Ginni Rometty, Satya Nadella, and MIT Media Lab director Joichi Ito – who was also on the panel – ceded that there will be seismic changes ahead directly as a result of advances in artificial intelligence.
See also: How to watch Davos 2017 keynotes
The talk was peppered with acknowledgements that there are risks ahead and that AI presents a world of unknown unknowns – that this is new territory, and even those ahead in development are trying to figure out the way forward.
In particular there were concerns over the societal impact that could emerge from these new technologies.
“Odds are, there are some jobs that will be wholly replaced by automation,” Rometty said, before adding that “most of us will be working with these systems.” To drive home her point, Rometty insisted that the main drivers behind at least IBM’s vision for artificial intelligence are purpose, transparency, and augmentation – in short that AI should be designed with a clear purpose in mind, to complement human behaviour, and the data sets that the machine learns from should be made known.
Taking a transparent approach is one way to gain trust in people who might be sceptical or fearful of artificial intelligence. “AI can be trained very differently,” Rometty said. “And the human needs to remain in control of these systems – to always pull the plug.”
Satya Nadella said that for him the key to AI is in how to make it “broadly accessible”.
“That is the true benefit of AI,” Nadella said. “In our case, one of the things that inspires me is the state I was born in, in India, and the state I now live in, in the US, are both using statistical machine learning to be able to improve the college high school outcomes and use the scarce state resources smartly. That, to me, is democratising AI.”
But both recognised that there will naturally be concerns about the impact of AI on society the world over. Their views somewhat mirror the World Economic Forum Global Risks report for 2017.
In the document, the WEF notes that there are “governance dilemmas” to come in navigating the regulation of AI and autonomous vehicles. Respondents to the survey also cited AI and robotics the most when asked about 12 emerging technologies that could pose the most risk – whether in the economic, geopolitical or technological spheres.
Nadella addressed this from Microsoft’s perspective, but his response might not be considered particularly reassuring in the wider context.
“It’s augmentation not replacement – that’s a design choice,” Nadella said. “You can come at it and say replacement is the goal, and augmentation is the goal – in our case [it’s] augmentation.”
Both Nadella and Rometty agreed that there needs to be a degree of collaboration between private industry and government. MIT Media Lab’s Joichi Ito suggested it’s irresponsible to leave the governance of AI to the market alone: “Understanding each other is the role of these partnerships in AI and some of the research we’re doing,” he said. “But it’s beyond a panel discussion – you need to intensively go in because most engineers don’t understand the law, and most engineers don’t understand why government exists.”
Ito also raised the question of bias in data sets and algorithms – a topic explored in Cathy O’Neil’s Weapons of Math Destruction - and how this can adversely increase inequality due to the inherent biases of the people writing the code, who are often white males.
This led to a deeper discussion of the elephant in the room, inequality, with chair Robert F. Smith of Vista Equity Partners raising the recent Oxfam study that claimed eight people own over half of the world’s wealth. Will AI be used for the good of the many or to bolster the wealth of the few?
Rometty insisted that the positives outweigh the potential negatives. She mentioned her concept of ‘new collar’ workers – that significant training programmes need to be undertaken the world over to ensure people are equipped with the right skills for a changing future.
“On a macro basis, there are five million open jobs in America,” she said. “On a macro level this sounds good... [but] on a micro level, when you have been displaced or your skills have not kept up, this to me is the issue of our time. Skills is the issue of our time.”
She went on to say that neither government or businesses can solve this alone, and recommended that public-private partnerships should seek to bust myths about AI as well as train people in the relevant skills to cope with the change that will be underway.
Nadella and Rometty agreed that there will be new jobs created by a society that’s more deeply integrated with artificial intelligence – but that this will require the skills to match.
“Take this very seriously and understand, this is a once in a generation change in skill type,” Rometty said.
Nadella said overall GDP growth is “far from stellar” and so “we need technological breakthroughs, we need AI”.
He brought up the first industrial revolution and Rousseau’s theory of the social contract – in other words, democratic consensus in the way society is run – and said that understanding this is the key to navigating the challenges posed by AI.
“We should do our very best to work in helping train people for the jobs of the future,” Nadella said. “None of us can sit here and predict all these jobs, but the lump of labour fallacy will be disproven, there will be new jobs. But how do we know what are the skills? This is where we need new breakthroughs.”