Everything and everywhere will change at once

Last week I had an experience as extraordinary as it was disturbing.

Craig Mundie, former director of research and strategy at Microsoft, was giving me a demo of GPT-4, the most advanced version of the AI ​​chatbot. ChatGPT, developed by Open AI and released in November.

Craig was preparing to inform the museum board about my wife, word planet, of which he is a member, on the effect ChatGPT will have on words, language and innovation.

“You have to understand,” Craig warned me before starting his demo, “this is going to change everything about how we do everything.

I think it represents mankind’s greatest invention until date. It’s qualitatively different and it’s going to be transformative.”

Craig added that large language modules like ChatGPT will grow their capabilities constantly and they will move us “toward a form of artificial general intelligence,” providing efficiencies in operations, ideas, discoveries, and insights “that have never before been achievable across all domains.”

Then he made a demo.

And I realized Craig wasn’t up to it.

First, he asked GPT-4, of which Craig was a select advanced tester and which had just been released to the public, to summarize Planet Word and its mission in 400 words.

He did it perfectly, in seconds.

Then he asked her to do the same in 200 words.

Just a few more seconds.

Then he asked her to do the same in Arabic. Just as fast.

Then in Mandarin.

Two more seconds. Later, in English, but in the form of a Shakespearean sonnet.

A few more seconds.

Craig then asked GPT-4 to write the same description to a alphabet versewhere the first line begins with the letter A, the second with B, and so on until the end of the alphabet.

He did so with astonishing creativity, starting:

From the origins of the word to the art of the pen. (From the origins of the word to the art of the pen.)

And so on up to Z.

That night I could hardly sleep.

Watching how an AI system – its software, microchips and connectivity – produces that level of originality in multiple languages ​​in seconds at a time, well, the first thing that came to mind was the science fiction writer’s observation Arthur C Clarke that “any sufficiently advanced technology is indistinguishable from magic”.

The second thing that came to mind was a moment at the beginning of “The Wizard of Oz”:

the tornado scene where everything and everyone is thrown into a whirlwind, including Dorothy and Toto, and then swept from mundane black-and-white Kansas to shimmery, futuristic land of ozwhere everything is in colour.

We are about to experience such a tornado.

We have entered a Promethean moment:

one of those moments in history when certain new tools, ways of thinking or sources of energy are introduced that are such a departure and advance from what existed before that you can’t change just one thing, you have to change everything.

That is, how it creates, how it competes, how it collaborates, how it functions, how it learns, how it governs and, yes, how it cheats, commits crimes and wages wars.

We know the main Promethean eras of the last 600 years:

the invention of pressthe scientific revolution, the agricultural revolution combined with Industrial Revolutionthe revolution of nuclear energyTHE personal computers and the internet and… now this moment.

But this Promethean moment is not guided by a single invention, such as the printer or the steam engine, but by a technological supercycle.

It is our ability to feel, digitize, process, learn, share and act, all increasingly with the help of AI.

This loop is introduced into everything – from the car to the fridge, to the smartphone or fighter jets – and drives more processes every day.

That’s why I call our era Promethean “The era of acceleration, amplification and democratization“.

Never have there been more humans with access to cheaper tools that amplify their power at an ever-faster rate, as they permeate the personal and work lives of more and more people simultaneously.

And it’s happening faster than almost anyone expected.

The potential of these tools to solve seemingly impossible problems, from human biology to fusion energy to climate change, is staggering.

Let’s think of an example that most people probably haven’t even heard of:

the way in which deep mind, an artificial intelligence laboratory owned by Alphabet, The parent company of Google, recently made use of its artificial intelligence system AlfaFold solve one of science’s most complex problems, at a speed and with a scope that amazed scientists who had spent their careers slowly and painstakingly approaching a solution.

The problem is known as “protein folding“.

Proteins are large, complex molecules made up of chains of amino acids.

And as my colleague from the Times explained half falls In an article in AlphaFold, proteins are “the microscopic mechanisms that direct the behavior of the human body and all other living things.”

What each protein can do, however, largely depends on its unique three-dimensional structure.

Once scientists can “identify the forms of proteins,” Metz added, “they will be able to accelerate the ability to understand disease, create new medicines, and otherwise probe the mysteries of life on Earth.”

But second science newsit took “decades of slow experiments” to reveal “the structure of more than 194,000 proteins, all housed in the Protein Data Bank”.

In 2022, however, “the AlphaFold database exploded with predicted structures for more than 200 million proteins.”

For a human being that would be worth a Nobel prize.

​ maybe two.

And with it our understanding of the human body has taken a giant leap. As a 2021 scientific paper, “Unfolding AI’s Potential,” published by the Bipartisan Policy Center states, AlphaFold is a metatechnology:

‘Metatechnologies have the ability to… help find patterns that contribute to discoveries in virtually all disciplines.’

ChatGPT is another one of those metatechnologies.

But as Dorothy discovered when she was suddenly transported to Oz, there was a good witch and an evil witch there, both fighting for her soul.

The same will happen with ChatGPT, Google Bard and AlphaFold.

Are we prepared?

It does not seem:

We’re debating whether to ban books at the dawn of a technology that can summarize or answer questions about virtually any book for everyone in a second.

Like so many modern digital technologies based on software and chips, AI is dual-use: it can be a tool or a weapon.

The last time we invented a technology this powerful, we created nuclear energy, which could be used to light up an entire country or destroy an entire planet.

But the problem with nuclear power is that it was developed by governments, who collectively created a system of controls to stop its proliferation by bad actors – not perfectly, but not bad.

AI, by contrast, is driven by private for-profit companies.

The question we must ask, Craig said, is how to manage a country and a world where these AI technologies “can be weapons or tools down the line” while being controlled by private companies and accelerating their power by the day. day.

And do it so the baby doesn’t get thrown out with the bathwater.

We will have to develop what I call “complex adaptive coalitions”where corporations, governments, social entrepreneurs, educators, competing superpowers, and moral philosophers come together to define how to get the best out of and buffer the worst from AI.

No actor in this coalition can solve the problem alone.

It requires a very different model of government from traditional left-right politics.

And we will have to make the transition between the worst tensions between the great powers since the end of the Cold War and the culture wars raging in virtually all democracies.

We better get this sorted out quick because, Toto, we’re not in Kansas anymore.

c.2023 The New York Times Society