Why It’s Now Time to Start Taking Artificial Intelligence Seriously
If government regulation is any indication of how seriously we should take developing technologies, then this month sees a shift in the world of artificial intelligence. In the UK, the House of Commons science and technology committee has recommended that the UK establish a commission on artificial intelligence to provide global leadership on the social, legal and ethical implications of AI, in much the same way as it did in the field of human reproductive technology in the 80’s (MPs call for UK to launch artificial intelligence commission). On this side of the Atlantic, the US government has, through the Executive Office of the President and the National Science and Technology Council, released a report entitled Preparing for the Future of Artificial Intelligence, which makes 23 recommendations as to how the government should respond to AI, including the need for carefully balanced regulation. Accompanied by a strategic plan for federally funded research and development, the White House report aims to summarize both the benefits and opportunities presented by AI, as well as the concerns which the future of AI creates. It looks at AI from three core perspectives:
Economic impact – primarily increased productivity versus job losses;
Fairness, safety and governance – in other words, avoiding unintended consequences of widespread AI adoption; and
Global considerations and security – these speak for themselves if you know what Skynet is. If you don’t, here’s a quick intro.
Practical ideas surfaced in the report by way of recommendations include:
Taking Advantage of AIOpening up government data in order to train AI through the process of machine learning and thereby accelerating the development of AI that addresses social challenges.
Avoiding silos of expertise by pooling the experience and learning of AI “practitioners” across federal government.
Engaging with foreign governments and international organizations on the opportunities and threats of AI.
Identifying milestones that could represent or foreshadow significant leaps in AI capabilities.
Keeping AI developments under review both in the US and elsewhere and reporting regularly to the Administration, especially in the context of milestones.
Ensuring that where the federal government uses AI for decision-making, fairness of outcomes is assured (“AI needs good data. If data is incomplete or biased, AI can exacerbate problems of bias.”)
Emphasizing ethics as a component of educational curricula for AI-related disciplines.
Encouraging effective regulation, with technical expert input, to ensure human safety.
Investigating the effects of AI on the US job market and developing policy responses.
Recognizing that the private sector will dominate AI research, the report recommends federal funding for basic and long-term AI R&D which it anticipates will be largely ignored by the private sector.
Developing actions that ensure an appropriately sized, trained and diverse workforce of AI researchers, specialists and users.
Accounting for the influence of AI on cybersecurity and establishing policy on autonomous weapons systems (which brings us back to Skynet).
The recommendations are to some extent a hodgepodge but it’s good to see governmental recognition of the multiple challenges that lie ahead in terms of AI. Given the diversity of AI technologies and applications currently under development, it’s perhaps not surprising that in identifying the issues, this report seems like a potpourri of related, but only loosely related, ideas.
You can view the report in full here.