Beyond self-driving cars and Google Translate software, there is much talk about how artificial intelligence (AI) technologies can be used to transform society for the better. A conference in Geneva this week is examining the pros and cons of AI, and the way forward.
Some 500 policymakers, academics and executives are gathered in Geneva this week for the inaugural “AI for Good Global Summitexternal link”, co-organized by the International Telecommunications Union (ITU) and the Xprize Foundationexternal link, a Silicon Valley non-profit group.
The conference has a grand title and a lofty aim – “to chart a course for AI that will benefit all of humanity”. The UN, which is represented by 20 agencies such as the children’s fund UNICEF, wants to refocus AI on sustainable developmentexternal link and see how it can contribute to global efforts to eliminate poverty and hunger, and to protect the environment.
“Artificial Intelligence has the potential to accelerate towards a dignified life, in peace and prosperity, for all people,” UN Secretary-General António Guterres told event attendees in a video address. “The time has arrived for all of us – governments, industry and civil society – to consider how AI will affect our future.”
AI has been steadily grabbing the headlines. Last month, a Google AI programme defeated a Chinese grand master at the ancient board game Go. Unbeknownst to many people, forms of AI are already transforming daily lives.
“It’s all around us. It’s not the future but is already happening,” declared Jürgen Schmidhuberexternal link, scientific director of the Swiss AI laboratory, whose researchers have developed software for companies like Apple, Google, Microsoft and Amazon. “All of us have AI in our pockets when we use our smartphones to do speech recognition.”
Yoshua Bengioexternal link, a computer scientist from the University of Montreal, acknowledged that much progress had been made in the areas of speech recognition, and the ability of computers to recognize objects and use and understand human language.
“More researchers are studying this area, so we are making rapid progress,” he declared. “Right now there is low-hanging fruit like using Al for medical image classification, or detecting cancer cells from video.”
Facebookexternal link, for example, is mapping the world’s population by using image-recognition software to read satellite images for signs of human habitation. IBM is using AI techniques to predict pollution levelsexternal link and scientists at Stanford Universityexternal link are employing AI and satellite remote-sensing data to predict crop yields months ahead of harvest.
David Salomão, a software engineer who works at The National Communications Institute of Mozambique (INCM), has a similar idea.
“We're trying to combine mobile telecommunications with artificial intelligence to support agriculture,” he explained. “The idea is to support farmers with automated irrigation systems that decide when to turn off or on and that can assess the moisture of the soil.”
In her conference speech in Geneva, former World Health Organization chief Margaret Chan described AI as the “new frontier” for the health sector.
“Health information is often messy and poorly structured,” she said. “It’s not systematically analysed. AI can give data structure, speed up the reading of results from electrocardiograms and give more precise predictions. This can cut down on healthcare costs.”
Chan said that AI has massive potential but urged caution.
“Medical decisions are complex, and rely on care and compassion. I doubt that a machine will ever be able to imitate genuine human compassion.”
She pointed out that developing countries have little use for a smart system to diagnose illness if there are no doctors and nurses, clean running water and electricity or medicines.
Best or worst thing ever
The future of AI
A new surveyexternal link of machine learning forecasts that AI will outperform humans in many activities in the next decade, such as translating languages (by 2024), writing high-school level essays (by 2026), driving a truck (by 2027), working in retail (by 2031), writing a bestselling book (by 2049), and working as a surgeon (by 2053). Researchers believe there is a 50% chance of AI outperforming humans in all tasks in 45 years and of automating all human jobs in 120 years, with Asian respondents expecting to reach these dates much sooner than North Americans.
Last year, Professor Stephen Hawking warned that the creation of powerful AI would be “either the best, or the worst thing, ever to happen to humanity”.
Ensuring that AI is used to redistribute wealth and reduce inequalities between states is just one of many huge challenges for policymakers considering how to govern AI at the global level while technological development accelerates.
“Discussion on AI is at the fork in the road,” Salil Shettyexternal link, secretary general of Amnesty International, told the audience.
He described AI’s potential for “huge economic disruption” leading to hundreds of millions of jobs lost to automation. He also warned of the risks for democracy and elections as seen in recent reports of the impact of AI and big data on the US elections and Brexit.
“In the future, robot cops will patrol our streets and wars will be fought by killer robots,” Shetty said. “Will AI systems become the gatekeepers deciding who can access healthcare and who qualifies for a job or mortgages? Those in power who have access to the fruits of the data economy, a handful of countries and companies, will be those who continue to gain while the vast majority of people will be left behind…we can do better.”
On Wednesday, Shetty announced that Amnesty International had launched an “AI and Human Rights Initiative” to “enshrine the protection of human rights in the use and development of AI”. It joins a host of other global AI initiatives, such as the Partnership on AIexternal link, which includes Google, Microsoft and Human Rights Watch, or OpenAI,external link a non-profit research company launched in 2015 by philanthropists and entrepreneurs to develop “safe” AI systems.
The UN is throwing its hat into the ring as well; ITU Secretary-General Houlin Zhao told the conference that Secretary General Guterres had proposed to start a special advisory committee for AI. The ITU plans to hold informal consultations to see how the UN can work together with industry and academics to promote AI innovation and create a good environment for its development.
All sides seem to agree that there is a pressing need for a clear roadmap, but some observers warn about giving too much power to the private sector.
“The primary danger of the AI for Good Global Summit is not that it distorts perceptions of AI risk; it is that Silicon Valley will wield greater influence over AI governance with each successive summit,” researcher Kyle Evanoff wrote on a blog external linkon the Council of Foreign Relations website.
“Policymakers should recognize AI’s unprecedented transformative power and take a more proactive approach to addressing new technologies. The greatest risk of all is inaction.”