Session M3: Invited Session: Artificial Intelligence: Existential Risk or Boon to Humanity

Cognitive Computing: From Breakthroughs in the Lab to Applications in the Field

In the last decade, the availability of massive amounts of new data, and the development of new machine learning technologies, have augmented reasoning systems to give rise to a new class of computing systems. These ``Cognitive Systems'' learn from data, reason from models, and interact naturally with us, to perform complex tasks better than either humans or machines can do by themselves. In essence, cognitive systems help us penetrate the complexity of big data, reason using rich models, and enable each of us to perform like the best. We believe this will transform every industry for the better. Artificial Intelligence (AI) has a rich history, dating back to 1950's. The desire to build a Turing machine that would imitate a human dominated the psyche of the AI researchers for many decades [1]. The complex reality of human beings and the limitations of early symbolic approaches narrowed the success of early AI technology to a few specialized and small-scale applications [2, 3]. Over the last decade, new kinds of unstructured data from social networks, streaming data, and online publications, as well as massive data emitted from sensors from the physical world have outpaced traditional forms of structured data. This will continue to grow exponentially. The insights embedded in this massive amount of data can provide unprecedented opportunities for business and social value. Data has indeed become one of our most precious resources, and with its accelerated pace of evolution, it will determine the future trajectory of business and society [4]. New tools are being developed to extract insights out of the big data, which is abundant, unstructured, noisy, and unreliable. These tools have not relied on the same techniques that helped us exploit clean, structured data of the past, using small-scale models of the world and explicitly specified reasoning mechanisms. The new tools are using more automated statistical pattern-matching techniques, called machine learning, that have come of age in the last decade [ref]. In addition to reasoning from explicitly specified models of the world, these new machine learning techniques have given rise to a new class of systems that effectively learn from patterns in big data, and simultaneously augment their world models. Such systems can also interact naturally with us, on human terms, through natural language (i.e., unstructured text data), speech (i.e., unstructured audio data), vision (i.e., unstructured video data), and other modalities. We call this emerging class of systems that reason, learn, and interact naturally with us ``Cognitive Systems''. IBM's Watson is a family of cognitive systems targeted to a variety of domains. The first Watson system was capable of answering factoid questions as effectively as the best professionals in that field (as demonstrated by the Jeopardy! exhibition match, see illustration). Follow-on systems answer other types of questions, e.g., those that require passage answers, as well as other domains, e.g., healthcare, insurance, and education. Yet other cognitive systems in the Watson family go beyond question answering to support discovery of insights hidden in big data, such as in huge repositories of scientific literature, reasoning with evidence to support or refute topics of discussion, and to go beyond textual data to images and videos.   

Beneficial Smarter-than-human Intelligence: the Challenges and the Path Forward

Today, human-level machine intelligence is still in the domain of futurism, but there is every reason to expect that it will be developed eventually. A generally intelligent agent as smart or smarter than a human, and capable of improving itself further, would be a system we'd need to design for safety from the ground up: There is no reason to think that such an agent would be driven by human motivations like a lust for power; but almost any goals will be easier to meet with access to more resources, suggesting that most goals an agent might pursue, if they don't explicitly include human welfare, would likely put its interests at odds with ours, by incentivizing it to try to acquire the physical resources currently being used by humanity. Moreover, since we might try to prevent this, such an agent would have an incentive to deceive its human operators about its true intentions, and to resist interventions to modify it to make it more aligned with humanity's interests, making it difficult to test and debug its behavior. This suggests that in order to create a beneficial smarter-than-human agent, we will need to face three formidable challenges: How can we formally specify goals that are in fact beneficial? How can we create an agent that will reliably pursue the goals that we give it? And how can we ensure that this agent will not try to prevent us from modifying it if we find mistakes in its initial version? In order to become confident that such an agent behaves as intended, we will not only want to have a practical implementation that seems to meet these challenges, but to have a solid theoretical understanding of why it does so. In this talk, I will argue that even though human-level machine intelligence does not exist yet, there are foundational technical research questions in this area which we can and should begin to work on today. For example, probability theory provides a principled framework for representing uncertainty about the physical environment, which seems certain to be helpful to future work on beneficial smarter-than-human agents, but standard probability theory assumes omniscience about \emph{logical} facts; no similar principled framework for representing uncertainty about the outputs of deterministic computations exists as yet, even though any smarter-than-human agent will certainly need to deal with uncertainty of this type. I will discuss this and other examples of ongoing foundational work.