This month The European Commission (EC) released a report on Humans and Societies in the Age of Artificial Intelligence (AI). It is one of the first EC documents with long-term thinking about the impacts of AI and it talks clearly about AGI and safety. We are very pleased to have contributed to this report and believe it makes important contributions towards building a better future with artificial intelligence.
GoodAI’s contribution looked at the future impacts of AI on individuals and society at large. The contribution builds on GoodAI’s research of longer-term impacts, to which apart from Marek Havrda contribute COO Olga Afanasjeva, Communications Manager Will Millership and CEO, CTO Marek Rosa. Below you can find some of our highlights and contributions to the report.
The first part of the report outlines many challenges facing humanity with the advent of more powerful AI systems, and this includes the unintended effects of a race to AGI, a topic GoodAI has written significantly on    .
The report suggests, much like we have outlined, that “there is currently a race for AGI going on around the world with a general belief that the winner would take it all as such a breakthrough would further speed up the research effort. The motivation is therefore immense and there is a genuine risk that safety measures might be bypassed.”
This is a problem that we addressed in our Solving the AI Race round of the General AI Challenge, however, it remains open as to exactly how this “race” can be managed to ensure safe AI development.
The report proposes to “start reflecting about new ways of developing regulations where policymakers, with the support of regulatory scientists, could follow much more closely the development of new technologies and work in parallel at the elaboration of the needed regulations. This effort should be at least EU wide and even worldwide since the deployment of most new technologies has become instantaneous and global.”
The report also highlights the importance of values and well-being and urges the EU to support “the endeavour of UNESCO to develop universal ethical impact assessment principles.”
The implications of AI on work and jobs was another area explored by the report. It referenced a GoodAI piece  exploring the potential of AI to increase inequalities as “jobs in advanced economies involve increasingly unprotected temporary work, freelancing, and one-time gigs. It is quite likely that this trend will accelerate in the coming years with all the negative consequences on human wellbeing.”
This is coupled with the increased “empowerment of individuals who will be able to take advantage of available technology and build new business models…They could use the full potential of AI systems to develop all aspects from product or service design to marketing and after-sales care. Under these conditions, one could for example imagine that three people using a few online gig-workers (human or AI) could create a new service and ensure its smooth delivery for which a medium-size company would be needed today.”
The report puts forward a starting point to launch an investigation to explore several post-work society scenarios and examine their impacts on “wealth concentration or redistribution.” This should include dimensions such as “potential sources for public funds, different ways to provide for people’s basic needs, new forms of social status independent from work, as well as new ways to create a sense of purpose and of self-worth and to better value noneconomic activities.”
Policy-making and impact assessment tools
The work of GoodAI particularly contributed to two of the action points in the report:
- Action 8 – AI-empowered policy-making
- Action 10 – Ex-ante impact assessment tools
With regards to AI-empowered policy-making the idea, put forward by Marek Havrda , is to launch pilot projects that use AI models to deeply predict the impacts of policies on society. In the future, Marek suggested that a new type of experiments could be carried out in collaboration with video-game developers in order to “translate dry technical descriptions of the potential impacts of policy options and provide visual experiences of these impacts on countries, regions, and communities to policymakers as well as to citizens in order to facilitate debate and decision-making.” This may be in particular important for outlining trade-offs and informing related public debate before the decisions are taken.
Furthermore, the report suggested that the work carried out by Marek Havrda and Bogdana Rakova on enhanced well-being assessment  “could serve as one of the sources of inspiration for future work at EU level.” The work suggests creating new impact assessment tools based on rigorous indicators and includes as far as possible also well-being indicators. The assessment tools could be used with testing centers or testing reference sites which would apart from the technical properties of AI systems allow for assessment of the impacts on society. “They [the testing centers] should at least be co-created by public actors as most businesses will have potential conflicts of interest due to the prevailing business models, and engage all stakeholders.”
Takeaways and action points
The report comes up with 10 main takeaways:
- In the age of AI, ‘Intelligence’ will have many different meanings
- We must not underestimate the cultural dimension of AI
- Machines already manage our emotions much better than we do (this is not an opinion of GoodAI)
- Good intentions will not prevent AI from having extremely disruptive effects
- While we talk a lot about privacy, we continue to give our data away
- Freedom of choice is being eroded and liberal democracy and free markets are at risk
- We need to understand if and how the current mental health crisis is linked to the digital transformation
- Creativity will no longer be a human prerogative
- AI risks creating the most unequal societies that ever existed
- Education is our best bet
Once discussing each takeaway in-depth the authors suggest 21 action points grouped under the following themes:
- Strengthening the knowledge base
- Gearing-up policy-making
- Assessing the impacts of AI systems
- Empowering citizens
- Promoting AI based solutions for public good
- Exploring futures
We are delighted to have contributed to this report and pleased to see the European Commission and the European Union putting such great emphasis on these issues and coming up with concrete suggestions for a way forward.
 GoodAI. (2018). Roadmapping the AI race to help ensure safe development of AGI. Retrieved from: https://www.goodai.com/roadmapping-the-ai-race-to-help-ensure-safe-development-of-agi/
 GoodAI. (2017). AI Race Avoidance. Retrieved from: https://www.goodai.com/ai-race-avoidance/
 Rosa, M., Afanasjeva, O., Millership, W. (2017). Report from the AI Race Avoidance Workshop. Retrieved from: https://medium.com/ai-roadmap-institute/report-from-the-ai-race-avoidance-workshop-bd631b2bbb2c
 Afanasjeva, O., et al. (2017). AVOIDING THE PRECIPICE: Race Avoidance in the Development of Artificial General Intelligence. Retrieved from: https://medium.com/ai-roadmap-institute/avoiding-the-precipice-db720a805190
 Havrda, M., Millership, W. (2018). AI and Work a Paradigm Shift? Retrieved from: https://www.goodai.com/ai-and-work-a-paradigm-shift/
 Havrda, M. (2020). Artificial Intelligence’s Role in Community Engagement within the Democratic Process. International Journal of Community Well-Being 3(2):1-5. https://www.researchgate.net/publication/346534223_Artificial_Intelligence’s_Role
 Havrda, M., Rakova, B. (2020). Enhanced well-being assessment as basis for the practical implementation of ethical and rights-based normative principles for AI. In the Proceedings of 2020 IEEE International Conference on Systems, Man and Cybernetics (SMC). https://arxiv.org/abs/2007.14826