The Turing Test Is Bad For Business

WIRED IDEAS Daron Acemoglu, an Institute Professor at MIT, is the author of five books. Five books have been written by him, including the New York Times bestseller Why Nations fail and The Narrow Corridor, States, Societies and the Fate of Liberty (both co-authored with James A. Robinson). Michael I. Jordan is the Pehongchen Distinguished Professor at the Department of EECS, and the Department of Statistics of the University of California. His research interests include optimization, machine learning, and control theory. E. Glen Weyl, Founder of the RadicalxChange Foundation Microsoft's Office of the Chief Technology Officer Social Economist and Social Technonologist (OCTOPEST), and co-author of Radical Markets : Uprooting Capitalism & Democracy for a Just Society with Eric Posner.
The news is filled with fears about Artificial Intelligence: job loss, inequality, discrimination or even a superintelligence ruling the world. Everyone assumes that business will be the one to benefit, but data suggests otherwise. Despite all the hype, US companies have been slow to adopt the most advanced AI technologies. There is also little evidence that these technologies are increasing productivity or creating new jobs.

This is not due to AI technology's relative infancy. This is due to a fundamental mismatch in how AI technology is being conceptualized by many in the technology industry. It's also a result of the need for business and how AI is currently being designed. This mismatch can be traced back to Alan Turing’s groundbreaking 1950 "imitation game" paper, and the Turing test that he suggested therein.

The Turing test is a way to define machine intelligence. It involves imagining a computer program capable of imitating a human in a text conversation.

This was, at best, one way to describe machine intelligence. Turing and other pioneers of technology like Norbert Wiener and Douglas Engelbart understood that computers were most effective for business and society when they complement and enhance human capabilities. They should not compete with humans. Such complementary forms of information technology include databases, spreadsheets, search engines, and spreadsheets. Although they have had a tremendous impact on businesses, they aren't often called "AI" and the success stories that they represent have been diluted by a desire for something more "intelligent." Although this yearning is not well defined and there has been little effort to create an alternative vision, it has become a desire to surpass human performance in tasks like vision and speech. This view has been dominant in both public and private discussion as well as in capital investment in AI.

Social scientists and economists emphasize the fact that intelligence does not arise only in individuals, or even primarily in them, but also in groups such as markets, education systems, and cultures. Technology can play two important roles in supporting collective intelligence. As Douglas Engelbart's 1960s pioneering research and subsequent development of the field human-computer interaction show, technology can increase the participation of individuals in collective intelligence by giving them information, insights and interactive tools. Technology can also create new types of collectives. This last possibility has the greatest potential to transform society. This alternative frame for AI has major implications for human welfare and economic productivity.

Scale is possible when businesses are able to successfully split labor and bring different skill sets together in teams that create new products or services. Markets are successful when they bring together different sets of participants. This facilitates specialization to increase overall productivity and social welfare. Adam Smith knew this well more than two centuries ago. Technology should not be focused on imitation but the complementarity, as Smith intended.

There are many examples of machines that enhance productivity by performing tasks which complement those performed by humans. These include the huge calculations that support everything, from financial markets to logistics. They also enable the transmission of high-fidelity images over long distances in a flash of an eye and the sorting of reams information to find the relevant items.

Computers can do much more than just execute code lines written by humans. Computers can learn from data, and can interact with, infer and intervene in real-world issues, alongside humans. This breakthrough should not be seen as a chance to transform machines into silicon-based versions of humans. Instead, let's focus on how computers can make use of data and machine learning to create new markets, new services and new ways to connect people in economically profitable ways.

Recommendations systems are an early example of economics-aware machine-learning. This innovative form of data analysis gained prominence in the 1990s at companies like Amazon ("You may also enjoy") and Netflix ("Top picks"). Since then, recommendation systems have been ubiquitous and had a major impact on productivity. They connect people to products by leveraging the collective wisdom of the crowd.

SIGN UP Subscribe to WIRED to stay informed with your favorite Ideas writers.

Machine learning is being used to create direct connections between listeners and musicians, writers and readers, game creators, and players. This space has seen early innovators such as Uber, Shopify, YouTube and YouTube. The term "creator economy", which is growing in popularity, is being used. These collectives are markets. Economic value is associated with the connections among their participants. This is a key feature of these collectives. It is necessary to research how machine learning, sociology, and economics can be combined to create healthy markets that generate sustainable income for participants.

This innovative use of machine-learning can also help to strengthen democratic institutions. The digital ministry of Taiwan uses online participation and statistical analysis to increase the quality of deliberative discussions that can lead to team decisionmaking in the most well managed companies.