An extraordinary wave of new advances in AI is now underway - from recommendation engines to speech recognition, predictive tools in health to driverless cars, IBM Watson and Google DeepMind to hundreds of start ups. A crucial question is how to maximise the benefits of AI for the public and avoid the biggest risks.

There are strong institutions focused on developing the technology, but very few are attending to the human, social and public dimensions of AI in a serious way. This is where Nesta is most heavily involved, bringing together investment, research, convening and programmes, and using AI as a tool for research.

Our aim is to help technologies develop in ways that enhance humanity and improve our capacity to solve problems - which is far from guaranteed. As well as producing introductory guides and investigating the many dilemmas around law, health, education and ethics, our work falls into the following main categories.

Investment - We have invested in some of the most promising firms using AI to solve problems, including Featurespace (which tracks fraud) and CogBooks (which uses AI to help students learn in adaptive ways).

Using AI for research - We use AI to understand innovation, industrial and economic policy. Web data (for example, text from company websites or online job ads) offer an alternative to official data sources such as business surveys. We use state-of-the-art AI methods and tools including machine learning, text mining, topic modeling and deep learning to extract information from those data, track the growth of new sectors (as we did in Tech Nation 2016 and the video games map), segment online job ads to make the labour market more intelligent (in our Skills Map and in our work in the Economic Statistics Centre of Excellence for the ONS on occupational and skills taxonomies) and map local economies (in Arloesiadur, the innovation data platform for Wales that we are about to launch).

Ultimately, our goal is to transform innovation and economic policy with big, real-time data, AI tools and interactive visualisations

Understanding how AI can shape the future of work - We are using AI and predictive modeling to inform the priorities for reskilling the workforce. Nesta’s early contributions included the 2014 volume looking at the future ‘robot economy’, and our look at the history of automation.

We have used ML techniques to show that creative occupations are significantly more resistant to automation than other jobs, reinforcing our call for UK policymakers to grow the creative workforce.

We have recently published the most ambitious and detailed study of the future of work in the US and UK to date, which considers the full range of long-term trends impacting on the labour market, including population ageing, urbanisation and the greening of the economy, as well as automation and other forms of technological change. Our approach makes use of active learning techniques that optimise expert human insight, and ML algorithms to provide actionable insights. We are now extending the study to other countries and regions.

Regulation - Despite the huge potential impact of AI, regulation has been slow to catch up. Nesta has set out detailed designs for a Machine Intelligence Commission (MIC) to address these issues, guiding behaviours, understanding, norms and rules, with some formal powers of investigation and action, and drawing on experience of other fast changing fields.

In time, standards will also be needed though none currently exist. We have been working with the British Standards Institute (BSI) to investigate where and how standards could be used. Both the MIC and our work with BSI fits into a wider strand of work on the idea of future facing or ‘Anticipatory Regulation’, which we are developing both in terms of ideas and in practical application through the Challenge Prize Centre work with cities on managing drones and autonomous vehicles, which will make substantial use of AI.

AI and data in government - Nesta’s government innovation work has both researched and experimented with the potential of AI in government. This work has mainly focused on local government, with research including our Connected Councils, Wise Council and Local Datavores reports.

We have recently designed ambitious plans with partners to study the state of the art in policy design around the use of AI and algorithms across Europe. Our Offices of Data Analytics Programme has applied AI to help improve decision making and public service delivery at a city or regional scale with pilots in London and Essex using machine learning to analyse historic data on cases of housing violations and modern slavery in order to help predict future ones.

AI in health - Nesta’s Healthlab is heavily involved in digital tools for healthcare and we are currently looking in detail at the roles and dilemmas around AI (for example, the changing roles of doctors and patients as chatbots evolve, or automated tools for assessing mental stages).

AI in art - We are preparing an overview of the impacts of AI on arts and culture, particularly new methods of production, creativity and audience interaction, growing out of our Digital R&D Fund for the Arts.

AI in philanthropy - We’ve explored how AI could be used to help philanthropic foundations make grants, understand patterns and give feedback to applicants.

Books - Nesta staff have written major books on issues related to AI, including my own 'Big Mind' and Stian Westlake’s 'Capitalism without Capital' with Jonathan Haskel (both published late 2017).

Convening conversations - We host many events to discuss AI and its implications, and are also in dialogue with many governments about AI strategies including Estonia, Germany, France, the UK and the UAE. Our FutureFest - a festival bringing together thousands of people to discuss, experience and interrogate the future - has explored many dimensions of future AI, from emotionally intelligent robots to novel hybrids of human and machine, and this will again be a big theme at our next event in 2018.