Government, the private sector and non-profits should work together to apply technology to address the nation's problems and to enhance civic engagement, said President Barack Obama, delivering the keynote Friday at the South by Southwest (SXSW) Interactive technology conference here.
The first sitting president ever to address the conference, he said if we pull together, "There's no problem we face in this country that's not solvable."
His address was in the form of an interview with Evan Smith, editor-in-chief of the Texas Tribune, who brought up the recent controversy between Apple Inc. and the FBI over smartphone encryption. President Obama noted that privacy must sometimes be sacrificed to an extent for the public good, such as security searches at airports, but on the other hand he did not like the idea of the government having "willy-nilly" access to people's smartphones.
But the idea of problems becoming increasing soluble -- or not -- was a constant theme in other sessions of the conference, largely devoted to emerging technologies.
Venture capitalist Steve Jurvetson, for instance, noted that Moore's Law, predicting a thousand-fold increase in computing power every decade, merely represents the "semiconductor epoch" of a trend that goes back to 1900, with the power of calculating devices increasing geometrically while their prices decrease geometrically.
"Wars and depressions have had no impact on the graph -- why should it change in the future? This is the most important thing ever graphed; nothing is more important for humanity's future," he said in a session titled "Accelerating Change and the Future of Technology."
"Already we are building artifacts that exceed human understanding," he said, referring to AI and deep learning. "Programming will become more like parenting than engineering. Moore's Law will revolutionize every industry over time," added Jurvetson, whose ventures have included SpaceX, Tesla Motors and Hotmail.
Meanwhile, AI that can, for instance, recognize pictures of cats is a marvelous thing. But for real-world jobs AI needs to be able to explain its decisions before humans can be expected to trust them, cautioned panelists in a conference titled "Big Data and AI: Sci-Fi vs. Everyday Applications."
"It is dangerous to think that machines will understand something if they are just given enough data," said Doug Lenat, head of AI firm Cycorp.
"A reliance on numbers and statistics will kill us," agreed Kris Hammond, chief scientist at Narrative Science. "What is being done now is magnificent, but it is not trustworthy. IBM's Watson (AI system) can tell you that it is 85% certain that you have thyroid cancer, but it can't tell you why."
AI that presents questionable ads to search engine users is bad enough, "But when you are dealing with people's lives you need results that are justifiable and auditable," said Rayid Ghani, from the University of Chicago.
Sign up for CIO Asia eNewsletters.