How the DIF showed the positive side of AI
Artificial Intelligence (AI) has one again been thrown into the public limelight following the release of the Malicious Use of Artificial Intelligence report. The 100-page document, published by a US-based group of experts including OpenAI, Electronic Frontier Foundation, and Center For A New American Security, details the potential for exploitation of AI by criminals, and nefarious international actors.
Concerns about the potential negative impacts of AI predate even the technologies themselves, and there are of course plenty of legitimate reasons to be concerned.
The impact of technology on people and the economy, and fundamentally what it means to be human in an "age of automation", were key topics throughout the 2017 edition of the Disruptive Innovation Festival (DIF). In light of the latest headlines, here are five (mostly) positive things we learned about AI during the DIF last November.
1. Machines can know things without us ever having told them how to do it
In How Humans Can Thrive When Machines Are Smarter Than Us, technology journalist David Weinberger highlighted the remarkable transition in computing moving away from just building computers that we programme using logical models, to machines that can figure out new models based only on the raw data supplied to them.
On the face of it, this paves the way for a Terminator-style scenario, but Weinberger was quick to point out that so far its uses have mostly been quite valuable for people, including the diagnosing of diseases, navigation, weather prediction and much more.
2. Experts want robots to be able to explain why they did something (and not how they did it)
How might it be ensured that advanced intelligence in robots is developed safely? Dr Subramanian Ramamoorthy, Informatics professor at the University of Edinburgh argues for the principle that an intelligent system can explain to you why it did something as an underpinning criteria for developing AI that works alongside people in a thriving society and economy, and doesn't cause harm.
Critically, this is different from the ability to express the algorithm or code that it used to make the decision in a way that humans can understand, which becomes increasingly difficult with the level of computing complexity today.
3. The real dangers of automation may not have featured in a Hollywood film
Internet giants are utilising access to massive amounts of data and powerful algorithms to "corner culture and undermine democracy", according to Jonathan Taplin, author of Move Fast and Break Things. During his DIF session, he highlighted the dangers of a global digital economy that he suggests has been monopolised by its three largest companies - Amazon, Facebook and Google.
Taplin's DIF session was not without hope and optimism, but he does paint a somewhat stark picture and a modern vision of the kinds of risks that automation and new technology may truly pose to society.
4. Automation is a very different story in low-to-middle income countries
Is automation a threat or an opportunity for developing nations? So far, examinations of the impacts of automation, especially societal analysis, has focused on a "western paradigm", argued Daniel Riveong on the final day of the 2017 DIF. The potential implications of AI technology on jobs, growth, entrepreneurship and development scenarios are unsurprisingly wide ranging.
The answer to the question is of course that automation can be both a threat and an opportunity in this different context. Riveong doesn't stray too much into the area of trying to predict the future, but acts as a starting point for a broader discussion of the impacts of automation.
5. Why we need to embrace complexity, and how machine learning does it already
Our machines have knowledge that we'll never understand, and we should embrace that argues Weinberger. "We now have the lens and the ability to live in chaos and acknowledge it in a way that we didn't before the internet and machine learning".
He went on to argue that the reality of the world may actually be closer to the way our machines perceive it than humans. Watch his conversation with the DIF's Seb Read and Colin Webster in full here.