By Gilbert Wong
When is too much knowledge a bad thing? According to the International Association of Scientific, Technical and Medical Publishers there are about 10,000 publishers of scientific journals worldwide producing some 33,000 active peer-reviewed journals in English, plus a further 9400 non-English journals. Together they publish around 3 million research articles each year.
This is a problem. Professor Michael Witbrock of the School of Computer Science at the University of Auckland says, “Every few seconds another paper is published in molecular biology. Humans can’t keep up with this. Even with thousands of researchers in the field, there’s still a bandwidth problem. We are missing out on potential medical advances because we can’t read our own literature.”
Suppose you had a computer that could comb all the research literature for answers to specific problems, able to search the entirety of experimental, clinical, and epidemiological research and sift through the diagrams, text and images for patterns, clues and connections for novel treatments and therapies. It could mean the end to many of the diseases and disorders and save countless lives.
Michael, a leading light in the small world of AI scientists, returned to New Zealand to launch the Broad AI Laboratory at the University. His life has had a singular focus on Artificial Intelligence and he has built a distinguished career that includes leading the AI Reasoning Lab at IBM Research, and senior roles in Cycorp, the company behind the world’s longest lived AI project.
It began when his dad took the young Michael to 2001: A Space Odyssey in Dunedin. Like many, he found HAL, the intelligent computer, the most engaging character. “HAL’s end is tragic but the idea of making something as intelligent as a person struck me as something I wanted to work on.”
He was in the right environment. His father Jack, an Anglican priest turned high school teacher ran the science club at St Paul’s High School in Dunedin, which became notorious for dangerous achievements, like building a working CO2 laser powerful enough to set fire to objects put in its path. His mother Julia was a science fiction fan, actively sparking Michael’s boyhood imagination.
HAL was science fiction, but a HAL for real, without murderous intent, is something Michael expects to see in operation well before the end of the 21st century. A working model for HAL would be what Michael calls a Broad AI.
The term was coined by a former colleague Bowen Zhou at IBM Research, who now heads AI development for jd.com, a giant Chinese online retailer. The AI that we know today, can do one thing supremely well, play chess better than anyone, read medical scans with more accuracy than humans, figure out what you want to buy and when you might want to buy it online.
However to AI strategists like Michael, the Holy Grail is general or broad AI. “Broad AI is able to do all the thinking that humans do, but without the same limitations.” By limitation he means ‘bandwidth’.
“Our input and output is quite low. We only have one thread of attention; we can’t multi task. Computers can. Humans have a hardware issue, we have one focus and that’s inconvenient, we have unreliable memories.”
A Broad AI would reuse knowledge, build reasoning to figure out complex questions, research the answer and explain it in natural language. When it comes to that classic training ground, chess, Michael says that progress means it is unlikely a human will ever beat an AI. But we don’t have an AI that is great at chess and finding a cure for cancer.
Smarter than human
“We have accumulated enough evidence to show that AI is going to be smarter than humans. We’re the smartest animal on earth, but we got as far as we have compared to a bunch of animals similar to us that we got the jump on only relatively recently. If you think of the difference between us and our fellow primates or the differences between us and an octopus, a dolphin or a kea, the difference is not that great.”
Michael’s logic is that with limited examples of intelligence on earth, the likelihood that humans are as intelligent as anything can be is unlikely. He would set the standards of intelligence for Broad AI far higher, witness the task already cited of combing through all medical research for new ways to treat human disease and disorder. The Broad AI Michael envisages could do that and also figure out a way to sequester carbon and beat all comers at chess all at the same time. That’s the upside.
The downside is less palatable. “There’s the idea that while the advent of AI will take away jobs, new kinds of work will rise. That fundamentally misunderstands the goal of AI.
“The next step is that we aren’t just seeking to automate some of the mental and intellectual tasks we perform but all of them. Being a professor, being a reporter, whatever job you like, the purpose of AI is to teach computers to do what we do.”
The Industrial Revolution’s first disruption was to make manual labour unnecessary, doing away with whole sectors of work. The final disruption Michael foresees will do the same to labourers of the brain. It’s been called the fourth stage of the Industrial Revolution and as it nears maturity Michael has a further prediction.
“The inevitable transformation we are facing is that everything we are useful for and necessary for will go away. The idea that everyone should be useful or work, that is going to go away.”
Michael says the idea of valuing a person due to their utility, by what they do, rather than who they are, is relatively recent in human history. “We are going to need to return to find ways of treating people as valuable because of their inherent value as human beings, irrespective of whether they are useful to society. That’s the disruptive concept.”
Michael is not a demonstrative speaker, his voice is quiet and he looks quite unperturbed as he predicts the end of white-collar work and a major somersault for capitalism. That’s because New Zealand has time to figure out how to handle the disruption to come and has a real opportunity to lead the way to utopia rather than dystopia.
“I think New Zealand has a mission, and historically that has been that we have always done things earlier.”
He runs through a list: universal suffrage, labour protections, gay marriage, trans rights and unique ways to respond to the legacy of colonialism. A political party at the last election raised the idea of a Universal Basic Income.
These are all examples of how New Zealand has been more thoughtful about how to build a good and inclusive society than many other countries. “If there’s one thing about New Zealand it is that we try social experiments. So we might be better at dealing with AI than elsewhere and show the world how we do it.”
The ingredients are right. New Zealand is small, well-educated and receptive to technology. Young people can take risks here that they cannot in America. Michael notes that the main criteria linking the young tech entrepreneurs behind American start-ups is wealthy parents. Failure would not leave them destitute. In New Zealand the risks for tomorrow’s tech entrepreneurs are mitigated by a social welfare net and public health system.
“New Zealand is a good environment to succeed at working out consequences for AI and by doing that New Zealand can have an impact on the future.”
Again the ingredients are here: a critical mass of academic, start up and financial activity and supportive policy from Government. Michael says there is sufficient scale and New Zealand offers a pleasant environment for smart migrants.
As for a diverse culture, Auckland surpasses New York. Michael advocates for an active recruitment campaign to bring innovators and entrepreneurs to New Zealand. “A lot of people dream about living in New Zealand. We should make that easy for people who are going to make a real impact for this country.”
Nor should we miss out on home-grown talent. “There’s less than 5m of us, so there’s no excuse for not finding those kids who have the potential to be tech entrepreneurs or scientists and supporting them to leverage New Zealand’s future.”
This article was originally published as part of The Challenge series and was republished with permission.
The Challenge is a continuing series from the University of Auckland about how researchers are helping to tackle some of the world’s biggest challenges.
Michael Witbrock is a Professor in Computer Science at the University of Auckland. He is an expert in artificial intelligence.