The AI of 5-10 years time could be very different from today’s AI. The most successful AI systems of that time will not simply be extensions of today’s deep neural networks. Instead, they are likely to include significant conceptual breakthroughs or other game-changing innovations.
That was the argument I made in a presentation on Thursday to the Global Data Sciences and Artificial Intelligence meetup. The chair of that meetup, Pramod Kunji, kindly recorded the presentation.
You can see my opening remarks in this video:
A copy of my slides can be accessed on Slideshare.
The ideas in this presentation raise many important questions, for which there are, as yet, only incomplete answers.
Indeed, the future of AI is a massive topic, touching nearly every area of human life. The greater the possibility that AI will experience cascading improvements in capability, the greater the urgency of exploring these scenarios in advance. In other words, the greater the need to set aside hype and predetermined ideas, in order to assess matters objectively and with an independent mind.
For that reason, I’ve joined with Rohit Talwar of Fast Future and Ben Goertzel of SingularityNET in a project to commission and edit chapters in a forthcoming book, “The Future of AI: Pathways to Artificial General Intelligence”.
We’re asking AI researchers, practitioners, analysts, commentators, policy makers, investors, futurists, economists, and writers from around the world, to submit chapters of up to 1,000 words, by the deadline of 15th September, that address one or more of the following themes:
- Capability, Applications, and Impacts
- How might the capabilities of AI systems evolve in the years ahead?
- What can we anticipate about the potential evolution from today’s AI to AGI and beyond, in which software systems will match or exceed human cognitive abilities in every domain of thought?
- What possible scenarios for the emergence of significantly more powerful AI deserve the most attention?
- What new economic concepts, business models, and intellectual property ownership frameworks might be enabled and required as a result of advances that help us transition from today’s AI to AGI?
- Pathways to AGI
- What incremental steps might help drive practical commercial and humanitarian AI applications in the direction of AGI?
- What practical ideas and experiences can be derived from real-world applications of technologies like transfer learning, unsupervised and reinforcement learning, and lifelong learning?
- What are the opportunities and potential for “narrow AGI” applications that bring increasing levels of AGI to bear within specific vertical markets and application areas?
- Societal Readiness
- How can we raise society-wide awareness and understanding of the underlying technologies and their capabilities?
- How can governments, businesses, educators, civil society organizations, and individuals prepare for the range of possible impacts and implications?
- What other actions might be taken by individuals, by local groups, by individual countries, by non-governmental organizations (NGOs), by businesses, and by international institutions, to help ensure positive outcomes with advanced AI? How might we reach agreement on what constitutes a positive societal outcome in the context of AI and AGI?
- Governance
- How might societal ethical frameworks need to evolve to cope with the new challenges and opportunities that AGI is likely to bring?
- What preparations can be made, at the present time, for the introduction and updating of legal and political systems to govern the development and deployment of AGI?
For more details of this new book, the process by which chapters will be selected, and processing fees that may apply, click here.
I’m very much looking forward to the insights that will arise – and to the critical new questions that will no doubt arise along the way.
dw2 
Leave a comment