Artificial Intelligence (AI) has the potential to bring immense opportunities, but it also poses challenges.
AI is currently dominating the research and development agenda of the leading Internet industry. Echoing the buzz from the Silicon Valley, Accenture analysts believe that in some countries artificial intelligence could even double annual economic growth rates by 2035.
AI isn’t new, but its recent acceleration can be explained by its convergence with big data and the Internet of things (IoT), and the endless applications and services it allows. In the market, this translates into investments across all industries as stakeholders try to maximize the potential of AI for their own businesses. For instance, at the beginning of the year, Ford motors announced a plan to invest $1 billion over the next five years in Argo AI, an artificial intelligence startup that is focused on developing autonomous vehicle technology. It's an indication that AI is a hot topic beyond the traditional ICT sector.
The issue has also come at the top of policymakers’ agenda including at the G20, the ITU and the OECD where leaders gathered last week.
How our community feels about AI
There is a growing expectation on the part of many individuals that AI and machine learning will fundamentally reshape the future of the Internet and society around it.
This is one of the trends we’ve observed in The Internet Society’s project about the Internet’s Future, where AI, together with five other areas, have been identified as key “Drivers” of change in the coming 5 to 10 years. We’ve interviewed thousands of stakeholders and there is a sense that we may be experiencing a new technology Renaissance. Indeed, in ten years' time AI technologies could dominate all aspects of our day to day lives from driving to banking or even working.
Yet, the uncertainties raised by our community about this technology are extensive. These include the potential loss of human agency and decision-making, lack of transparency in how algorithms make decisions, discrimination, the pace of technological change outstripping governance and policy, and ethical considerations. A number of interviewees raised concerns related to the impact on industry and employment – and therefore society – noting the consequences of automation-led change across industries and business practices, and the possible increase in inequalities and societal disruption.
Will AI replace human labour?
Last week, the discussions at the OECD revolved around a specific issue: Will AI replace human labour?
What do humans do at work? They perceive their environment, learn, use language to communicate, plan and navigate tasks – all of them abilities that can be imitated to varying degrees by machines.
Looking back at the history of AI, the concept was born when a group of visionary researchers, including Marvin Minsky and John McCarthy, gathered in the summer of 1956 at Dartmouth College to create computers programmed to act as humans. So, the risk – or opportunity – was embedded, although perhaps not consciously, in the group’s objectives: replicating human intelligence.
Is it realistic that we could all be replaced by robots and algorithms?
It depends who you ask and how you analyse the challenge. So far the estimated impact on job displacement has had a broad range, somewhere between 9 and 47%. From the OECD to the University of Oxford, the measuring techniques are quite different. The numbers are alarming and should be taken seriously, but they also do not tell the whole story and should be used carefully.
Shaping a future we can look forward to
These fears are natural, but should be put into perspective. Lets think about how AI could improve human performance and lives.
Over the past few years, deep learning has made tremendous progress. See the example of the Go Game guru, Lee Sedol, who was defeated by the “Alpha Go” machine. He explained that beyond personal disappointment, he also experienced a positive feedback loop. Lee learned from AI Go patterns and techniques and raised his own performance level. This story shows that AI performing at the level, or higher, than humans is not necessarily a threat – it can augment intelligence and support human development.
AI can also have a positive effect on humanity notably by drawing inferences from enormous sets of data. For example, in the pharmaceutical field, the combination of AI and big data expands the industry’s ability to solve new scales of problems, which in turn enables the acceleration of research and can bring major breakthroughs in drug discoveries and disease diagnoses.
Energy efficient homes, personal assistants that make our lives easier, etc. There are many other reasons and fields where hope – and even excitement – is possible.
Going forward it is important that we broaden and demystify the debate in order to balance the headlines with insights and facts. To this end, The Internet Society recently published a Policy Paper on Artificial Intelligence and Machine Learning, introducing the fundamentals of the technology at hand and some of the key challenges it presents.
As one of our guiding principles from this paper clearly states: "The public’s ability to understand AI-enabled services, and how they work, is key to ensuring trust in the technology."