Betting on the Future

Technological and social change

Hi, it's Ben from the Tony Blair Institute. Today’s edition dives into GPTs, GPT3, the J-Curve and why we need to take bigger bets on technology.

💝 Like this post? Please share it to help us reach new readers! 

Betting on the Future 

By Benedict Macon-Cooney and Jess Northend 

“The triumph of the industrial arts will advance the cause of civilization more rapidly than its warmest advocates could have hoped.” Charles Babbage 

A tyranny of low expectations pervades much of our politics. On the extremes of the right, a bunkered down mentality has returned, which sees shrinking the pie as the route to national riches, while the far left increasingly promotes retrograde policies as the only way to address the challenges of climate change, migration, and technological transformation.  

This new steady state feels at odds with the modern world. Our scientific and technological capabilities have increased exponentially over the last three decades. We have created inter-connected links around the world and an abundance of information. An ongoing biotech revolution has allowed us to sequence the genome and scale the technology, which helped us to develop a vaccine in days in response to Covid-19. Through breakthroughs in gene-editing and computer science, software is now also the code of life.   

Our run on the hedonic treadmill has also given us cheap consumer goods, on-demand entertainment, and limitless communication. Yet discontent continues. Part of this is a by-product of the democratisation and greater diffusion of information. We all much have a greater understanding of global problems, including the great sword of Damocles that is climate change. And we’re better able to see how institutions are failing in this exponential age, railing against the ills of technology rather than trying to harness its opportunities.  

But if the last decade had a creeping sense of statis and stagnation, there is once again a growing chorus pushing for progress. 

Last year a16z’s Marc Andreessen wrote a calls to arms It’s Time to Build, while more recently, The Atlantic’s Derek Thompson called for an “abundance agenda” and the NYT’s Ezra Klein’s for “supply-side progressivism.” All evoke a similar sentiment and a disquiet with an increasingly inert status quo. As Thompson writes: “One lesson of the Everything Shortage is: You cannot redistribute what isn’t created in the first place.”  

In allo, there are also echoes of Winston Churchill’s invocation in the aftermath of the Second World War that:  

“If we are to bring the broad masses of the people in every land to the table of abundance, it can only be by the tireless improvement of all our means of technical production, and by the diffusion in every form of education of an improved quality to scores of millions of men and women.” 

As the world continues to contend with the fallout of the pandemic, it is a spirit we need today. As Joel Mokyr has written “progress isn’t natural”. It needs advocates and builders, ready to challenge and re-shape our systems and institutions, which often continue to fall short. It needs more people willing to take risks, bet on the future and on the promise of technology. History has shown the rewards of these endeavors, and given the tools we now have at our disposal, the prizes should be bigger than ever before.  

The forthcoming productivity boom 

This is in part a bet on the future, which for many has been a long-time coming. Nearly four decades ago, commenting on the apparent step-change in IT and computing, the economist Robert Solow stated, “[We] can see the computer age everywhere but in the productivity statistics.”  

He asked why, given the potentially transformative promise of computing, was this not feeding through to higher productivity and higher wages?   

Erik Brynjolfsson, Daniel Rock and Chad Syverson recently set out the case that such a boom is about to happen. After nearly two decades of sluggish productivity growth while inflation adjusted pay has flat-lined, we are—they suggest—on the brink of an upswing. The returns on the structural adaptations business have adopted, are set to be realised.  

They describe this process as the Productivity J-curve - the process by which a new General Purpose Technology is accompanied by a slowdown in productivity, before eventually being followed by a sharp increase in productivity growth. 

This is not all technologically driven, but innovations in biomedical science, driven by data, the deep decline in the cost of renewables and batteries are all part of it. And an increasing factor in the acceleration is AI.  

The labour market has started to adapt to the first wave of robotics and AI.. We can see this in warehouses, with a traditionally heavy ratio of manual to cognitive tasks, while Robotic Process Automation is ushering in productivity gains in office roles. COVID-19 has further compressed the timetable of technological adoption and reorganisation by firms. 

Foundation models are now being widely used and are part of the next generation of AI that will change our economy and our workplaces. Earlier this year, OpenAI published a new version of Generative Pre-trained Transformer 3 (GPT3) - a deep learning language model that can now translate human instructions into computer code. It has since been used to write plays, phishing emails, and even write an overview of GPT3 itself.  

AI and Machine Learning will also power the next wave of innovation in biotech, the future of food, autonomous transport, and clean energy and more, presenting profound new possibilities for future abundance. The question is then whether GPT3 can actually become another GPT i.e. a General Purpose Technology.  

GPTs are transformative technologies that have the potential to create radical change in economic models, the way businesses are organised and, in turn, social structures. A handful of characteristics serve to identify a technology as a GPT, including: (1) there are no close substitutes; (2) it has a wide array of applications across industries; (3) it is initially crude, but evolves in complexity over time; (4) it enables a cluster of broader innovations.  

Having had a few false dawns, AI is today rapidly evolved in complexity. It has a wide range of applications beyond its original industry and enables a cluster of broader innovations, including in drug discovery and self-driving cars. There are no close substitutes, it has become pervasive across the economy, and is radically changing the economic environment, the way we interact with technology on a daily basis, and the jobs we do. 

The lessons from history 

This bet on the promise of AI is unprecedented in terms of technological capability.  

But it is also a bet on history. Almost 200 years ago the UK economy went through a similar process of disruption: productivity flatlined, before new technologies were eventually introduced, adopted by businesses and integrated into society, and growth rebounded.  

In 1769 James Watt was granted a patent on his separate condenser - a major step in improving the efficiency of steam engines. Watt’s engine made it possible for steam engines to move beyond removing water from mines, and into the emerging mills and factories of Britain’s Industrial Revolution. By 1800 Watt’s patent had expired and, by 1830 steam had reached parity with water as a source of power.     

The steam engine transformed transport. George Stevenson’s locomotive travelled the span of the Stockton to Darlington railway in 1825; not long after, it was reaching 36 miles per hour on the Liverpool to Manchester railway. Rail was fully embraced as a core part of the British economy, with total rail miles increasing to over 400 million by 1910, up from 60 million per year in the early 1850s.   

The steam engine was one of the first truly transformative technologies of the modern age. It contributed to the development of Britain’s cities, the industrialisation of commerce, and enabled people and goods to travel further, faster. Yet, in the 1760s – when James Watt was first working with the technology of steam engines – there was little indication that his experiments would lead to the wholescale reorganisation of the British economy. Linking back to the defining characteristics of GPTs, it was initially crude, but its impact grew over time, enabling a cluster of broader innovations, across industries.  

It took the best part of 100 years for businesses to reorganise their ways of working, and for the economy to adjust to the massive economic change that steam had enabled. Indeed, it was only after that century-long delay that the benefits of steam started to be seen in the economic data, after a long stalling of productivity growth and wage growth.

Productivity and wages did eventually rebound but, in the meantime, the challenges for working people compelled a generation of social reformers to intervene. Their concerns: how to create safeguards and decent working conditions, how to create a social security system fit for the changing economy, and how to guarantee the rights of workers. These questions dominated domestic discourse for over a century.  

Even before AI, the technological revolution was the central question of our time. Over the last few decades we have built a whole new infrastructure for the world. But the breakthroughs we will witness in the coming decade will mean any linear projections of the world today will always fail.  

AI will provide new parameters and variables that can help us reshape everything around us. It should provide a far more symbiotic relationship with computers, provide us with more tools to augment, improve and hopefully provide abundance where this is otherwise lacking. 

Opportunities for progressives 

However, technology alone will not automatically cure regional disparity, low productivity, and sluggish wage growth. The lessons from Erik, Daniel and Chad’s paper—and indeed from technological shifts of the past—is that the productivity boom will come eventually, but there are choices in how we engage with it, how rapidly it is delivered, and who benefits from it.   

Industrial era-social reformers moulded the contours of political debate in the 20th century: globalism and nationalism; tax and spend; public and private. For progressives who see technology as a way to scale clean energy, world-class education, biotech, electric vehicles, flying cars and even manufacture next generation materials in space, we need a new theory of state that is a platform for opportunity.  

Next generation institutions  

In particular, we need new infrastructure oriented around the opportunities that tech presents today. Take the development of data architecture for biomedical research. Biobanks and bioreactors offer the promise of helping deliver a new generation of personalised, precision medicine, and their stores of specimens and tissue samples are unlocking new, data-driven research programmes. But building wider bio-infrastructure for the 21st C that includes labs, manufacturing, distribution channels, digital devices, home sequencing and data will be critical to accelerate the democratisation of health and build next-generation drugs and diagnostics. AI will be central to unlocking the vastness of this ideal.  

In a similar vein, our cities are increasingly connected, with data from smart infrastructure, wearables and transport networks creating mass, anonymised insights that can be used to improve urban living - from congestion, to air quality, to personal health. Driven by advances in AI and ML, the traditional institution of city government will need to adapt to a new role, with an active approach to technology governance, cyber security and interoperability. A global network of digital twins - enabling cities globally to share data with public and private actors around logistics, climate, and energy - is another example of the kind of next generation infrastructure that’ll be key to delivering on the promise of technology.   

Networks as propagators 

This new infrastructure shouldn’t all be down to governments, as the growth of new networks that operate beyond the reach of existing institutions continues. Blockchain, web 3, crypto, non-fungible tokens (NFTs) and decentralised autonomous organisations (DAOs) are at the heart of this trend towards decentralisation. Only recently a biopharma research project was funded via an NFT for the first time. A DAO now owns the IP to this work, giving interested parties a new route to back frontier research. This approach offers researchers and labs an entirely new way to access research funds and transparency over access / use of early-stage research. In the case of pharma research, innovative network-based funding offers a new approach to sharing risk and moving basic research beyond the ‘valley of death’ and into clinical trials. These types of innovation pose new questions for governments and traditional gatekeepers about when / where to give up control to encourage innovation. While network-based organisations may seem like a threat to centralised authority, they offer new models of co-operation though which people around the globe can work together in the common interest.  

Powering the diffusion engine  

One of the main features of the UK’s productivity challenge over the past decade and a half has been the major gap between ‘the best’ British firms and ‘the rest’. While every region and sector has highly productive companies, the UK has a larger tail of less productive firms. As has been argued, the UK’s diffusion engine has seized up. Despite ranking consistently in the top five for innovation, the UK ranks only 38th globally for knowledge diffusion. 

Building out a network to support technology transfer should be a priority for progressives. It’s an essential part of ensuring that the benefits of innovation are broadly spread across the country. Options proposed by Sir Charlie Mayfield, Andy Haldane and others include strengthening tech transfer via supply chains, creating ‘digital twins’ for businesses to show the impact of introducing new tech, and supporting management and leadership training in firms that wouldn’t normally seek it. The scale and scope of these mechanisms will change over the coming decades, but their importance will only increase. 

An adaptive approach to leading through the technological revolution 

The speed and depth of innovation happening simultaneously in several fields is unprecedented. This means there isn’t a mental map or instruction set for leaders - in public life or in industry - to follow. While some of the challenges we need to work through will be technical (solved by applying existing expertise or approaches that have worked before), many will be adaptive in nature - requiring us to quickly test new interventions and approaches. The next generation of progressive leaders will need to be frank about the trade-offs - often to jobs, identities and industries - and hold steady as we collectively navigate this future.      

--- 

The technology revolution is radically reshaping society. The question is whether this radical revolution warrants a radical political response and whether we believe that it can be shaped to benefit not just a small elite, but to help unpick some of the big challenges facing our generation. The shifts in technology we’re living through will need a complete realignment - not just in how government provides services, but in how we plan for the future. 

We will need to develop new institutions to help guide this change, coupled with a highly adaptive approach to navigating the policy choices we face. We need to demonstrate the promise of the next wave of innovation to people and their communities - in health, education, transport, and getting in and on at work. Our political ambition for technology should match the scale of this promise - not just in tackling the challenges and diseases of today, but in framing the possibility for the next generation.   

New political parties and movements have been borne out of technological change before. They need to again, with progressives establishing a new vision that harnesses science, innovation and technology for good. The future is already here; the real question is how we choose to meet it.