The HAL 9000 computer menaced humans in Stanley Kubrick’s “2001: A Space Odyssey.” But it’s becoming clear that this type of dystopian view of artificial intelligence is out of phase with humans’ real-world expectations and hopes for AI.
People know this for two reasons: The first has to do with results of a consumer survey we recently fielded; the second has to do with the power of broccoli (I’ll get to that in a moment).
Consumers, it turns out, may love dystopian fiction about AI’s impact (going all the way back to Mary Shelley’s “Frankenstein”), but they firmly believe that, in real life, AI will actually make the world a much better place.
>See also: AI: the possibilities and the threats posed
That’s what we found from an independent global survey we commissioned earlier this year, assessing the public’s attitude towards intelligent machines. Not only did it reveal that most people were in favour of AI technology, but they see it as bringing huge potential benefit. AI, in their eyes, will give us more flexibility and control, enhance safety and make our lives more rewarding.
The survey findings (download the full report here) were more positive than we expected and suggest a significant proportion of the public across all global regions now see AI as the ultimate realisation of what advanced technologies can do for us.
Ready for change
The survey was prepared by Northstar Research Partners, in partnership with ARM. It was distributed in May to nearly 4,000 respondents in US, Europe and APAC.
More than a third of consumers surveyed believe AI already has had a noticeable impact on their daily lives. That belief swells to more than two-thirds of respondents when asked about the anticipated impact just six years from now. And more than 60% of the people believe AI technology will make the world a better place, using words like “optimistic” and “excited” to describe the AI-shaped future.
Respondents were given various future possibilities to consider. In one health-care scenario, there was an even split between consumers choosing an AI machine or a regular human doctor. Taken a step further, respondents said they had no qualms about having an AI doctor perform an eye examination (57 percent) or even brain surgery (41 percent).
Half of consumers would trust an autonomous car to drive their family if their accident rates were demonstrably better than human drivers. In the next decade, this number rises to over two-thirds of consumers.
Respondents also see humans developing bonds with AI robots, perhaps in the way humans now love their pets, and using the technology to care for the sick or elderly.
There are, of course, concerns. The biggest perceived drawback relates to employment: Fewer or different jobs for humans (although those respondents see some jobs that robots actually should do instead of humans).
Security is also a concern, which is not surprising in an era in which high-profile hacks seem to occur almost daily. And over half have concerns about AI machines becoming more intelligent than humans. This was seen when AlphaGo beat the world Go champion, Lee Sedol, for the first time, and observers saw it as foreshadowing the uprising of the machines.
Eat your veggies
Here’s where the broccoli comes in. Broccoli has nearly twice the energy per gram of a lithium ion battery. To run your brain for a day, you would need 62 Lion batteries.
To play a four-hour game, AlphaGo (powered by 1,920 microprocessors and 280 graphics processors) consumed approximately 500kW. In human terms, that’s the equivalent of 1.7 million calories.
During the same contest, Lee Sedol consumed approximately 400 calories (less than one bunch of broccoli), making him 4,000 times more energy efficient for a similar level of intelligence.
>See also: Is business data AI compatible?
So a machine can triumph over humans, but at a tremendous energy cost. At the same time, humans aren’t viewing the man-machine dynamic as a zero-sum game: Shortly after Sedol’s defeat, the other world-class Go players competed alongside AlphaGo program in human-machine teams.
The players studied previous AlphaGo games and adopted new strategies. Here was an excellent example of human-machine collaboration, which we also see in medicine and other sectors.
Society is in the early, exciting phases of the next generation of artificial intelligence and machine learning, a phase where humans still are vastly more efficient computing machines and are learning to harness the machine’s power to work smarter. It’s quite the opposite of “2001: A Space Odyssey,” where humans found they couldn’t work with HAL and HAL found he couldn’t work with humans.
Sourced from Brian Fuller, Editor in Chief at ARM