AI: The promise and the peril

Artificial intelligence is here and its presence and associated disruption will increase in the future. Panelists at the MIT Sloan CIO Symposium discussed ways to ease that disruption.

Mommas, don’t let your babies grow up to be truck drivers. Or pretty much anything that a machine or a robot could do, if you want them to have a job. The list of those things will continue to get longer – in some cases rapidly – extending well beyond the assembly line on a factory floor.

The forecast is not all gloomy – artificial intelligence (AI), machine learning (ML) and automation are also expected to create jobs that will likely be much more interesting and creative than the repetitive tasks of the industrial age.

Indeed, it has been a growing component of cybersecurity technology, and therefore cybersecurity jobs, for several years. Former Symantec CTO Amit Mital (now manager at KRNL Labs), at a panel discussion sponsored by Fortune magazine in 2015, called AI one of the “few beacons of hope in this mess” – the mess being cybersecurity, which he contended is “basically broken.”

[Related: -->Machine learning: Cybersecurity dream-come-true or pipe dream?]

That, according to a number of experts on panel discussions of AI at the MIT Sloan CIO Symposium on Wednesday, illustrates both the peril and the promise of the technology. The enormous challenge, they said, will be to minimize the peril while maximizing the benefits.

According to Andrew McAfee, principal research scientist at MIT and co-director of the university’s Initiative on the Digital Economy (IDE), AI amounts to, “the largest disruption in labor and the way we work,” in generations. He called it the, “second phase of the second machine age,” and noted that while he and his co-panelist, Erik Brynjolfsson, have written two books on the topic, “we don’t know what’s coming at us.” The panel title was that of their forthcoming book: “Machine, Platform, Crowd: Harnessing Our Digital Future.”

McAfee cited an example of the increasing power of AI from this week, when a computer program, Google’s AlphaGo, defeated Ke Jie, China’s top player of the ancient strategy game Go, after which Ke said this was no fluke – that the program’s “understanding of Go and the judgment of the game is beyond our ability.”

Brynjolfsson, MIT professor and director of the IDE, agreed. He said the “second wave,” is machines moving beyond what they are “taught” by humans to learning on their own. “It is the most important thing affecting the economy and society,” he said.

Those warnings were somewhat offset by assurances that while AI is already better than humans at jobs that involve “patterns,” and will be getting much better, it is not even close to matching humans in areas like creativity, collaboration and even conversations – “smart” machines are still dependent on the datasets used to train them.

[Related: -->AI isn't just for the good guys anymore]

That capacity to absorb and analyze massive datasets is one of the things that makes AI effective in cybersecurity, It can spot anomalies much more quickly than humans.

But as Joi Ito, director of the MIT Media Lab and moderator of a panel titled, “Putting AI to Work,” put it, the fear that machines will become smarter than humans and take over the world is tempered by the reality that “they’re stupid and they’ve already taken over the world.”

There was general agreement that AI is now generating and will continue to generate massive disruption. It will require massive adaptation if AI is to benefit society at large, and not just a few big winners. Some panelists were optimistic that, as has been the case with other technological revolutions, there will be new jobs created that can’t even be imagined now.

However, Ryan Gariepy, cofounder and CTO of Clearpath and OTTO Motors, was dubious that the same will happen with the revolution now under way. “My opinion is that we will not see net new job creation,” he said. “If I and other people do our jobs, you won’t need as many people to keep the world moving. There needs to be some social consideration of that.”

He said he expects millions of jobs to become obsolete, and for that trend to accelerate, adding that retraining is not always a practical option. “Truck drivers can’t go back to school,” he said, “and 90 percent of those jobs will disappear in a generation,” when autonomous vehicles become standard.

Brynjolfsson warned that it won’t just be low- to medium-skilled jobs affected. “There is the potential for it to take over many other jobs,” he said. “Machines can read MRIs and other medical images. People with 20 years training may find their skills are irrelevant.”

Ali Azarbayejani, CTO of Cognito Corporation, noted that while the current technology revolution will likely create many new jobs, they will be different jobs that require different skills.

Some of those jobs are already apparent – in cybersecurity. As has been well documented, robots and machines can be hacked. There have been high-profile demonstrations of hacks of self-driving vehicles. So those machines, devices and vehicles, and the individual users and companies that depend on them, will require an expanding security workforce for protection.

Seth Earley, CEO of Earley Information Science, while agreeing there will be, “an enormous amount of disruption," from AI, was more optimistic about retraining for the jobs of the future. “The thing that is causing the problem is part of the solution, because of improvements in training with robot simulation,” he said. “Imagine the best teacher you ever had. Imagine that being developed into a program.”

The least disruptive scenario, Ito said, would be for AI to “augment” rather than “automate” the workplace. “Augmentation doesn’t mean you’ve given up your agency,” he said. “I don’t think letting the machine decide is optimal.”

[Related: -->AI will transform information security, but it won’t happen overnight]

Azarbayejani said augmentation is one of the services his firm provides – listening to workers in large call centers, “not only for measuring (customer service) but how to improve in real time. It’s very much augmenting – it doesn’t replace humans, but helps them do their jobs better,” he said.

For those left unemployed, there was some discussion of the societal implications of providing a UBI (universal basic income) to all people whether they are working or not. But McAfee contended, “we are nowhere near peak labor,” and Brynjolfsson said most people want to work and “be engaged in their community. “We’re not in a world where we’re short of work that humans can do,” he said. “That’s decades out.”

If there is a way to prepare for what is already under way, several panelists said it will have to involve re-thinking education. “Kids should talk to each other and play with one another,” Brynjolfsson said. “Right now they are well trained for the first machine age, but not for collaboration and creativity.

McAfee agreed. “An amazing number of entrepreneurs were dropouts,” he observed.

Show Comments