SFF2023: From Productivity Paradox to Productivity Polarity

Below is a readout of my prepared remarks from Singapore Fintech Festival 2023

Hello. Thank you so much for having me today. I'm Thomas, and I'm the CEO of GitHub.But I like to say, I’m Thomas, and I’m a developer. I’ve loved building things with technology, with software, almost my entire life. 

Beginning with my childhood in East Berlin, back before the wall fell. And that love only increased when I got access to the internet in the mid-1990s and even more so with the rise of open source.

Today, with artificial intelligence, we stand at an inflection point that has been brewing since the advent of the home computer. All of our advances in the past fifty years have led us to the intersection we now stand in. 

And, the biggest question is, what will we do with the opportunity – and the risk at hand. 

Now, if you recall, in the 1970s and 1980s, the world was riding high on a hype similar to the one we are experiencing today with artificial intelligence. 

It was the home computer. It was new. It felt radical. Until this point, computers were shared resources in big companies, governments, and universities. You had to book a time slot to get access and run a computation. 

The home computer looked to promise a new age of productivity for the everyday worker. Even in the early days, growth was exponential across computers that today, almost no one remembers. 

The Mac and the PC came and the gold rush flowed in. The demand signified the early hope. That something novel was about to revolutionize our way of work and life. And it did. 

Over the decades, investments in IT shot through the roof in the United States, and around the world. But despite the gold rush of IT investments, the rate of productivity increases did not go up. In fact, at that moment in time, with the birth of the PC, productivity growth did not accelerate at all. It flatlined. 

This is what triggered the famous quote from Robert Solow in 1987, that “computers are everywhere except in the productivity statistics.” This statement reflected a new way of thinking about this point in time, called the “productivity paradox.”

It’s as simple as this: Investments in tech went up, productivity stayed flat, especially in the United States. Now over time, we have certainly experienced gains in productivity. 

From making software developers more productive, to the benefits technology is bringing to us in our personal lives, or the software that is powering our vehicles. And many Asian countries have definitely stood to benefit greatly over the past decades. 

But looking back, the home computer enabled us to read, write, and edit our work in a digital form. It started the age of computation. 

The internet then was the vessel that connected humanity together, it allowed us to share our work and communicate digitally. The internet, and then the smartphone, the computer in our pockets, became the foundation for global collaboration. 

And it's the information that we created, edited, and shared together in the last fifty years that is now making artificial intelligence machines possible. Our life is so much better because of computers. They were the change agent for a bigger picture. The strides and leaps we’ve taken in technology were all meant to get us to this point.

Last December, the big bang happened with ChatGPT. This made us realize that machines have become intelligent enough to work with us. To generate for us. Directed by us. 

Recently, Harvard released a new study that found that workers using ChatGPT were 40% more productive with AI. Forty percent is remarkable, think of what would have meant during the industrial age. What would forty percent in an automaker assembly line have meant, still mean today? But it goes further. 

In June 2022, we launched GitHub Copilot, the world’s first at-scale tool that writes code for software developers. You can think of it like an AI assistant that predicts the code developers want. It’s integrated into their environment, so developers don’t have to change their workflow. They type code as usual, and Copilot predicts what the next line is. 

Or multiple lines, boilerplate code, but also complex algorithms.Less than six months on the market, our research found that developers were already fifty-five percent faster when using Copilot. Although it’s not exact, during the industrial revolution in the late 1800s – the steam engine improved productivity by around twenty-two percent. And it foundationally changed our way of life. 

The productivity gains from AI are not only undeniable, they are finally fulfilling the expectations that we had ever since the personal computer appeared on our desks. It is already safe to assume that with AI we are not emerging into another generation of the productivity paradox. 

But rather, we are entering an age of productivity polarity. We live in a fragmented, divided world, with contradicting approaches to quite nearly everything. And it's because of this dissonance, that I believe we will enter an age of productivity polarity between industries, private sector investments and different types of regulation. 

Countries and organizations that figure out how to purpose AI – with reduced risk – will exist in a different spectrum of productivity entirely. A different spectrum of economic growth, entirely. Those who don’t, will fall behind. 

Let’s start with industries. 

Now, it’s clear that there is a boon in investments by enterprise organizations. This is another fact. Fifty percent of all organizations have adopted AI in at least one business area. Let’s crack the hood, what industries are those actually being made in? 

For example, we are already seeing the gains go disproportionately to the tech sector, as you can see in this study issued by Accenture. 

Of course, it makes sense that the technology sector would be first to adopt new technology. If other industries don’t follow, they will sit on the wrong side of the polarity. We must generate a moment of productivity equilibrium. 

I am certain that AI has the potential to light up the global economy in a way we have not seen since the industrial revolution. But for us to achieve this, we must invest in AI to benefit all workers: Blue collar AND white collar. From manufacturers to mathematicians, from traders to technicians. All to accelerate human progress with AI. 

The early evidence of polarity is even starker in private sector investments across countries. 

According to Goldman Sachs, the productivity gains that we will experience with AI could surge the global GDP by 7 trillion US dollars. And, I think this is a conservative number. The reality will be much higher.

At GitHub, our own research found that just AI-powered developer tools ALONE could boost the economy by 1.5 trillion US dollars by 2030. This is just with the AI-powered tools that we have today. And just with that one sector of workers, without any innovation in the next seven years. 

And, it’s private sector investments, new startups turned to global unicorns, that will fuel even more of this economic gain. There are thousands of AI startups out there that could be the next Microsoft, Google, or Tesla. But those giants are all U.S. based. And for now, it's looking like most of the AI innovation is concentrated in Silicon Valley and on the U.S. West Coast again. 

This report from Stanford says it all. As of now, the United States is dominating the world in private sector investments in AI. And while I live in Seattle, I am a German citizen, I deeply love my home country. We are the largest economy in Europe, yet there are little investments in AI. I want Germany to succeed.And I want this for Singapore, Asia, the rest of the world, the rest of the world’s citizens, too. 

But we must activate public-private partnerships, fostering innovation by individuals where new startups can become the GDP-driving tech giants of tomorrow. 

The greatest change happens at the roots. I believe there will be a greater wave of AI startups than during the internet wave in the 1990s or the smartphone wave in 2008. 

But we must nurture a startup ecosystem where founders can innovate from wherever they live. 

And accelerating this individual innovation, goes to the heart of my third point of polarity. Which is regulation. 

In February, I traveled to Brussels to speak out as the EU was drafting the world’s first comprehensive attempt at regulating AI with the AI Act. Policymakers everywhere are taking note. And it’s encouraging: they are focused on reducing the risks from AI so that we can seize its benefits. 

From the EU to the US, Canada, Japan, Brazil and beyond, policymakers are working with humility and seeking diverse input. Striking the right balance in regulation is tough, especially in a world as fast moving and fragmented as ours. 

It’s never been more important to create a set of globally aligned standards. Efforts like the AI Verify toolkit here in Singapore, or the NIST risk management framework in the US, are great examples of how to get it right.

What we don’t want is workers in different countries benefiting from AI at different levels. Imagine, your software developers in the US being fifty five-percent faster, but those in Singapore only twenty-two percent faster due to different regulatory requirements.

This is also why I believe the regulatory compliance burden should not be one size fits all.

Individual researchers and small groups of innovators should NOT be held to the same standards as big international corporations that have large compliance teams. 

So, it is encouraging to see regulatory efforts focus on the high-risk AI models and use cases. And not directly on the community of developers and researchers who sit at the heart of innovation.

Otherwise, we will extinguish individual innovation before the flame is lit. 

You know, polarity is a truly confounding word. There is as much upside as there is downside. 

But you shouldn’t come away from this scared or worried. We are just at the beginning of the Age of AI. We’ve just stepped foot into this new reality. 

When you think about the rise of the home computer and the internet, the companies that have gone on to define our lives, did not reach the stars right away. 

Look at Amazon, and e-commerce: 

The internet was not conceived until 1983. Amazon was founded a year after the first popular internet browser came out in 1993. And Amazon did not have a profitable year until 2003. Twenty years later. And now, here we are. 

There is still so much opportunity for everyone to get it right. Every person. Every organization. Every country can take share in the Age of AI! We just have to get to work. And we have to get to work now. As CEO of GitHub, as a developer, I am dedicated to accelerating human progress through the power of AI. 

I hope you'll join me.Thank you, Singapore! Thank you, Fintech Festival! It’s been an honor. See you next time!