This year’s CogX Festival of AI and Emerging Technology put the focus squarely on the role of innovation to create a better world.
At the start of London’s Tech Week, CogX 2019 set up camp in the regenerated King’s Cross district of London. The event, now in its third year, convenes the artificial intelligence and emerging technology industries. 15,000 visitors gathered to hear talks across a huge range of topics from more than 600 speakers across ten stages.
At the heart of this year’s agenda was a focus on the Global Goals (the seventeen Sustainable Development Goals agreed by the United Nations in 2015) and the need to ensure that our best tools, technology and talent are pointed at the world’s greatest challenges.
Technology won’t solve all the problems we face, but how can it help?
Protecting Our Planet
Climate change is without doubt the most pressing of our problems. For context, Dr Gail Bradbrook, co-founder of Extinction Rebellion took to the stage to paint a desperate picture of ecological collapse, and of economic and societal disintegration following in its wake. “I expect to die,” said Bradbrook, “I expect my children to die. How are we to live in these times?”
Economist Kate Raworth believes that we should redesign our economies for a start. She proposes a model called ‘the doughnut,’ which delineates a safe space in which nations can meet the needs of their citizens while ensuring they don’t overshoot planetary boundaries. “If AI is not in service of this,” said Raworth, “what the hell is it doing?”
If apocalyptic scenarios are not motivation enough, Celine Herweijer, partner in PwC’s Innovation and Sustainability practice, shared research that forecasts a $5.2 trillion opportunity for AI across the agriculture, water, transport and energy sectors alone, as well as a potential 4% reduction in greenhouse gases.
Across the conference, speakers highlighted many of the ways artificial intelligence (AI) is already contributing to the climate fight. Google.org, the philanthropic arm of the tech giant, is funding a project which uses satellite imagery to track, quantify and publicise carbon emissions from power plants. The Small Robot Company is using AI-powered robots to help farmers minimize the use of chemicals and limit carbon emissions. For the near term, Global Thermostat has developed a cutting-edge climate solution that captures carbon from the air, then puts it to commercial use or stores it permanently underground.
Better Lives for All
The Global Goals are also about creating better lives for people: improving wellbeing, tackling inequalities and providing better life outcomes.
Data science is helping by powering better decision-making. Atlas.AI, a social impact tech startup, is using machine learning to gather economic and agricultural data which can help decision makers in the development sector. Its AI platform is using satellite imagery and machine learning to build valuable datasets on prosperity, crop yields and other indicators in sub-Saharan Africa.
Technology is also propelling novel solutions in healthcare. Global estimates suggest that more than 900 million people worldwide suffer from some form of mental disorder. Maja Pantic, professor of Affective and Behavioural Computing at Imperial College shared her work on human-centric AI that can detect and respond to emotion. Pantic has developed a promising facial recognition algorithm that can identify the signs of depression, a condition that is notoriously hard to diagnose. By analysing the patient’s facial expressions, in an experiment it was able to correctly detect depression in 92% of cases.
Thomas Reardon, CEO and co-founder of CTRL-labs, wants to empower all of humanity, making individuals more capable by allowing our brains to talk directly to technology. Rather than connecting tech directly to our noisy brains (like Elon Musk’s NeuraLink for example), their CTRL-kit device uses a comfortable wristband to decode individual motor neuron signals that point to user intentions, allowing the user to control technology with their brain. While the device could help those with neurological conditions regain motor control, Reardon insists that the device is more than therapeutic, saying “If you build a technology for eight billion people, you’re going to solve a ton of problems.”
More Ethical Technology
With much of the conversation on advanced technology now centring on ethics, it’s no surprise that CogX devoted a whole stage to the theme.
CTO at anti-fraud startup Shape Security, Shuman Ghosemajumder led one of the more compelling panels on the Ethics stage, called Deep Fakes and Disinformation, which looked at the threats posed by fake online content. Shuman warned that any platform can be hijacked with fake content, and in fact, most major companies are being hacked and they don’t even know it. What’s more, this fake content is becoming easier to produce due to rapidly improving AI technology. The most troubling revelation came from panellist Giorgio Patrini, CEO of Deeptrace, the Amsterdam-based startup that is developing software to detect fake videos. He explained that deepfakes have introduced significant doubt in our minds, as the sheer possibility of fake content compels us to now question whether real content is real.
Lack of diversity continues to plague tech and much discussion centred on the need for interdisciplinary teams. Astro Teller, who leads ‘moonshots’ at the Alphabet division X that focuses on audacious tech ideas, outlined the added benefits that can come from a collision of diverse mindsets. The design of their high-altitude stratospheric Loon balloons was achieved partly thanks to the help of a fashion designer, while the diplomatic skills of former military veterans are proving useful for retrieving balloons that go astray in some parts of the world.
Perhaps the biggest debate when it comes to AI centres on what might happen if we succeed in creating super intelligent machines. Alan Turing’s view was that we would lose control, according to AI pioneer and UC Berkeley professor Stuart Russell. At the stage named for Turing, he outlined his recommendations for a better form of AI, one that can work for us, rather than against us.
In Russell’s view, humans are incapable of accurately specifying objectives for AI, which can lead to unforeseen consequences when AI gets smarter. In a modern-day version of the King Midas problem, a robot instructed to acquire coffee might do so at all costs, potentially killing anyone who stands in its way or tries to switch it off. Russell advocates introducing elements of uncertainty into the AI’s objectives, which will then encourage the AI to observe and learn more, leading to better outcomes. Russell advocates shifting the focus, from pure intelligence towards a field that is concerned with AI that is “provably beneficial” to humans.