General Purpose Technologies of the Future

AI, Biotechnology and Nanotechnology are the technologies most likely to develop into GPTs in the future.

It is important to identify and assess the likely impact of potential future General Purpose Technologies (GPTs) due to their ability to affect productivity, reshape society and disrupt existing businesses. The most likely GPT candidates are biotechnology, nanotechnology and Artificial Intelligence (AI).

Criteria

The criteria used to identify potential GPTs (as discussed in General Purpose Technologies) are:

• A single, recognisable technology

• The technology initially has significant scope for improvement

• Pervasive (eventually used across many sectors)

• Creates many spill over benefits

The criteria used to identify GPTs are subjective though, leading to debate about what technologies actually qualify as GPTs. These criteria try to identify technologies that improve over time and become broadly use across the economy, creating wide spread spill over benefits so that aggregate productivity is boosted.

Candidates

Using these criteria, a list of candidate technologies can be narrowed down to those most likely be considered GPTs in the future.

AI

• A single, recognisable technology - Yes

• Significant scope for improvement - Yes

• Pervasive - Yes

• Many spill over benefits - Yes

Nanotechnology

• A single, recognisable technology – Debatable, as it refers to any technology operating below a certain scale

• Significant scope for improvement - Yes

• Pervasive – Yes, if the technology improves sufficiently

• Many spill over benefits – Yes, if the technology becomes widely used

Biotechnology

• A single, recognisable technology – Yes, it is a broad categorisation but this is not dissimilar to computers as a GPT

• Significant scope for improvement - Yes

• Pervasive – Mainly used in healthcare at the moment but biotechnology has the potential to spread into areas like agriculture and manufacturing

• Many spill over benefits – Yes, if the technology becomes widely used

3D Printing

• A single, recognisable technology - Yes

• Significant scope for improvement - Yes

• Pervasive - No, 3D printing is a niche manufacturing technology that is unlikely to spread beyond use cases with relatively small batch sizes

• Many spill over benefits - No, while it can enable the restructuring of supply chains and new types of products the spillover benefits are likely to be fairly small

Robotics

• A single, recognisable technology – No, robotics is difficult to separate from computers, AI and other machinery

• Significant scope for improvement - Yes

• Pervasive (eventually used across many sectors) – Likely to be widely used eventually

• Many spill over benefits – Yes

Blockchain

• A single, recognisable technology – Yes, although software is generally not treated as a GPT

• Significant scope for improvement – Yes, but so far improvements have been incremental

• Pervasive – Not as yet but potentially could be, although this is dependent on the technology improving and would require significant social and institutional changes.

• Many spill over benefits – Yes, if the technology becomes widely used

Clean Energy

• A single, recognisable technology – No, there are a range of technologies used to harness clean energy and clean energy is largely a subset of electricity

• Significant scope for improvement - Yes

• Pervasive - Yes, electricity from clean sources is used widely

• Many spill over benefits – No, electricity is already pervasive and low-cost. There will be environmental benefits but not the type of productivity gains typical of a GPT.

Reusable rockets

• A single, recognisable technology - Yes

• Significant scope for improvement - Yes

• Pervasive - Unlikely

• Many spill over benefits – Unlikely

Battery systems

• A single, recognisable technology - Yes

• Significant scope for improvement - Yes

• Pervasive - Yes

• Many spill over benefits – Questionable as batteries have enabled a broad range of technologies already but are generally not considered a GPT. This may be because batteries are considered an extension of electricity.

Artificial Intelligence

AI is a somewhat loosely defined technology, and is often considered to be any software that can perform tasks that have historically been difficult for computers. As a result, the goal posts of what is considered AI tends to move as the capabilities of algorithms improve. AI is also a misleading term as it tends to convey a sense that there is something more advanced than basic mathematical operations occurring. The computer simply optimises parameters in an algorithm based on some objective function with no semantic understanding of the problem it is solving.

Dramatic progress has been made in a number of fields (language and vision) in recent years due to neural networks, large datasets and GPUs. This has created a surge in interest in AI and in particular, neural networks. While the name may imply similarities with the function of the brain, neural networks are again based on simple mathematical operations.

Large data sets have been a key enabler of advances in AI, particularly for large neural networks which continue to improve with data volume. Hardware improvements, particularly GPUs, which allow highly parallel processing have also been critical to recent advances in AI. Continued improvements in hardware (potentially including neuromorphic computers, quantum computers, ASICs and FPGAs) and algorithms along with access to larger and more varied data sets have the potential to significantly improve the capabilities of AI.

AI’s potential as a GPT largely comes from its ability to automate what previously may have been considered non-routine tasks. So far many applications of AI have been incremental improvements or novelty use cases, but in the near-term this could begin to change. Autonomous vehicles are one application where AI could dramatically impact society and further deployment of AI in call centres has the potential to eliminate many jobs. As robotic process automation improves it is also likely to cause significant productivity gains due to its ability to automate tasks performed by knowledge workers.

AI is likely to take a long time to fully diffuse through the economy though as each application generally requires its own data for training and this data can be costly and time consuming to collect. The more generalised AI becomes, the less this will be a problem though.

Nanotechnology

Nanotechnology refers to the use of matter on an atomic, molecular or supramolecular scale. It is a category that is inclusive of all technologies that operate at a scale less than 100 nanometers. The ability to manipulate structures on a nanoscale provides new material properties, such as greater reactivity, unusual electrical properties, enormous unit strength, self-healing, self-cleaning, memory (revert shape) and piezoelectricity (convert pressure into energy).

Commercial applications are currently limited to nanomaterials, like graphene and carbon nanotubes. Graphene is a two-dimensional honeycomb lattice made of carbon atoms that is considered useful due to its exceptionally high tensile strength, electrical conductivity and transparency. Carbon nanotubes, as the name suggests, are tubes made of carbon with diameters typically measured in nanometers. Carbon nanotubes have unusual electrical properties, exceptional tensile strength and high thermal conductivity. These properties make carbon nanotubes suited to applications like electronics, optics and use in composite materials. Nanomaterials are likely to find use in diverse applications including medicine, manufacturing, fuel cell electrocatalysis, organic photovoltaic cells, low-friction coatings, high-strength composites, new types of displays and super-efficient batteries.

There is also the potential to develop nanoscale machines or robots, although this has so far proven difficult. This is in part due to the fact that most useful structures require complex and thermodynamically unlikely arrangements of atoms. In general it is difficult to build atomic scale devices as atoms of comparable size and stickiness must be positioned together. Self-assembly based on molecular recognition occurs in biology though, so it is a question of whether these principles can be used to engineer new constructs in addition to natural ones.

Biotechnology

Biotechnology is being driven by an increasing ability to understand and engineer biology. The use of biotechnology is currently limited to healthcare and to a lesser extent agriculture though. Medicines are generally not considered GPTs as their impact is fairly narrow, meaning that even if biotechnology resulted in cures for diseases like cancer and Alzheimer’s or a treatment that slowed aging, it would not be considered a GPT unless it was deployed more broadly.

Recent advances include dramatically lower DNA sequencing costs, advances in multi-omics and the emergence of new techniques to edit genes and reprogram cells. McKinsey has grouped biotechnology innovations into four categories:

• Biomolecules - the mapping, measuring, and engineering of molecules

• Biosystems - the engineering of cells, tissues and organs

• Biomachines - the interface between biology and machines

• Biocomputing - the use of cells or molecules such as DNA for computation

Biotechnology has the potential to be adopted in areas like agriculture, healthcare, IT and manufacturing as the technology develops. Biology could be used to produce up to 60% of the physical inputs to the global economy. Approximately one-third of these inputs are biological, like wood or livestock, and the other two-thirds are non-biological but potentially could be produced or substituted using biotechnology. The ability to engineer organisms also creates the potential to improve disease prevention as well as agricultural productivity. The feasibility of interfaces between computers and biological systems is also improving. This could eventually result in brain-computer interfaces or utilising DNA to store data.

The adoption of biotechnology is heavily dependent on societal attitudes and the regulatory landscape. Without a social license to operate biotechnology is unlikely to live up to its potential, as indicated by pushback against GMOs in agriculture and hesitancy regarding mRNA vaccines.

Conclusion

AI, biotechnology and nanotechnology are three potential GPTs of the future. Although biotechnology and AI are not new technologies, their use is still narrow relative to their potential and they both have significant potential for further improvement. The future importance of AI seems relatively assured at this stage but biotechnology and in particular nanotechnology still need to improve significantly before they become broadly adopted.