Summit: Artificial Intelligence Is Humanity-Changing, Build Safeguards Now

Warning: Attempt to read property "post_title" on null in /home/clients/58f2a29976672af522a8f4d82ffa28b6/web/wp-content/plugins/better-image-credits/better-image-credits.php on line 227

Warning: Attempt to read property "post_title" on null in /home/clients/58f2a29976672af522a8f4d82ffa28b6/web/wp-content/plugins/better-image-credits/better-image-credits.php on line 227

For artificial intelligence enthusiasts, the future is bright. Soon intelligent machines will help humankind solve most problems, and according to one speaker at an artificial intelligence summit in Geneva this week, humans will be outsmarted by robots in the foreseeable future, in an artificial intelligence bliss. For others, artificial intelligence is far from delivering a fully positive outcome, and for several United Nations representatives, such as the World Health Organization, the world should not be entrusted to robots just yet.

The AI (Artificial Intelligence) for Good Global Summit is taking place from 7-9 June, and is organised by the International Telecommunication Union (ITU), and XPRIZE Foundation, self-defined as an “innovation engine.” The event is also convened in partnership with 20 other United Nations agencies, such as the World Health Organization, UNICEF, World Food Programme, UN Conference on Trade and Development, and the World Intellectual Property Organization.

Robot attendee at the AI for Good Global Summit

Over the three days, a number of main sessions and several parallel tracks are seeking to address questions linked to artificial intelligence (AI), such as privacy and ethics, societal challenges, sustainable living, poverty reduction, capacity building, and investment and economic aspects.

Houlin Zhao, ITU secretary general, said at the event opening today that AI will have an important role in the achievement of the UN Sustainable Development Goals (SDGs), and dialogue between UN agencies and industry is essential. He added that UN Secretary General António Guterres proposed to establish a special advisory committee on AI, and suggested that the outcome of the summit this week would provide valuable input for the advisory committee.

Exponential Trend, Wisdom of the Crowd

According to Marcus Shingles, CEO of the XPRIZE Foundation, problem-solving was traditionally the affair of governments and industry. However, a new type of problem solver appeared in recent years: the individual. The question is how to empower individuals to help reach the sustainable development goals (SDGs) using tools like AI.

The computing power of the world is steadily increasing, said Shingles, following what is known as Moore’s Law, which states that overall processing power for computers doubles every two years. Once something becomes digital and can take advantage of a doubling pattern, he said, it gains an exponential capability.

“Now you can buy CRISPR gene-editing software, and edit genes,” he said, adding that even an individual can do that, when only big companies could do this a few years ago.

A new study published in Nature a few days ago shows unexpected mutations in mice after CRISPR, which is short for clustered regularly interspaced short palindromic repeats, a gene-editing technique (IPW, Health and IP, 16 February 2017).

Exponential trends now include autonomous driving cars, 3D printing, and AI, which is behind most of these trends, he said.

Three billion people are connected to the internet today, about 50 percent of the world population, and this will sharply increase in the near future, he predicted, underlining ongoing initiatives “to put the world online in the next seven to ten years,” such as the Google Loon project (involving global balloons), Facebook internet drones, and Qualcomm satellites.

That would be 7 to 8 billion people walking around with supercomputers in their pockets with access to AI, 3D printers, biotechnology, and gene editing – and all connected, he said.

The collective wisdom of the crowd will solve the problems of the world, according to Shingles.

Automated Driving Safer, Robots Taught to Be Good Citizens

Rupert Stadler, CEO and chairman of the Board of Management at Audi AG, and founder of the Audi “Beyond” initiative asked how AI can be harnessed to solve core challenges, and make the world a better place.

There is a need for awareness of the change in the relationship between humans and machines, he said, touting the many advantages of AI and robots for mankind, taking as example medical robots facilitating diagnosis, helping with cancer research by coupling all computers to speed up research.

Stadler called for fair algorithms maintaining a plurality of opinions and allowing information self-determination, as opposed to the “filter bubble of social networks, which isolate us from facts and opinions opposing our viewpoints.”

Automated and Autonomous driving are leaving more decisions to machines, he said, adding that AI could make driving more efficient and “our lives safer”, given that 90 percent of all car accidents are due to human error.

However, the legal framework around those new technologies needs to be shaped and harmonised, Stadler said, adding that public acceptance of those technologies is also key. Ethical concerns exist, he said, and should be taken seriously.

AI will allow us to make our lives easier by collecting and interpreting huge amount of data and by predicting situations of the future.

For Jürgen Schmidhuber, scientific director of Swiss AI Lab, at IDSIA (Istituto Dalle Molle di Studi sull’Intelligenza Artificiale), robots will soon outsmart humans.

AI will impact the way people live and work, and transform society, he said. Robots can be educated to become valuable members of society, said Schmidhuber, who added, “We cannot prove they will always be doing the right thing, but we punish them for doing the wrong thing and we reward them for doing the right things and that’s how they become valuable members of society.”

In the long run, it is clear that robots are going to transcend us, he said.

WHO, Other UN Agencies, Call for Caution

WHO Director General Margaret Chan told the conference that proponents of the power of AI often refer to two specific examples: self-driving cars, and the delivery of health care.

Margaret Chan addresses AI for Good Global Summit

However, the pace of technological advances has left very little space to reflect on those advances in public policies and address the ethical questions that arise.

The enthusiasm for the use of smart machines to improve health care reflects the perspectives of wealthy countries and private companies, she said, adding that during her travels as the WHO director general, she has been in many countries where health facilities lack basics such as electricity or running water.

Noncommunicable diseases, such as cancer, diabetes, and heart diseases, have overtaken infectious diseases as leading killers worldwide, she said, affecting all income groups in all places, and are the most costly.

Among advantages of AI, she cited supercomputers that can accelerate the screening of novel molecules in the search for new drugs, and speed up the reading and interpretation of results from radiographies, electrocardiograms, and the analysis of blood samples, while at the same time reducing the likelihood of human error.

However, Chan called for caution in the “midst of all this exciting potential.”

Medical decisions are complex, and they depend on context and values such as care and compassion, she said, adding, “I doubt that a machine will ever be able to imitate genuine human compassion.”

She then underlined the need for health systems to be able to respond to early diagnosis of an illness. “What good does it do to get an early diagnosis of skin or breast cancer if a country offers no opportunity for treatment, has no specialists or specialised facilities and equipment, or if the price of medicines is unaffordable for both patients and the health system,” she asked.

Medicines and medical devices are heavily regulated, “with good reason,” she said, remarking that a machine cannot be sued for medical malpractice. Regulatory issues must be solved before a new AI technology reaches the market, she said.

She also noted that the mining of “huge amounts of data” raises serious issues of patient privacy and the confidentiality of medical records.

Izumi Nakamitsu, UN High Representative for Disarmament Affairs, also called for caution and underlined issues such as the absence of technical barriers to developing autonomous weapons, like creating a machine or an algorithm making decisions to take a human life without human involvement.

Frank La Rue, UN Educational, Scientific and Cultural Organisation (UNESCO) assistant director-general, said technology should remain at the service of humans, not the other way around. UNESCO supports technological progress, he said, but all technological policies should be balanced with social policies.

Machine Learning Needs Data, Human Labour

Peter Lee, corporate vice president of Microsoft AI and Research, said there is a high dependence in the area of AI on machine learning, which relies on data. In commercial practices today, data is mined, “it is the digital exhaust of human activities.”

For example, he said, to achieve a machine able to provide speech recognition, many hours of human speech are recorded and then human labour is necessary to label those recordings. Machine learning systems today are very efficient and of high quality, but depend on very large amount of data, and are slow learners.

Luckily, there are very specialised algorithms with large amounts of computing power which can take thousands of hours of training data and feed data into machine learning systems, Lee said.

Yoshua Bengio, a professor at the University of Montreal, remarked on the limitation of the knowledge that can be fed to computers, since a lot of human knowledge is intuitive, and uncommunicable.

“Deep learning” in AI is part of machine learning, and is inspired by the brains, through which computers learn from data how to have good representation of the information, he said.

Deep learning has been very successful in the last few years, he said, with computers being able to extract good representation from data. Successes have been achieved in speech recognition, and computer vision, he said. A computer can now look at an image and to some extent make sense of it.

Recently progress has also been made on use and understanding of language, but this is at an early stage, Bengio said, very far from a computer able to hold a general purpose dialogue with a human, he said.

“The low-hanging fruit” is in medical images, he said, with the ability of computers to detect cancer cells or colon polyps from videos for example. Computers can do better than the best doctors, he said.

However, Bengio said medical data is hard to get because it is hoarded by hospitals, doctors, and organisations protecting the privacy of patients. “Right now, we are not striking the right balance” of building something for the good of everyone’s health, and protecting individual privacy, he said.

On computers equalling humans, he said that likelihood is “very, very far,” adding that most major breakthroughs that we have today are based on human-supervised learning, in which the computer has to be told what is the right answer on millions of cases.

The systems we currently have can make all kind of “crazy mistakes,” said Bengio, adding there are limitations to the current science, and the current state of the art is not magic, “far from that.”

Even though the technology is still limited, he said, it can still have huge impacts.

 

Combat the infodemic in health information and support health policy reporting from the global South. Our growing network of journalists in Africa, Asia, Geneva and New York connect the dots between regional realities and the big global debates, with evidence-based, open access news and analysis. To make a personal or organisational contribution click here on PayPal.