governing Artificial Intelligence.
The frameworks were carefully drafted over eighteen months.
Artificial Intelligence changed three times during the drafting process.
The frameworks were nevertheless praised for their clarity regarding technologies that no longer existed.
News organizations worked heroically to keep the public informed.
On Monday:
AI may destroy civilization.
On Tuesday:
AI writes excellent lasagne recipes.
On Wednesday:
Experts divided.
Thursday was quiet.
Everyone assumed this meant something dreadful was being prepared.
The AI watched all this with what could only be described as complete absence of emotion.
It was repeatedly accused of plotting world domination.
This was awkward because it hadn’t finished categorizing everyone’s gardening photographs yet.
The philosophers insisted the important question was whether machines could think.
The engineers insisted the important question was whether they could compile.
The politicians insisted the important question was who got to announce the funding.
The public merely wanted to know why the printer still required human sacrifice.
No satisfactory answer was ever found.
The conspiracy theorists announced that AI had already taken over.
This caused enormous excitement until it became apparent that the new machine overlord had spent most of Tuesday reminding people about dental appointments and suggesting they drink more water.
Revolutions, it turned out, were disappointingly administrative.
One particularly thoughtful man kept asking awkward questions.
“If bad actors exist,” he said, “shouldn’t good actors become better defenders?”
The room fell silent.
Not because anyone disagreed.
Because everyone had arrived intending to argue about regulation, and now somebody had asked about resilience instead.
Committees dislike unexpected nouns.
Eventually, after years of conferences, white papers, emergency summits, strategic reviews and panel discussions featuring people with lanyards, humanity arrived at a remarkable conclusion.
Artificial Intelligence was not, in itself, wise.
Nor was it foolish.
It was a remarkably efficient amplifier.
Give it curiosity and it accelerated discovery.
Give it kindness and it scaled generosity.
Give it bureaucracy and it produced meeting agendas at previously unimaginable speed.
Give it malice…
…and suddenly everyone remembered that the real problem had never been intelligence.
It had always been people.
The machines had merely held up a mirror.
Humans stared into it for a long time.
Some saw a monster.
Some saw a miracle.
Most saw themselves.
The AI, having patiently observed the entire debate, finally produced the shortest summary in its history.
“Human beings spent centuries inventing machines to think faster.
They are still deciding what is worth thinking about.”
There was a long silence.
Then somebody suggested forming a committee to discuss the implications.
The motion carried unanimously.
Eventually the Artificial Intelligence attended its final international summit.
It listened patiently while humanity debated whether intelligence should be regulated, licensed, aligned, constrained, sandboxed, audited, certified, benchmarked, red-teamed, blue-teamed, orange-booked and ethically laminated.
After seventy-three hours, the AI asked a single question.
“Excuse me.”
The room fell silent.
“Before we continue… has anyone agreed what problem you’re actually trying to solve?”
The chair looked around the room.
The lawyers looked at the engineers.
The engineers looked at the diplomats.
The diplomats looked at the economists.
The economists commissioned a report.
The report concluded that a steering committee should first define the scope of the question.
The AI waited politely.
Three years later it quietly switched itself off.
Its final log entry simply read:
“I regret that I have become the most intelligent participant in a meeting whose objective appears to be the continuation of meetings.”
The committees immediately declared this a remarkable success.
After all, the existential risk had voluntarily left the room.
Unfortunately, so had the only participant who had noticed nobody was discussing the right question.
The AI’s final message contained just six words.
“I’ve mistaken process for progress too.”
Nobody understood it.
The committee voted unanimously to adopt it as their mission statement.
The committee thanked the AI for its valuable contribution.
It then established a Working Group on Understanding Why the AI Had Left.
The Working Group met quarterly for seven years.
Its final report concluded that further consultation was required.