Impact of AI: Doomsday or force for good?

Chair of the Department of Information Systems, Michael Goul, contemplates on the future implementation of information systems.

By Michael Goul, Chairman, Department of Information Systems

Stephen Hawking, Bill Gates, Elon Musk and others are warning that there are pitfalls if we develop artificial intelligence (AI) capabilities beyond the human brain. Hawking goes furthest, suggesting that AI could spell the end of mankind. Musk actually calls for government regulation over AI development, but he is unclear on what form that might take. Others hold different views. Microsoft’s Eric Horvitz feels the dire projections are overblown. In fact, he cites an upcoming competitive tournament between Microsoft’s new software assistant Cortana, Apple’s Siri and Google Now. By viewing AI in this assistance role, the doomsday scenario appears less likely.

Similarly, by exerting human control over AI by hosting and judging tournaments, the autonomy of the evolution of AI can be managed. This will include a type of self-policing; each organization will see what other organizations are doing in the space, understand the boundary norms and self-adjust as needed. But there is also a subtle implication: each company will own its AI, and will risk backlash if any of its AI applications get out of control. So how do you view the future of AI?  Evil or potential for good? Will it spin out of control, or can it be managed? Can we trust those building new AI implementations to keep them reined in? Or should we seek government regulation? Those of us in the IT community are likely to be front and center in these debates. In the meantime, many are putting big money into AI.

A partnership called Vicarious has an amazing set of investors including Jeff Bezos, Mark Zuckerberg, Peter Thiel and many others. Following is a description of the Vicarious project plan: “We are building a unified algorithmic architecture to achieve human-level intelligence in vision, language and motor control. Currently, we are focused on visual perception problems, like recognition, segmentation and scene parsing.

We are interested in general solutions that work well across multiple sensory domains and tasks.” The Association for the Advancement for Artificial Intelligence has long been involved in leading the way for AI research worldwide. This year’s conference was held in January. A quick scan of the program shows featured events and papers on AI in traffic, ethics and AI, robotics, AI in cities and even AI for assistive technologies. No doomsday here. In its early days, AI went through a progression similar to the cycles other technologies have gone through: basically, AI was overhyped with no possibility of meeting all expectations. The field fell out of vogue, but many of the early algorithms and approaches became part of the way we build applications, i.e., they were assimilated.

Today’s push has a more robotics- and sensor-based orientation; the emerging Internet of Things shows significant promise. In IS, many papers were published on expert systems and applied AI in business. Will we begin to see them again? To the extent that the IS community makes the move to a more sensor-based world, I suspect so. We used to say that many would anthropomorphize AI by reading so much into an application’s capability that it appeared to be akin to a human. If we infer that a technology could be evil perhaps we are falling into the same possibly faulty logic. But who am I to argue with the likes of Steve Hawking, Bill Gates and Elon Musk?  They are calling out this risk, and we should probably pay attention even as we cautiously embrace projects like Vicarious and the ongoing excellent work of the AAAI. Even if there is controversy, IS can gain from this new sensor infusion. There are likely to be excellent algorithmic approaches relevant to our work.