Independent thinktank Demos has launched an interim report for Research 4.0 which looks at the potentially transformative impact of fourth industrial revolution technologies on the research sector.
The world of research is rapidly changing. The past 20 years have seen huge developments in the way research is conducted, with dramatic changes in tasks across the research lifecycle.
Much of this change is driven by the start of the fourth industrial revolution driven by the rise of the Internet of Things, 3D printing, nanotechnology, biotechnology, 5G, new forms of energy storage and quantum computing.
Artificial intelligence (AI), and deep machine learning in particular, stands out as a class of fourth industrial revolution technologies likely to be impactful. This is due to their position as a general purpose technology.
The changing face of research
Today, artificial intelligence is an increasingly influential part of our daily lives and is developing at a rapid pace. This is no less the case than in the research sector, where we found researchers have developed and deployed a wide range of AI methods to augment their research practices.
Natural language processing (NLP), a technique that makes common use of AI, is the analysis of unstructured language data, essentially enabling computers to extract meaningful information from language as used by humans. Researchers at MIT have developed an NLP tool that can read scientific papers and produce a short summary in plain English. This can be used by researchers to scan a large number of papers and get an understanding of what they say.
Lancaster University researchers have used Tagtog, an AI platform that utilises NLP and machine learning, to annotate and extract information from historical documents that relate to early colonial Mexico. Previously it would take scholars years to fully understand just a small section of these documents.
However, the use of computational techniques such as NLP can allow for much quicker analysis, enabling previously impractical or cost-prohibitive research to be undertaken.
Changing research processes
Technological advances are not only changing the way that research is conducted but also how data is captured, shared and evaluated.
AI-assisted technologies could speed up the bid writing process for researchers. At present these tasks can be extremely time-consuming with significant administrative burdens, taking researchers away from research. Natural language processing mean machines are able to analyse unstructured data, such as written content for bids, and generate content themselves.
They could also dramatically increase the effectiveness of peer review. Today, humans are at the centre of the peer review system for academic papers. The current review process can be incredibly time-consuming. However, an automated system that reviews data standards and other methodologically laborious elements of the review process could free up time for other more qualitative tasks, such as ensuring the research sits in the broader context.
Considering the risks
However, it is critical that the risks posed to the research sector and wider society by the rise of automated research are given proper consideration.
Deep-learning algorithms allow humans to take a step back in analysing information, as computers are given the task of finding meaningful relationships in the data. However, as researchers become more distant from the analysis, they lose understanding of the underlying processes and may not be able to explain exactly what an algorithm is doing; the systems are black-boxed.
More alarmingly, 50% of AI researchers in a 2018 survey forecasted that high-level machine intelligence (HLMI) would be achieved within 45 years.
HLMI is achieved when machines can accomplish every task better and more cheaply than human workers. Reaching that threshold in research tasks has the potential to spark exponential technological progress, with AI systems quickly becoming vastly superior to humans in all tasks. The survey acknowledges the risks associated with this; researchers gave a 15% probability to a bad or extremely bad outcome (eg ‘human extinction’).
From finding new solutions to speeding up research processes, research is already being reshaped around the world by the emergence of a fourth industrial revolution.
However, it is vital to consider how these changes can be harnessed for the good of the research sector and wider society. The second phase of Research 4.0 will explore how this can be achieved, setting out a number of forecasts and policy recommendations for government.