Tom Rachman is a contributing columnist for The Globe and Mail.
We’ve spent years frazzled by social media, with YouTube nudging the disgruntled into conspiracy lunacy, Instagram tempting teens to self-harm, and Twitter converting newshounds into screen-checking bores. But blaming everything on social media misses the point. The deeper issue is what I’d call “the velocity imbalance”: that tech keeps hurtling at us, but human culture adapts gradually.
Already, another revolution is rushing at us: artificial intelligence. This could – and I’m not exaggerating – save our planet. It could also replace us with paper clips. It’s probably worth figuring out which. Except, we can’t. It’s happening too fast, and with unpredictable ends.
Consider a recent roundup from the Last Week in AI newsletter: San Francisco debates allowing police robots to kill suspects; AI invents materials that didn’t exist; AI is programmed with someone’s childhood diaries, allowing the grown author to chat with her younger self; AI composes top-level student essays; and AI experts openly worry about what they’re creating.
That was seven days.
Our current AI tools – personal assistants such as Siri and Alexa, for example, or e-mail autocomplete – are nifty enough. But AI’s potential is far greater: creating new pharmaceuticals, making travel by road and sky safer, offering cutting-edge medical diagnosis to the most deprived corners of the planet. Among the keenest hopes is that AI might rescue us from the climate nightmare that we humans seem so incapable of averting.
Every few months, another AI advance bursts into view, leaving tech-watchers agog. The latest is ChatGPT, a text-based AI program that has absorbed much of what is on the internet and can instantly create any piece of writing you request, from Shakespearean sonnets to sitcom scripts to recipes. You can converse with it, too. It’s not conscious, just eerily bright.
Already, AI has bettered human levels of reading comprehension, language understanding, image recognition and speech recognition. AI gets our humour. It can produce artworks, and high-definition photos of people who never existed, though you couldn’t tell.
The magic (and danger) of AI is that it improves upon humanity, bypassing our clunky mental processes, our needs for sleep and payment for our work, and our forgetfulness. A long-standing concern is that our superintelligent servants could be too literal. What if we ask an AI to make us smile, and it obliges by sending shocks into our cheeks?
Presumably, we could pull the plug. But, as the philosopher Nick Bostrom points out, we cannot simply reverse course on technological advances – try turning off the internet, for example. Even if humanity agreed that a superintelligent AI had become harmful, we might lack the brains to halt it. As Mr. Bostrom puts it: “Why haven’t the chimpanzees switched the off-switch to humanity?”
Tech giants such as Google, Meta and OpenAI (co-founded by Elon Musk) are far ahead of public awareness or government regulation. When a new product is ready, the wizardry is integrated into industrial practices and devices, subtly tweaking society without seeking permission, or quite knowing the impact.
“We’re on a very steep curve,” the head of OpenAI, Sam Altman, told The New York Times. “We will hit limits, but we don’t know where those will be. We’ll also discover new things that are really powerful. We don’t know what those will be either.”
In that case, full steam ahead!
Governments inch toward regulation like caterpillars watching trains pass. Canada is working on an Artificial Intelligence and Data Act (AIDA). The European Union has proposed a law on AI risks. The U.S. is working up an AI Bill of Rights.
Regulation is vital. But it’s too narrow. We need something that transcends national frontiers and political preferences – something like a Constitution of Human Dignity, proclaiming first principles about the AI outcomes we dread, what we must not cede, and what we cherish. Any tech innovation should meet those constitutional standards, or face penalties – even destruction.
A lesser-noted fact of this era is how much we’re reconsidering humanity itself, more so than since the theory of evolution. Often, we seem glum about ourselves, noting all the human frailties exposed by behavioural economics, by neuroscience, by the superior powers of tech. But we should also contemplate what we admire about humankind, putting aside for a moment the aspirations of competing groups, such that we safeguard the better parts of our nature against the better products of Silicon Valley.
In recent years, we’ve obsessed over each update, mesmerized by what our devices can do, mindless of what we are. Yet we sense the velocity imbalance from a pang in the gut: awe but dread. This isn’t just about social media. This is how life will be from now on.
Either we assert what humanity should be – or we let the machines decide for us.